JPH09331469A - Video camera device, video signal processing device, and gradation conversion method of video signal - Google Patents

Abstract

(57) [Abstract] [Problem] To effectively use a limited dynamic range. SOLUTION: The signal level distribution information (histogram table) detected during an effective video period of an input video signal is accumulated and normalized in a histogram to obtain a normalized cumulative frequency distribution, and thus a complete histogram equalization. The amplitude transfer characteristic (conversion information indicating the correspondence between the level of the input video signal and the level of the output video signal) is obtained. Next, this amplitude transfer characteristic (solid line a in FIG. 10D) is adjusted to obtain an amplitude transfer characteristic (solid line c in FIG. 10D) in which histogram equalization is adjusted. Next, the black code offset B _OF is subtracted from the whole to save the black code (broken line d in FIG. 10E). Next, peak preservation processing is performed so that the luminance level of the input luminance A does not change even by conversion, and an amplitude transfer characteristic (solid line e in FIG. 10F) for adaptive gradation conversion by histogram equalization is obtained. . Luminance conversion of the input video signal is performed by using the amplitude transfer characteristic as a gain of luminance.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a video camera device, a video signal processing device, and a video signal gradation conversion method. More specifically, the present invention relates to a video camera device or the like which attempts to effectively use a limited dynamic range by preferentially compressing an unused gradation area of an input video signal.

[0002]

2. Description of the Related Art FIG. 37A shows an ideal television system 300A from an image picked up by a camera to a recording system, a transmission system and an image reaching a viewer at a receiver. This television system 300A is a camera system,
It is composed of a recording system, a transmission system, and an image receiving system.

In the camera system 300A, the light incident through the image pickup lens 301 is separated by the color separation prism 302 into red, green, and blue color component light, and the red, green, and
The blue color component light is the CCD solid-state image sensor 303R, 303
The red image, the green image, and the blue image of the subject are incident on G and 303B to be imaged on the image pickup surface. CD for the red, green, and blue image signals output from the image sensors 303R, 303G, and 303B, respectively.
A correlative double sampling process is performed by an S (corelated double sampling) circuit 304, and red, green, and blue color signals R,
G and B are taken out.

The color signals R, G, B taken out by the CDS circuit 304 are amplified by an amplifier 305, gamma-corrected by a gamma correction circuit 306, and then supplied to a signal processing circuit 307. Then, in the signal processing circuit 307, matrix processing is performed on the color signals R, G, B to form a luminance signal Y, a red color difference signal CR, and a blue color difference signal CB. And the color modulation processing is performed on the color difference signals CR and CB to form the carrier color signal C.

In the recording system, the luminance signal Y and the carrier color signal C output from the signal processing circuit 307 of the camera system are recorded and reproduced by a VTR (Video Tape Recorder) 308.

In the transmission system, the recording system VTR 308 is used.
The reproduced luminance signal Y and the carrier color signal C are supplied to the encoder 309 to form the video signal SV, and the video signal SV is modulated by the modulation circuit 310 to be an RF signal. The RF signal is transmitted by the transmitting antenna 311. Sent by. Then, the RF signal received by the receiving antenna 312 is demodulated by the demodulation circuit 313 to obtain the video signal SV.

In the image receiving system, the demodulation circuit 31 of the transmission system is used.
The luminance signal Y and the carrier color signal C are obtained by the decoder 314 from the video signal SV obtained in 3, and the luminance signal Y and the carrier color signal C are supplied to the signal processing circuit 315. In the signal processing circuit 315, color demodulation processing is performed on the carrier color signal C to obtain color difference signals CR and CB, and matrix processing is performed to the luminance signal Y and color difference signals CR and CB. R, G, B are formed. And
The color signals R, G, B output from the signal processing circuit 315 are supplied to a CRT (cathode-ray tube) 316, and the C
An image captured by the camera system is displayed on the RT 316.

According to the ideal television system 300A shown in FIG. 37A, although the signal system includes non-linear processing, there is an inverse conversion to it, so that the subject to the viewer's eyes are linear. Therefore, the image displayed on the CRT 316 is a faithful reproduction of the subject.

[0009]

However, the image pickup device 3
The dynamic range of 03R, 303G, 303B, and the dynamic range defined by the recording system and the transmission system are limited, and the configuration shown in FIG. 37A cannot be actually adopted. The restrictions on the dynamic range are narrowest in the recording system and the transmission system as signal standards, and it is necessary to take measures to accommodate the vast dynamic range of natural light.

Therefore, the current television system 30
In 0B, as shown in FIG. 37B, the pliny circuit 321 is inserted between the amplifier 305 and the gamma correction circuit 306, and the gamma correction circuit 306 and the signal processing circuit 307 are inserted.
By inserting a knee circuit 322 between the
The signal levels of G and B are kept within the standard.
The signal level according to the broadcast standard is defined as the signal level of the color signals R, G, B, and thus can be directly included in the standard. Note that in FIG.
7A is designated by the same reference numeral.

By thus knee-compressing the high-luminance region, the vast dynamic range of natural light can be contained within the range of the television signal standard. However, in this case, since the high-luminance region is simply compressed, there is a disadvantage that the contrast in a bright portion of the image becomes difficult to see.

Therefore, an object of the present invention is to preferentially compress the unused gradation range to effectively use the limited dynamic range.

[0013]

A video camera device according to the present invention comprises a cumulative frequency distribution detecting means for detecting a cumulative frequency distribution of an input video signal, and an input video signal level and an output video image based on the cumulative frequency distribution. First conversion information generating means for generating first conversion information indicating correspondence of signal levels,
By adjusting the first conversion information so that the output video signal obtained by converting the input video signal with the first conversion information when the input video signal has the predetermined level has the second conversion level. Second conversion information generating means for generating the conversion information and level conversion means for converting the signal level of the input video signal based on the second conversion information are provided.

Further, the video signal processing apparatus according to the present invention includes a cumulative frequency distribution detecting means for detecting a cumulative frequency distribution of the input video signal, and a level of the input video signal and a level of the output video signal based on the cumulative frequency distribution. The first conversion information generating means for generating the first conversion information indicating the correspondence between the input video signal and the output video signal obtained by converting the input video signal by the first conversion information when the input video signal is at a predetermined level. By adjusting the first conversion information so that the predetermined level is obtained,
Second conversion information generating means for generating second conversion information;
Level conversion means for converting the signal level of the input video signal based on the second conversion information.

Also, the video camera device according to the present invention detects the signal level distribution information regarding the signal level distribution of the input video signal during the valid video period of the input video signal, and the invalid video of the input video signal. During the period, the only detection that performs the second operation of detecting the conversion information for converting the gradation of the input video signal based on the signal level distribution information of the input video signal detected in the first operation Means, control means for controlling the operation of the detecting means in accordance with the timing of the input video signal, storage means for storing information including at least the signal level distribution information and conversion information generated during the operation of the detecting means, And a gradation converting means for converting the gradation of the input video signal based on the conversion information.

Further, the video signal processing apparatus according to the present invention includes the first operation of detecting the signal level distribution information regarding the signal level distribution of the input video signal during the valid video period of the input video signal, and the invalidation of the input video signal. Only the second operation of detecting conversion information for converting the gradation of the input video signal based on the signal level distribution information of the input video signal detected in the first operation during the video period Detecting means, control means for controlling the operation of the detecting means according to the timing of the input video signal, and storage means for storing information including at least the signal level distribution information and conversion information generated during the operation of the detecting means. , And a gradation conversion means for converting the gradation of the input video signal based on the conversion information.

Further, according to the gradation conversion method of the video signal of the present invention, the cumulative frequency distribution of the input video signal is detected, and the level of the input video signal and the level of the output video signal are associated with each other based on the cumulative frequency distribution. The first conversion information is generated, and when the input video signal is at a predetermined level,
Second conversion information is generated by adjusting the first conversion information so that the output video signal obtained by conversion by the conversion information of the second conversion information becomes the predetermined level, and the signal level of the input video signal is changed to the second conversion information. And a conversion step based on the above.

The cumulative frequency distribution of the input video signal is detected. For example, the cumulative frequency distribution of the input video signal is detected based on the signal level distribution information detected during the effective video period of the input video signal. Then, based on this cumulative frequency distribution, first conversion information indicating the correspondence between the level of the input video signal and the level of the output video signal is generated. By converting the signal level of the input video signal using this first conversion information, an output video signal in which the unused gradation region of the input video signal is preferentially compressed can be obtained.

When the input video signal has a predetermined level, the first conversion information is adjusted so that the output video signal obtained by converting the input video signal with the first conversion information has the predetermined level. Second conversion information is generated. Then, the signal level of the input video signal is converted on the basis of the second conversion information, so that the gray level of the input video signal which is not used is maintained while the signal level of the predetermined level of the input video signal is stored. An output video signal is obtained in which the region is preferentially compressed.

[0020]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 shows a video camera device 100 as a first embodiment.

The video camera device 100 is a microcomputer (hereinafter referred to as "microcomputer") 1 that functions as a system controller for controlling the entire system.
25. The knee point, knee slope, white clip level, normalization constant, total gain, time constant, histogram strength, black code and the like used when creating a controller table described later are given from the microcomputer 125.

The video camera device 100 also includes a lens block 101, a color separation prism 102 for separating the light incident through the lens block 101 into red, green, and blue color component lights, and the color separation prism. CCD solid-state image pickup devices 103R, 103G, on which red, green, and blue color component lights decomposed by 102 are incident to form a red image, a green image, and a blue image of a subject on the image pickup surface, respectively.
And 103B.

In this case, the spatial pixel shift method is adopted to improve the resolution. That is, as shown in FIG. 2, the image sensors 103R and 103B are
Are shifted in the horizontal direction by a half pixel pitch (P / 2). In this spatial pixel shifting method, the sampling points of the green imaging signal output from the imaging element 103G and the sampling points of the red imaging signal and the blue imaging signal output from the imaging elements 103R and 103B are 180 points.
It has a phase difference of ゜.

Further, the video camera device 100 includes an analog process circuit 10 for performing correlated double sampling processing and level control processing for red, green and blue image pickup signals output from the image pickup elements 103R, 103G and 103B, respectively.
It has 4R, 104G, and 104B. The reset noise can be reduced by performing the correlated double sampling process. In the level control processing, level control such as white balance and black balance is performed.

The video camera device 100 also includes A / D converters 105R, 105G, 105B for converting the red, green, and blue color signals output from the analog process circuits 104R, 104G, 104B into digital signals. Have Image pickup elements 103R, 103G, 10 described above
3B outputs red, green, and blue image signals at a rate of fs1 (eg, 14.31818 MHz), the A / D converters 105R, 105G, and 105B output red, green, and blue color signals at a sampling frequency fs1. It is sampled and held and converted into a digital signal.

Further, the video camera device 100 has an A / D
The level detector 12 detects the levels of red, green, and blue color data output from the converters 105R, 105G, and 105B, and supplies the detection outputs to the microcomputer 125.
6. The output of the level detector 126 is used for controlling an aperture (iris), for example.

Further, the video camera device 100 has an A / D
A pre-process circuit 106 that performs image processing such as black-and-white balance control, shading correction, and defect correction on the red, green, and blue color data output from the converters 105R, 105G, and 105B, and an output from this pre-process circuit 106. The up-converters 107R, 107G, and 107B obtain red, green, and blue color data having a sampling frequency 2fs1 that is twice as high as the red, green, and blue color data.
In this case, the red, green, and blue color data of the sampling frequency 2fs1 are processed so that they are in phase with each other.

The video camera device 100 also has a color correction circuit 108 for performing a linear matrix process on the red, green and blue color data output from the up converters 107R, 107G and 107B. In the linear matrix processing, the calculation processing of the expression (1) is performed, and red, green, and blue color data in which the color reproducibility of the captured image is corrected are obtained. Note that DRin, DGin, and DBin are red, green, and blue input color data, DRout, DGout, and DBout are red, green, and blue output color data, and a to f are coefficients. DRout = DRin + a (DRin-DGin) + b (DRin-DBin) DGout = DGin + c (DGin-DRin) + d (DGin-DBin) DBout = DBin + e (DBin-DRin) + f (DBin-DGin) (1)

In the video camera device 100, the red and green color data D output from the pre-processing circuit 106 are also output.
It has an image enhancer 109 for generating contour enhancement signals Da and Dc for enhancing the contour portion of the image from R and DG. In this case, the contour emphasis signal Da emphasizes the high frequency side, and the contour emphasis signal Dc emphasizes the low frequency side.

Further, the video camera device 100 subtracts the black code BC supplied from the microcomputer 125 from the red, green and blue color data output from the color correction circuit 108 to obtain the red, green and blue stimulus values R. , G, B convertor 110
R, 110G, 110B and the subtractors 110R, 11
From the luminance conversion calculator 111, a luminance conversion calculator 111 that performs a calculation for converting only the luminance to the stimulus values R, G, and B output from the 0G and 110B without affecting the hue and saturation. It has a saturation conversion calculator 112 that performs a calculation for converting only the saturation without affecting the luminance and hue of the output stimulus values R, G, B.

In order to convert only the luminance without affecting the hue and the saturation, the luminance gain kw is set to 3 as shown in the equation (2).
It can be applied commonly to the channels. In equation (2),
Ri, Gi, Bi are input side stimulus values, and Ro, Go, Bo are output side stimulus values.

[0032]

[Equation 1]

Here, for the stimulation values R, G, B,
W, x, y are obtained from the equations (3) to (5). At this time, W is brightness, and x and y have only color information independently of brightness W. When x = y = 0, it is a gray pixel with no color, the angle of the vector (x, y) represents the hue, and the magnitude of the vector (x, y) represents the saturation.

[0034]

[Equation 2]

The change when the luminance is changed by the conversion according to the equation (2) is examined. By substituting the equation (2) into the equations (3), (4), and (5), the equations (6) to (8) are obtained. (6)
In formula (8), Wo, xo, yo correspond to the stimulus values Ro, Go, Bo on the output side, and Wi, xi, yi correspond to the stimulus values Ri, Gi, Bi on the input side. Is.
As a result, it is understood that only the brightness is changed to kw times and the color is not affected at all.

[0036]

(Equation 3)

Based on the equation (2), the brightness conversion calculator 111
Are configured as shown in FIG. That is, the brightness conversion calculator 111 multiplies the stimulus values Ri, Gi, Bi by the gain kw to obtain the stimulus values Ro, Go, Bo 113R, 1
13G and 113B.

Further, in order to convert only the saturation without affecting the luminance and the hue, a linear operation as shown in the equation (9) may be performed. In the equation (9), Ri, Gi and Bi are input side stimulus values, Ro, Go and Bo are output side stimulus values, and kc is a saturation gain.

[0039]

(Equation 4)

Using the luminance Wi, the equation (9) becomes (1
It is also expressed as shown in equations (0) to (13). Ro = Wi + kc (Ri-Wi) ... (10) Go = Wi + kc (Gi-Wi) ... (11) Bo = Wi + kc (Bi-Wi) ... (12) Wi = 0.59Gi + 0.30Ri + 0. 11Bi (13)

The change when the saturation is changed by the conversion according to the equation (9) is examined. In equations (3), (4), and (5), (1
By substituting the equations (0) to (12), the equations (14) to (16) are obtained. In equations (14) to (16), Wo,
xo and yo correspond to the stimulus values Ro, Go and Bo on the output side, and Wi, xi and yo are the stimulus values Ri, Gi and Gi on the input side.
It corresponds to Bi. As a result, it can be seen that there is no change in luminance and hue, and only the saturation is changed to kc times.

[0042]

(Equation 5)

Based on the equations (10) to (12), the brightness conversion calculator 112 is constructed as shown in FIG. That is, the brightness conversion calculator 112 subtracts the brightness Wi from the stimulus values Ri, Gi, Bi.
4B and multipliers 115R and 115 for multiplying the output signals of the subtractors 114R, 114G and 114B by a gain kc.
G, 115B and the multipliers 115R, 115G, 11
Luminance Wi is added to the output signal of 5B, and stimulation values Ro, Go,
It is configured to include adders 116R, 116G, and 116B that obtain Bo.

Returning to FIG. 1, the video camera device 100.
Adds the black code BC and the pedestal level correction value PED supplied from the microcomputer 125 to the red, green, and blue stimulus values R, G, and B output from the saturation conversion calculator 112, and the image enhancer. Adders 117R and 117 for adding the contour enhancement signal Dc output from
G, 117B. In this case, the stimulation values R, G,
By adding the black code BC to B, it is converted into a code value. Further, by adding the pedestal level correction value PED to the stimulus values R, G, B, it is possible to adjust the red, green, and blue color data values when the iris (not shown) is closed, that is, the black level. .

In addition, the video camera device 100 has the red, green, and red output from the adders 117R, 117G, and 117B.
Gamma correction circuit 118 for performing gamma correction on blue color data
R, 118G, 118B and this gamma correction circuit 118
The contour enhancement signal Da output from the image enhancer 109 is added to the color data output from R, 118G, and 118B.
And adders 119R, 119G, and 119B for adding.

In addition, the video camera device 100 has the red, green, and red output from the adders 119R, 119G, and 119B.
Clip circuits 120R, 120G, 120B that perform clipping processing on blue color data at a constant level, and matrix processing for color data red, green, and blue color data output from the clipping circuits 120R, 120G, 120B. The matrix circuit 121 for forming the luminance data, the red color difference data, and the blue color difference data, and the luminance data, the red color difference data, and the blue color difference data output from this matrix circuit 121 are clipped at a constant level to obtain the luminance. Data DY, red color difference data DCR, blue color difference data DC
It has a clip circuit 122Y, 122R, 122B for obtaining B.

Further, the video camera device 100 has low-pass filters 123R and 123R for limiting the band of the red and blue color data output from the preprocessing circuit 106.
B, and an interpolation filter 123G for obtaining green color data in phase with the red and blue color data from the blue color data output from the preprocessing circuit 106. As the low-pass filters 123R and 123B, for example, [12221] type filters having frequency characteristics shown in FIG. 5B are used. Further, the interpolation filter 123G has, for example, the frequency characteristics shown in FIG. 5A [1344.
31] type filters are used. The low-pass filters 123R and 123B and the interpolation filter 123
FIG. 5C shows an overall frequency characteristic with respect to the luminance components of the red, green, and blue color signals whose G pixels are shifted.

Further, the video camera device 100 includes low-pass filters 123R and 123B and an interpolation filter 123G.
Based on the red, green, and blue color data output from the above, the gain k of the brightness used in the brightness conversion calculator 111 described above.
w, and a controller 124 for obtaining the luminance Wi on the input side and the gain kc of the saturation used in the above-described saturation conversion calculator 112. In this case, the controller 12
4, the luminance conversion calculator 111 and the saturation conversion calculator 1
12, so that the knee compression, DCC plus function, white clip, flare correction, and adaptive tone conversion by histogram equalization according to the present invention are performed by kw, W.
i, kc are formed.

Next, the video camera device 100 shown in FIG.
Will be described.

The light from the subject incident through the lens block 101 is supplied to the color separation prism 102 and separated into red light, green light and blue light, and the image pickup device 1 respectively.
03R, 103G, and 103B. Image sensor 10
On the imaging surfaces of 3R, 103G, and 103B, a red image, a green image, and a blue image of the subject are formed and imaged. Then, the imaging devices 103R, 103
The red, green, and blue image pickup signals output from the G and 103B are supplied to the analog process circuits 104R, 104G, and 104B to be subjected to correlated double sampling processing and level control such as white balance and black balance.

Also, the analog process circuits 104R, 1R
The red, green, and blue color signals output from 04G and 104B are converted into fs1 rate color data by A / D converters 105R, 105G, and 105B. The red, green, and blue color data are supplied to the pre-process circuit 106 and subjected to image processing such as black and white balance control, shading correction, and defect correction.

Then, the red, green, and blue color data output from the pre-processing circuit 106 are supplied to the up-converter, and the sampling frequency 2 which is twice in phase with each other.
Red, green, and blue color data of fs1 is formed. And
The red, green, and blue color data are supplied to the color correction circuit 108 and linear matrix processing is performed to obtain red, green, and blue color data in which the color reproducibility of the captured image is corrected.

By the way, for example, in the D1 code, the stimulus value 0 is defined as 16 (binary number), and the red output from the color correction circuit 108 is calculated by the brightness conversion calculator 111 and the saturation conversion calculator 112. If the green and blue color data (signal code) is used as it is, the black code will change. Therefore, the red output from the color correction circuit 108,
Subtractors 110R, 110G, 11 from the green and blue color data
At 0B, the black code BC is subtracted and converted into red, green and blue stimulus values R, G and B.

Then, the subtracters 110R, 110G and 11
The luminance conversion calculator 111 performs a calculation for converting the luminance on the red, green, and blue stimulus values R, G, and B output from 0B, and the saturation conversion calculator 112 converts the saturation. An operation for performing is performed. As a result, in the luminance conversion calculator 111 and the saturation conversion calculator 112, the knee compression, DCC plus function, white clip,
Flare correction, adaptive gradation conversion by histogram equalization, and the like are performed.

Further, adders 117R, 117G and 117
In B, the black code BC is added to the red, green and blue stimulus values R, G and B output from the saturation conversion calculator 112 to be converted into red, green and blue color data. The correction value PED of the pedestal level is added to adjust the black level. Further, the adder 117R, 117G, 117B adds the contour emphasis signal Dc output from the image enhancer 109 for emphasizing the low frequency side.

Further, the adders 117R, 117G and 117
For the color data output from B, the gamma correction circuit 1
Gamma correction is performed by 18R, 118G, and 118B, and the contour enhancement signal Da that enhances the high frequency side output from the image enhancer 109 is added by adders 119R, 119G, and 119B. The adders 119R and 119
The color data output from the G and 119B is clipped by the clipping circuits 120R, 120G, and 120B, and then supplied to the matrix circuit 121 and subjected to matrix processing. The clipping circuits 122Y, 122R, and 122B perform clipping processing on the luminance data, the red color difference data, and the blue color difference data output from the matrix circuit 121 to obtain the luminance data DY and the red color difference data DC.
R and blue color difference data DCB are obtained.

Next, knee compression performed by the luminance conversion calculator 111 and the saturation conversion calculator 112, DCC plus function, white clip, flare correction, adaptive gradation conversion by histogram equalization, and the like will be described, and luminance conversion will be described. Details of the controller 124 for obtaining the luminance gain kw, the luminance Wi, and the saturation gain kc used in the calculator 111 and the saturation conversion calculator 112 will be described.

(1) Knee compression First, the relationship between R, G, B levels and colors will be described. This description is based on the signal before any non-linear processing such as gamma correction. FIG. 6A shows an example of the levels of colorless pixels. When colorless, R:
G: B = 1: 1: 1. At this time, the level of each channel and the brightness W are equal, and R = G = B = W ((1
See 3) formula).

In a colored pixel, R, G and B are spread around W and distributed. For example, in a flesh color pixel, the distribution of R, G, and B is as shown in FIG.
Equation (13) is a positive coefficient whose sum of coefficients is 1, R, G, B
As you can see from the linear combination of R,
At least one channel of G and B is larger than W, and at least one channel is smaller than W.

When the saturation is reduced to half (the color is lightened) while maintaining the hue, the distribution of R, G, B becomes as shown in FIG. 6 (c). As the color is gradually reduced in this state, the level of each channel converges on W. On the other hand, if the iris (aperture) is opened from the state of FIG.
The distributions of G and B are as shown in FIG. In this case, the brightness W is increased, but the hue and the saturation are not changed.

In FIG. 6 (d), the level of the R channel exceeds not only the knee point but also the clip level, and since it does not meet the television signal standard, some compression processing is required. Therefore, in the conventional camera system, this restriction has been satisfied by performing knee compression for each channel as described above. FIG. 6 (e) shows
The distribution of R, G, B when knee compression is performed for each channel with respect to the distribution of R, G, B of FIG. 6D is shown.

By performing knee compression for each channel,
Certainly, the R, G, and B levels come to satisfy the television signal standard. However, the distribution of R, G, and
Looking at the balance of B, R, G, B of the distribution of FIG.
It can be seen that the balance has changed. This change extends to the hue, and when viewed in the image, the flesh color becomes yellowish, and it looks as if it was a health hazard.

Therefore, in the knee compression according to the present embodiment, the process of keeping the level of the over-run channel within the signal standard is performed in the following two stages. That is, the knee compression is performed on the level of the luminance W (luminance knee).
Then, for the channels that are still over,
Saturate until the highest level channel fits the standard (saturation knee).

In the distribution of R, G, B in FIG.
Similar to (d), the R channel level exceeds the clip level. FIG. 7F shows R, G, B of FIG. 7D.
R, G, when luminance knee processing is applied to the distribution of
The distribution of B is shown. In addition, FIG.
The distribution of R, G, and B when saturation knee processing is performed on the distribution of R, G, and B of (f) is shown.

The above-mentioned luminance knee processing and saturation knee processing are specifically carried out as follows.

In the brightness knee processing, the knee compression is applied to the brightness level based on the equation (2). The brightness gain kw is uniquely determined from the input brightness level when the knee curve is determined. If the knee having a slope of 0 is considered to be a clip, a white clip operation can be performed on the input luminance in the same manner.

In the processing of the saturation knee, (9) based on the luminance Wi and the stimulus values Ri, Gi, Bi of red, green and blue.
Calculate the expression. The saturation gain kc is determined by the channel level limit value CM and the maximum channel level MAX (R
i, Gi, Bi) is calculated by the equation (17).

[0068]

(Equation 6)

As described above, by performing the two-stage processing of the luminance knee processing and the saturation knee processing,
As shown in (g), it is possible to perform gradation compression in the high luminance range without changing the hue and without exceeding the channel level.

Although not described above, by setting the knee point instead of the luminance Wi and the knee slope instead of the gain kc of the saturation, the saturation conversion calculator 112 shown in FIG. It becomes an arithmetic unit. Therefore, the conventional knee compression for each channel can be selectively realized. This means that there is no increase in the circuit scale from the conventional system due to the saturation conversion processing.

(2) DCC plus function Even in the highlight portion, it is possible to add color by forcibly increasing the chroma level (the color signal after the I and Q matrix is called "chroma"). In the conventional knee compression for each channel, when the highlight becomes high, the hue changes to converge to white. The DCC plus function is a method that deviates from the strict television signal standard because the red, green and blue color signals demodulated in the receiver exceed the specified dynamic range. However, this DCC plus function is installed as an optional function mainly for commercial cameras because it has the advantage that colors are added to the high brightness region and there are no problems in actual operation.

Conventionally, after the gamma correction, the loose knee compression, and the white clip, the red color difference signal RY and the blue color difference signal B-
The DCC plus function is realized by applying the knee compression to the luminance signal Y converted by the matrix of Y, the I signal, and the Q signal, and not applying the knee compression to the color signal.

However, considering the color difference signals RY and BY, for example, there are the following problems. That is, since the luminance signal Y and the color difference signals RY and BY do not independently have luminance and color information but are dependent on each other,
When the luminance signal Y is changed, the color is also affected. Further, since it is a signal that has passed through non-linear processing such as gamma correction, strictly speaking, the hue changes.

Signals after matrix (Y, RY, B-
It is assumed that Y) is a linear signal that is not subjected to gamma correction or the like. Signal (Y, RY / Y, BY / Y)
Indicates luminance and color independently. On the other hand, the signal (Y, RY, BY) has a shape in which Y is applied to the color channel, and thus the color (hue, saturation)
The value of the color channel (R
-Y, BY) changes occur. On the other hand, if Y does change but (RY, BY) does not change,
The color will change. Why signal (Y, R-
A signal format such as Y, BY) is adopted. To obtain a signal (Y, RY / Y, BY / Y), division is necessary, and it is realized in a circuit. Because it is difficult.

In order to realize the DCC plus function, of the signals (Y, RY, BY) after the matrix, only the luminance signal Y is knee-compressed to reduce the level, and the color difference signal RY is obtained. , BY are left as they are, the actual color (RY / Y, BY / Y) has a smaller denominator, so the saturation is higher than the actual color, and an unnatural image appears. Become. Therefore, the knee point cannot be lowered so much by such a process. Further, in practice, since this processing is performed after non-linear processing such as gamma correction, the above-described color theory formula does not apply correctly, and strictly speaking, not only saturation but also hue changes.

Therefore, in the present embodiment, (1
After setting the channel level limit value CM of the equation (7) to 110% or more and suppressing the luminance W to within 110%, R, G,
The channel level of B is exceeded by allowing the limit value CM to be the upper limit. Although not mentioned above, the television signal has a limitation that the level of each channel of R, G, and B is suppressed to 110% with respect to the 100% reference white level.
By performing the processing as in this embodiment to realize the DCC plus function, the hue is maintained and the saturation is automatically adjusted to be as faithful to the original image as possible within a given range.

In the distribution of R, G and B in FIG.
Similar to (d), the R channel level exceeds the clip level. FIG. 8F shows R, G, B of FIG. 8D.
R, G, when luminance knee processing is applied to the distribution of
The distribution of B is shown. In addition, FIG.
With respect to the distribution of R, G, and B in (f), the distribution of R, G, and B when the saturation knee processing is performed by setting the channel level limit value CM to be larger than the clip level is shown. . As is apparent from FIG. 8 (h), it is understood that coloring in the high luminance range can be significantly obtained by relaxing the channel level limit value CM while keeping the luminance W limit unchanged.

(3) White Clip Conventionally, white clips are also applied to each of the R, G and B channels. Therefore, when the R, G, and B levels are applied to the clip level, the hue is naturally changed because the single channel is completely cut without considering the balance of the R, G, and B levels.

Therefore, in the present embodiment, as described in the above-mentioned knee compression, white clipping is applied to the luminance, and a single channel over that is handled by the saturation knee processing. To do. As a result, it is possible to perform processing that does not change the hue even with the white clip.

On the other hand, for example, the brightness white clip is set to 1
Set the channel level limit value CM to 109
By setting it to%, 9% in the meantime can be used for coloring, and the above-mentioned DCC plus function can be realized without departing from the television signal standard. The user can allocate a limited dynamic range to the gradation expression and the color expression.

(4) Flare Correction Conventionally, flare correction has been performed by reducing the pedestal level. In this case, how the color is affected will be described below.

It is assumed that the pedestal level a is added to a certain pixel (Ri, Gi, Bi) to obtain (Ro, Go, Bo). At this time, the expressions (18) to (22) are established. Here, Wi is the brightness according to the stimulus values Ri, Gi, Bi,
Wo is the brightness according to the stimulus values Ro, Go, Bo.

[0083]

(Equation 7)

The saturation SATi before adding the pedestal is expressed by the equation (23), and the saturation SATo after adding the pedestal.
Is expressed by equation (24).

[0085]

(Equation 8)

Here, if Wo / Wi = k is set, (24)
The formula is as shown in formula (25).

[0087]

[Equation 9]

Since SATo ≧ 0 and SATi ≧ 0, SATo = SATi / k. Therefore, the saturation becomes Wi / (Wi + a) times after the pedestal is added. That is, when a pedestal is added in the lifting direction, the color fades, and conversely, when the pedestal is added in the pulling direction, the color increases.

On the other hand, the hue HUEi before adding the pedestal is represented by the equation (26), and the hue HUEo after adding the pedestal.
Is expressed by equation (27). Therefore, the hue is preserved even by adding the pedestal.

[0090]

(Equation 10)

As described above, when flare correction is performed by adding a pedestal, the hue is preserved, but the saturation increases more than it actually is.

Therefore, in the present embodiment, (2)
In the equation, flare correction that does not affect the color is realized by performing control to reduce the gain kw in the gradation range in which black floating occurs. This operation is automatically performed when adaptive gradation conversion by histogram equalization described later is performed.
A correction operation is performed according to the occurrence of flare.

(5) Adaptive gradation conversion by histogram equalization In keeping the vast dynamic range of natural light within the range of the television signal standard, a method of compressing a high-luminance region by knee compression and blackening caused by flare are corrected. The method of compressing is as described above. In the present embodiment, by preferentially compressing the gradation range that is not used in the current image,
Further effective compression is performed.

Here, it is considered that a wide gradation area is used. That is, by taking the frequency of appearance of each brightness range in the screen, compressing the brightness range with a low frequency of appearance, and expanding the brightness range with a high frequency of occurrence, more gradations are added to the brightness range actually present in the screen. Allows compression to be assigned.

By this processing, the following effects can be obtained. That is, when the histogram is divided into a bright area and a dark area, such as when the shade and the sun are reflected together in the room and outside the window, the dark area is conventionally crushed (so-called black crushed). , The bright area was skipped (so-called whiteout), but this process makes it possible to make both areas visible. In addition, when black floating occurs such as when flare occurs, the histogram of the black region is low, so that the image is compressed and automatically adjusted to a black image quality. Further, even when the illumination is good, more gradation is assigned to the subject reflected in the image, so that a fine image is obtained.

Adaptive gradation conversion by histogram equalization will be described with reference to FIGS. 9 and 10.

The appearance frequency is expressed as a histogram in which the horizontal axis represents the luminance and the vertical axis represents the number of appearing pixels as a bar graph. The higher the value, the more gradation should be assigned. FIG. 9A shows an example of the histogram.
In this example, the illumination condition is relatively good, and the histogram is concentrated up to about 100%. If a differential gain proportional to the histogram is given, it is possible to give more gradations to the high luminance area of the histogram value. That is, if the integral of the histogram is used as the amplitude transfer characteristic, the differential gain becomes proportional to the histogram.

This integration of the appearance frequency in the horizontal axis direction is called the cumulative frequency distribution. The number of appearing pixels is integrated over the entire interval, that is, the right shoulder of the cumulative frequency distribution is always equal to the total number of pixels and is therefore constant. Further, the value of the histogram does not become negative, so it always becomes a monotonically increasing curve. FIG.
FIG. 9B shows a cumulative frequency distribution corresponding to the histogram shown in FIG. In this way, the cumulative frequency distribution is obtained by accumulating histograms.

If the luminance conversion is performed by using the curve of the cumulative frequency distribution as the amplitude transfer characteristic, complete histogram equalization is performed. That is, the histogram of the processed image is completely flat. FA (Factory Au
in the first stage of the binarization process of the sensor camera of
Although the histogram equalization is performed so far, it is often unfavorable because it is overemphasized in the video for viewing. Therefore, a procedure is added to adjust the method of histogram equalization.

First, a method of assuming the cumulative frequency distribution as the amplitude transfer characteristic will be described. As mentioned above, the right side of the cumulative frequency distribution is the total number of pixels (total number of histogram points).
be equivalent to. This value is normalized so as to be equal to the maximum value of the video signal code. The normalization constant for this is the video maximum code / total number of pixels. For example, if the video signal is 12
If the histogram is taken for 188928 pixels in bits, the normalization constant = 4095/188928
Is multiplied by the cumulative frequency distribution to normalize
It becomes a curve of amplitude transfer characteristics. FIG. 9 (c)
The amplitude transfer characteristic obtained by normalizing the cumulative frequency distribution of (b) is shown. The broken line shown by the broken line in FIG. 9C is the amplitude transfer characteristic of perfect histogram equalization.

Next, the method of histogram equalization is adjusted. As shown in FIG. 10 (d), the amount of multiplication can be adjusted by internally dividing between the complete histogram equalization (solid line a) and no histogram equalization (dotted line b). The solid line c in FIG. 10 (d) shows the amplitude transfer characteristic with the effect reduced to 1/3.

The histogram equalization processing up to this point is histogram equalization for the input video signal code. As shown in FIG. 10D, the black code changes due to this processing. The luminance conversion according to the equation (2) must be performed not for the video signal code but for the luminance stimulus value. Therefore, black code consistency must always be guaranteed. Therefore, a process of saving the black code is performed by subtracting the black code offset B _OF at the stage where the histogram equalization in FIG. The solid line c in FIG.
This is the same as the solid line c in (d), and the broken line d in FIG. 10 (e) shows the amplitude transfer characteristic after the black code offset B _OF is subtracted.

In this way, practical histogram equalization for viewing can be performed. If the signal complies with the television broadcasting standard, the processing up to this point is sufficient, but for the signal in the camera, if knee compression is performed after this, the priority of the high-luminance region to be knee-compressed is lowered. Can be considered. Therefore, the signals in the camera do not have the same priority throughout the code assignment. For example, in the histogram of FIG. 9A, the illumination condition seems to be good, and the histogram is gathered in the area of normal light amount. In this case, the histogram equalization processing shown in FIG.
In the amplitude transfer characteristic of the broken line in (c), the histogram gathered in the normal region is spread over the entire signal code to effectively utilize the code region. However, as mentioned above,
It is said that the high-luminance region has a lower priority, and if the histogram equalization process is performed as it is, an image with good lighting conditions will be knee-compressed. That is, it is not always good to adjust the in-camera signal to use the entire signal code.

It can be said that, in the case of a signal like this example, performing the histogram equalization processing while keeping the current peak value of brightness is an image quality improvement that does not contradict the photographer.

Therefore, in FIG. 10F, the brightness level of the input brightness A having the maximum histogram value is stored so as not to change even by conversion. This allows
If it was converted as it was, it was converted to the level of p1 and was knee-compressed contrary to the photographer's intention, but p
It is stored in 2 levels and the brightness is exactly as the photographer intended.

Specifically, the following processing is performed.
Conversion gain p2 / that sets the brightness level of the input brightness A to p2
Find p1. This conversion gain is multiplied by the entire amplitude transfer characteristic obtained in FIG. However, if the signal code is directly multiplied when being multiplied by the conversion gain, the black code saved by the above-described black code saving process again changes. Therefore, also in this peak preservation process, the luminance stimulus value should be converted.
Therefore, when the input luminance is Win, the output luminance is Wout, the conversion gain is kwh, and the black code is BC, the calculation of equation (28) is performed. Wout = (Win-BC) · kWh + BC (28)

Looking at the amplitude transfer characteristic after the peak storage processing shown by the solid line e in FIG. 10 (f), even if there is a constraint to store the luminance by this processing, the gradation is redistributed according to the histogram. You can see that it is.

In an area brighter than the input brightness A,
Although the differential gain is further reduced as compared with the amplitude transfer characteristic after the black code storage processing indicated by the broken line d in (e), FIG.
As can be seen from the histogram in (a), since the high-brightness region is less important in this image, the knee slope is equivalently laid down, which is reasonable.

In this way, when the processing up to the peak storage processing is completed, a series of camera histogram equalization processing is completed and the amplitude transfer characteristic is obtained as shown by the solid line e in FIG. 10 (f). In the present embodiment, the amplitude transfer characteristic is used as the gain kw of the brightness, and the brightness is converted by the equation (2), whereby the adaptive gradation conversion by the histogram equalization is realized.

(6) Manual Control of Saturation kc in the equation (17) is a saturation gain for narrowing the channel level when it exceeds the standard. K in equation (9)
c is a gain for adjusting the saturation in a broader sense. That is, if kc is set to 1.0, the saturation does not change, but if kc is set to 1.2, the color becomes slightly darker, and if kc is set to 0.8, the color becomes slightly lighter. By thus setting the saturation gain, the user can freely adjust the coloring according to the situation.

Therefore, in the present embodiment, the operation of reducing the saturation with the saturation knee is given priority, and in other cases, the color density can be adjusted by the gain of the saturation set by the user. In order to realize this, kc in equation (17)
Then, the minimum value of kcn set for saturation adjustment is taken.

As described above, the controller 124 includes the luminance conversion calculator 111 and the saturation conversion calculator 112.
According to Knee compression, DCC plus function, white clip, flare correction, adaptive gradation conversion by histogram equalization, and manual control of saturation,
kw, Wi, kc are formed.

FIG. 11 shows the detailed structure of the controller 124.

The controller 124 uses the filter 123.
A matrix circuit 201 for calculating the brightness W from the red, green, and blue color data R, G, B output from R, 123G, 123B according to equation (13), and the brightness W output by this matrix circuit 201. Luminance gain generator 202 for generating a luminance gain kw1 corresponding to, and an amplifier converter for up-converting the luminance gain kw1 at the fs1 rate output from the luminance gain generator 202 to obtain a luminance gain kw at the 2fs1 rate. 203, and a pixel averaging circuit 204 for averaging the brightness W output from the matrix circuit 201, for example, every 4 pixels or every 8 pixels to obtain a brightness Wh for obtaining a histogram.

The luminance gain generator 202 has a RAM (random access memory) 205 in which gain data of luminance corresponding to each of a plurality of sections divided into sections is written. In the present embodiment, the luminance range (for example, 16
000 to 3FF) is 0 to 60 as shown in FIG.
The RAM 204 functions as a table in which the gain data of the brightness for the 61 sections is stored. Further, the 61 sections are divided into three areas I to III as shown in FIG. 12, and are set so that the fineness of the section division in each area is different. For example, each section of the area 0 to 15 is 4 / step, and the area II is
Each section of 16 to 47 is 16 / step, and the area I
The section of 48 to 60 of II is 32 / step. In addition, regarding the number of interval division and the fineness of interval division,
It is not limited to the example described above.

Further, the brightness gain generator 202, based on the brightness W output from the matrix circuit 201, section data sec-1 indicating which of the 61 sections the brightness W is, and further within that section. Of the offset data ofs1 indicating the position of the pixel position, and based on the brightness Wh output from the pixel averaging circuit 204, the interval data sec-2 indicating which of the 61 intervals the brightness data Wh is in is output. Section generator 206 and section data sec1 output from this section generator 206
The address generator 207 outputs, as read address data, data sec-d sequentially indicating the section indicated by the section data sec-1 and the section before the section. In this case, when the section data sec1 indicates the section of 0, the address generator 207 outputs 0 section data s indicating that the brightness W is in the section of 0.
ec-0 is also output.

The brightness gain generator 202 outputs the read address data sec output from the address generator 207.
-d or read address data rad output from the sequencer to be described later is selectively taken out to RAM 205
To the brightness W by the interpolation circuit using the switch circuit 208 supplied to the RAM 205, the gain data q _n and q _{n-1 of the} brightness read by the read address data sec-d from the RAM 205, and the offset data ofs1 output from the interval generator 206. Interpolation calculator 2 for obtaining the brightness gain kw1 corresponding to
09 and.

The interpolation calculation in the interpolation calculator 209 will be described with reference to FIG. When the brightness W output from the matrix circuit 201 is Wa and is in the section of n,
Data sec-d output from the address generator 207
Based on the RAM 205, gain data q _n
And gain data q n-1 in the interval of _n-1 are output. Here, if the interval of n is m / step, (2
Interpolation calculation as shown in equation 9) is performed. In the interpolation calculator 209, when n = 0, the address generator 2
Based on the 0 section data sec-0 output from 07, q _n is used as the gain data q _{n-1 in the} equation (29).

[0119]

[Equation 11]

Further, the controller 124 uses the filter 1
23R, 123G, 123B outputs the maximum data MAX (R, R, G, B of the color data R, G, B
A maximum value circuit 210 for extracting G, B), a data MAX (R, G, B) extracted by the maximum value circuit 210, and a brightness W output from the matrix circuit 201, values corresponding to stimulus values, respectively. And a brightness gain multiplier 211 that multiplies the brightness gain kw1 output from the brightness gain generator 202.

The brightness gain multiplier 211 has a function of MAX (R,
G, B) or the brightness W, and a subtracter 213 for selectively extracting the brightness W and subtracting the black code BC from the output data of the switch circuit 212 to change the stimulus value.
And the gain of the brightness kw1 in the output data of the subtractor 213
A multiplier 214 for multiplying by, and MAX 'obtained by multiplying the maximum value of the stimulus values of red, green, and blue by the gain kw1 and data W'by multiplying the brightness W by the gain kw1 from the output data of the multiplier 214. And a switch circuit 2215 for outputting the output.

In this case, in the brightness gain multiplier 211, the switch circuits 212 and 215 are switched every 1/2 pixel cycle, and MAX (R, G, B) and brightness W are dot-sequentially processed. . As a result, one multiplier can be used, and the circuit scale can be reduced. The switch circuit 21
2, 215 and other switch circuits are switched by the sequencer 223 described later.

Further, the controller 124 has a saturation gain generator 216 which obtains a saturation gain based on the equation (17) from the data of MAX ′ and W ′ output from the luminance gain multiplier 211. ing. This saturation gain generator 216
Is a subtracter 217 that subtracts W ′ from MAX ′ and a channel level limit value CM provided from the microcomputer 125.
A subtracter 218 that further subtracts W ′, a divider 219 that divides the output data of the subtractor 218 by the output data of the subtractor 217, a saturation gain kc1 that is output from the divider 219, and is set by the user. A minimum value circuit 220 that outputs the smaller one of the saturation gains kcn.

Since the equation (17) includes division, there is a singular point. The singularity occurs when MAX '= W', that is, a colorless pixel. The divider 219 of the saturation gain generator 216 processes this and removes it as follows. That is, when MAX '= W', MAX '<CM
If so, kc1 = kcn, and if MAX '= CM, k
c1 = 1.00, and if MAX '> CM, kc1 =
Set to 0.00.

Further, the controller 124 up-converts the saturation gain of the fs1 rate output from the saturation gain generator 216 to obtain the saturation gain kc of the 2fs1 rate, and the luminance gain multiplier. And an up-converter 222 that obtains the luminance W'of the fs1 rate output from 211 and the luminance Wi of the 2fs1 rate.

Further, the controller 124 uses the RAM 20
5, the sequencer 223 that controls the operation for writing the gain data of the luminance to create the table, the RAM 224 used when creating the table, and the section data sec2 or the sequencer 223 output from the section generator 206 described above. Output address data ad
A switch circuit 225 for selectively extracting r and supplying it to the RAM 224, an arithmetic logic unit (ALU) 226 used when creating the table, and a sequencer 223.
Luminance data x from the address data adr output by
Luminance data generator 2 for generating and supplying ALU 226
And 27.

Next, the operation of the controller 124 shown in FIG. 11 will be described.

First, the operation for obtaining the luminance gain kw used by the luminance conversion calculator 111, the saturation gain kc used by the saturation conversion calculator 112, and the luminance Wi will be described. FIG.
Is an excerpt of a circuit portion of the controller 124 for obtaining kw, kc, Wi.

The operation for obtaining the gain kw of luminance is as follows. Red, green, and blue color data R, G, output from the filters 123R, 123G, 123B (see FIG. 1).
B is supplied to the matrix circuit 201, and the brightness W is calculated for each pixel. The brightness W for each pixel is supplied to the interval generator 206 of the brightness gain generator 202, and from this interval generator 206, interval data sec1 indicating the interval to which the brightness W belongs and the interval of the brightness W for each pixel. The offset data ofs1 indicating the position of is output.

Further, corresponding to the section data sec1 output from the section generator 206 for each pixel, the address generator 207 outputs data sec-d sequentially showing the section to which the luminance W belongs and the section before it. , This data se
cd is supplied to the RAM 205 as read address data. Therefore, the gain data q _n and q _{n-1 of} the brightness corresponding to the section to which the brightness W belongs and the section before that are read from the RAM 2 for each pixel. Then, the interpolation calculator 209 performs an interpolation calculation for each pixel using the gain data q _n and q _n−1 supplied from the RAM 205 and the offset data ofs1 supplied from the interval generator 206 (( 29)), and the gain kw1 of luminance is obtained. Then, the brightness gain kw1 obtained for each pixel from the interpolation calculator 209 is converted into a rate of 2fs1 by the up converter 203, and the brightness gain kw used by the brightness conversion calculator 111 is obtained.

The operation for obtaining the saturation gain kc and the luminance Wi is as follows. Filters 123R, 123G,
The color data R, G, B of red, green, and blue output for each pixel from 123B is supplied to the maximum value circuit 210, and the maximum data MAX (R, G, B) is taken out. Then, for each pixel, the data MAX extracted by the maximum value circuit 210
(R, G, B) is supplied to the brightness gain multiplier 211, the black code BC is subtracted to be converted into a stimulus value, and the brightness gain kw1 output from the brightness gain generator 202 is multiplied to obtain the data MAX. ′ Is obtained. The brightness W output from the matrix circuit 201 for each pixel is supplied to the brightness gain multiplier 211, the black code BC is subtracted to be converted into a stimulus value, and the brightness gain output from the brightness gain generator 202 is further increased. The data W'is obtained by multiplying kw1.

Then, for each pixel, the brightness gain multiplier 211
The data MAX ′, W ′ output from the output is supplied to the saturation gain generator 216. In the saturation gain generator 216, the saturation gain kc1 is calculated for each pixel using the data MAX 'and W'and the channel level limit value CM (see equation (17)). Further, in the saturation gain generator 216, the saturation gain k is calculated by the minimum value circuit 220 for each pixel.
The smaller one of c1 and the saturation gain kcn set by the user is taken out. Then, the saturation gain generator 21
The saturation gain output for each pixel from 6 is converted to a rate of 2fs1 by the up converter 221, and the saturation gain kc used in the saturation conversion calculator 112 is obtained.

Further, the data W'output from the brightness gain multiplier 211 for each pixel is converted into 2 by the up converter 222.
It is converted to the rate of fs1 to obtain the luminance Wi used in the saturation conversion calculator 112.

Next, in the RAM 205, as described above,
The operation of writing the gain data of the brightness corresponding to one section and creating the table will be described. FIG. 15 is an excerpt of a circuit portion related to the table creation of the controller 124. In FIG. 15, the RAM 205 and the sequencer 22
3, a portion other than the RAM 224, the switch circuit 225, and the luminance data generator 227 constitutes an ALU 226.

The ALU 226 is a switch circuit 230-2.
33, clipping circuits 234 to 236 for clipping overflow due to calculation, an adder / subtractor 237 serving as an adder or a subtractor, a register 238, and a comparator 2.
39, a division controller 240, a white clip circuit 241, a black code offset register 242 for temporarily storing a black code offset B _OF, and a peak preservation ratio register 243 for temporarily storing a peak preservation ratio.
, A multiplier 244, and a subtractor 255.

The RAM 224 functions as a work RAM. As will be described later, a histogram is taken in the RAM 224 during the effective pixel period, and during the vertical blanking period, the RAM 224 is used as a temporary storage of data during calculation.

The operation of table creation is performed by the sequencer 223.
It is partitioned by and is sequentially performed in the order of step 0 to step 15 shown in FIG. Sequencer 22
In No. 3, the effective pixel period is in step 0, and at this time, the external circuit operates to take a histogram.

When the sequencer 223 enters the vertical blanking period, the sequencer 223 proceeds to step 1 where it repeats sequences 0 to 7 while changing the address from 0 to 60, and then executes steps 2 to 15 in the same manner. And create a table. Here, the sequences 0 to 7 are f
It is sequentially performed at a rate of s1 (horizontal driving frequency of the image pickup elements 103R, 103G, 103B).

In step 12, since the division is performed, the sequencer 223 starts the division sub-sequencer in sequence 2 (divstart).
Thereafter, the sequence 3 is temporarily stopped (stop), and the completion of the division sequencer is waited for.

Further, the steps 3 and 4 have a slightly different operation from the other steps. In step 3, when the luminance range includes a black code (adr = blksec),
In order to capture the black code offset B _OF used in the black code storage processing (see FIG. 10E) in the adaptive gradation conversion processing by the above-mentioned histogram equalization into the register 242, a write enable is output to the register 242 in sequence 3. (Blkwr).

In step 4, when the luminance range is in the luminance range A instructed to save the peak (adr = hlds)
ec) The calculation for _obtaining the conversion ratio K _hold for performing the peak preservation processing (see FIG. 10F) in the adaptive gradation conversion processing by the above-mentioned histogram equalization is performed. Since the division is performed also here, the sequencer 223 reads the RAM 224 in sequence 4 (memr).
d) In sequence 5, the division sub-sequencer is started (divstart), then stopped in sequence 6 (stop), and the termination of the division sequencer is waited. Then, in sequence 7, the conversion ratio K _hold is stored in the register 243 (hldwr).

The table creation process in steps 0 to 15 of FIG. 16 will be described below.

(1) Step 0: Take Histogram (See FIG. 9A) Step 0 is performed during the effective pixel period. In this step 0, a histogram table is created in the RAM 224. Only at this time, as the address data of the RAM 224, the section data sec2 corresponding to the luminance value of the pixel
Is given. The adder / subtractor 237 in the ALU 226 adds 1 to the histogram value up to the present at the address corresponding to this section data sec2, and the histogram value is stored again in the RAM 224 at the same address. As a result, the histogram value is incremented as shown in equation (30). Here, RAM1out is output data of RAM224, and RAM1in is input data of RAM224. RAM1in = RAM1out + 1 (30)

This is performed during the effective pixel period by the section generator 206.
By repeating this for each section data sec2 output from (see FIG. 11), a histogram table for that field is created in the RAM 224.

FIG. 17 shows the operation of the ALU 226 in the histogram acquisition in step 0, and the related signal paths are indicated by broken lines. The same applies to the drawings showing the operations of the following steps. In this case, ALU 226
The adder / subtractor 237 of function as an adder.

(2) Step 1: Accumulation and normalization (see FIGS. 9B and 9C) Step 1 and subsequent steps are performed during the vertical blanking period. R
Address data adr is supplied from the sequencer 223 to the AM 224. In this step 1, the histogram is accumulated and normalized to be converted into a normalized cumulative frequency table. Accumulation is performed by register 23 in ALU 226.
8, the cumulative value up to that interval is multiplied by the normalization constant K _CCD by the multiplier 244, and the multiplication result is again stored in the RAM 22.
Store in 4. This is shown in the equations (31) and (32). Regin = Regout + RAM1out ... (31) RAM1in = Regout * K _CCD ... (32)

Here, Regout is output data of the register 238, and Regin is input data of the register 238. Then, in step 1 of FIG. 16, “m
"emrd" indicates reading from the RAM 224, and "reg"
“wr” indicates writing to the register 238, and “memw”
“R” indicates writing to the RAM 224. The same applies to the following steps. However, in step 13,
14, “memrd” indicates reading of the RAMs 224 and 205, and “memwr” of step 14 indicates the RAM 20.
5 shows writing.

FIG. 18 shows the operation of the ALU 226 in step 1 accumulation and normalization. ALU22
The adder / subtractor 237 of 6 functions as an adder. In this case, as the histogram value increases, the differential gain in the amplitude transfer characteristic increases, but the clipping process is performed by the clipping circuit 234 so that the differential gain does not become excessively large.

In the sequence 6 of step 1, the histogram read from the RAM 224 is stored in the register of the histogram information reporting circuit (histwr) as described later.

(3) Steps 2 and 3: adjustment of histogram equalization (see FIG. 10D) In steps 2 and 3, adjustment of histogram equalization is performed. That is, the strength of histogram equalization is designated as K _WC , and calculation is performed as shown in equations (33) and (34). Regin = RAM1out-x ··· (33 ) RAM1in = Regout * K WC + x ··· (34)

When K _{WC is set} to 1.00, the histogram equalization is completed. If _KWC is 0.00, no histogram equalization is performed. The x in the above equation is the brightness data generated by the brightness data generator 227 and corresponding to the section. This indicates the luminance data in the case that did not perform the conversion, when K _WC is 0.00, the x directly becomes the RAM1out.

FIG. 19 shows the operation of the ALU 226 in adjusting the histogram equalization in step 2. In step 2, the calculation of equation (33) is performed. Therefore, the adder / subtractor 237 of the ALU 226 functions as a subtractor. FIG. 20 shows the operation of the ALU 226 in adjusting the histogram equalization in step 3. In this step 3, the addition of the calculation of the equation (34) is performed. Therefore, the adder / subtractor 237 of the ALU 226 functions as an adder.

(4) Step 4: Black code storage processing (see FIG. 10E) In step 4, black code storage processing for removing the black code offset B _OF is performed. Here, in the section including the black code (adr = blksec), the difference between the table value after the addition / subtraction processing of the histogram equalization and x, that is, the black level offset B _OF is set in the register 242, and this is set in the entire section. By subtracting from the table with, the table value in the section including the black code is made equal to x. The black level offset B _OF is set in parallel with Regout * K _WC during the calculation of equation (34) in step 3.
Is stored in the register 242. The operation of subtracting the black level offset B _OF from the entire table is the operation of Expression (35). RAM1in = RAM1out-B _OF ... (35)

FIG. 21 shows the operation of the ALU 226 in the black code saving process of step 4. In step 4, the equation (35) is calculated. Therefore, A
The adder / subtractor 237 of the LU 226 functions as a subtractor.

(5) Steps 5 and 6: Peak Saving Processing (See FIG. 10F) In steps 5 and 6, peak saving processing is performed. Although not described above, in step 4, the peak preservation ratio K is calculated by dividing the conversion result p1 in the luminance range A to be fixed and the value when the conversion is not performed here, that is, x.
_Hold is calculated and stored in the register 243. That is, in step 4, the section of the brightness range A (adr = h
In (ldsec), the calculation of equation (36) is performed to obtain the conservation ratio K _hold . K _hold = (x-BC) / (RAM1out-BC) (36)

In general, division is represented by b / a = c, where numerator is b, denominator is a, and quotient is c. When this equation is modified, b = ac, and the quotient c can be obtained as the number x when it becomes equal to the numerator b when the denominator a is multiplied by the number x. Here, to obtain the quotient c,
It is sufficient to sequentially change x so that it converges on the quotient c.
For example, when the quotient c is obtained by n-bit data, x is n-bit data and MS is set so that ax does not exceed b.
It is only necessary to confirm in order from B, and finally confirmed n
The bit data x becomes the quotient c.

As a concrete example, b = 1010 and a = 111.
Then, a division process for obtaining 4-bit data from the 2 ¹ digit as the quotient c will be described. Consider 4-bit data x = [b3, b2, b1, b0] for obtaining the quotient c. First, b3, which is the MSB, is confirmed. b3 =
1, b2 = b1 = b0 = 0, and ax and b are compared. Since ax = 1110> b, b3 = 0 is set. Next, the confirmation process of b2 is performed. b3 = 0, b2 =
1, b1 = b0 = 0, and ax and b are compared. ax
Since = 0111 <b, it is determined that b2 = 1.
Next, the confirmation process of b1 is performed. b3 = 0, b2 = 1, b1 =
1, b0 = 0, and ax and b are compared. ax = 1
Since 010.1> b, b1 = 0 is confirmed.
Next, the confirmation process of b0 is performed. b3 = 0, b2 = 1, b1 =
Ax and b are compared with 0 and b0 = 1. ax = 10
Since 0.111 <b, it is determined that b0 = 1.
As a result, the quotient c = 0.01 is obtained.

FIG. 22 shows the operation of the ALU 226 at the time of calculating the peak storage ratio in step 4, and the peak storage ratio is calculated by the division processing as described above. At this time, the adder / subtractor 237 of the ALU 226 functions as a subtractor.

In this case, the brightness data generator 227 outputs the brightness data x corresponding to the section including the brightness A, and the subtracter 255 subtracts the black code BC from the brightness data x.
Is subtracted and supplied to the comparator 239. Also,
Luminance data R in the section including the luminance A from the RAM 224
AM1out is read out, the black code BC is subtracted from the luminance data RAM1out by the adder / subtractor 237, and the multiplier 24
4 is supplied. Then, in the multiplier 244, the output data of the adder / subtractor 237, that is, the RAM1out-BC is multiplied by, for example, 12-bit data b (11) to b (0) set in the register 238, and this multiplier 244 is used. Output data is supplied to the comparator 239. In the comparator 239, the output data of the subtractor 255 described above,
That is, x-BC is compared with the output data of the multiplier 244, and the comparison result is supplied from the comparator 239 to the division controller 240.

In this state, under the control of the sequencer 223, the division controller 240 first clears the register 238 for creating the peak conservation ratio K _HOLD (b (11) to b (0) = 0. ), Followed by the MSB b (11)
Is set to “1”. Then, the division controller 240
On the basis of the comparison result from the comparator 239, when the output data of the multiplier 244 is larger than x-BC, b (1
1) is changed to "0", while when the output data of the multiplier 244 is less than or equal to x-BC, b (11) remains "1",
Confirm b (11). Hereinafter, the division controller 240
"1" is set to b (10) to b (0) in order, and b (11) described above is set.
The same confirmation processing as in the case of is performed. Then, stored this 12-bit data b (11) where the determined such ~b (0) is supplied to the peak storage ratio register 243 from the register 238 as the peak storage ratio K _{HOL D.} Next, the preservation ratio K
_{Put hold} on the whole table. However, since it is necessary to keep the black level constant, the equation (37),
The equation (38) is calculated. Regin = RAM1out−BC (37) RAM1in = Regout * K _hold + BC (38)

FIG. 23 shows the operation of the ALU 226 in the peak storage processing (1) of step 5. In step 5, the calculation of equation (37) is performed. Therefore, the adder / subtractor 237 of the ALU 226 functions as a subtractor. Further, FIG. 24 shows the operation of the ALU 226 in the peak storage processing (2) of step 6. In this step 6, the addition of the calculation of the equation (38) is performed. Therefore, the adder / subtractor 237 of the ALU 226 functions as an adder.

(6) Steps 7 and 8: Knee compression processing (1) (see FIG. 25 (g)) In steps 7 and 8, the first knee compression processing is performed. To be precise, a table is created that performs knee compression. The process of applying knee to the level table is (3
9) to (42). Here, K _p is the knee point and K _S is the knee slope. Then, in the knee compression process (1), K _p = K _p 1 and K _S = K _S 1. When it is _{RAM1out ≧ K p Regin = RAM1out-} K p ··· (39) RAM1in = Regout * K S + K p ··· (40) RAM1out < when a _{K p Regin = RAM1out-K p} ··· ( 41) RAM1in = Regout * 1.00 + K p ··· (42)

FIG. 27 shows the operation of the ALU 226 in the knee compression process (1) in step 7. In this step 7, the operation excluding + K _P in the expressions (39) and (40) or the expressions (41) and (42) is performed. Therefore, the adder / subtractor 237 of the ALU 226 functions as a subtractor. FIG. 28 shows the operation of the ALU 226 in the knee compression process (1) of step 8. In step 8, the remaining + K _P is calculated. Therefore, the adder / subtractor 237 of the ALU 226 functions as an adder.

(7) Steps 9 and 10: Knee compression processing (2) and white clip (see FIG. 25 (h)) In steps 9 and 10, the second knee compression processing and white clip processing are performed. The process of applying the knee to the level table is the same as the knee compression process (1) (39)
~ The calculation is performed by the equation (42). In this knee compression process (2), K _p = K _p 2 and K _s = K _s 2. In this case, since the knee is applied twice, the slope of the final knee curve is K _s 1 * K _s 2. The knee is rounded off by this two-stage knee.

In the white clip processing, the white clip processing is executed by supplying white clip level data from the microcomputer 125 to the white clip circuit 241 at the stage of step 10. Therefore, in the other steps, the white clip circuit 241 does not function.

FIG. 29 shows the ALU 22 in the knee compression process (2) and the white clip process of step 9.
6 shows the operation of FIG. In step 9, (39)
Expression and (40) Expression, or (41) Expression and (42)
The calculation is performed except for + K _{P in the} expression. Therefore, AL
The adder / subtractor 237 of U226 functions as a subtractor. 30 shows the operation of the ALU 226 in the knee compression process (2) and the white clip process of step 10. In step 10, the remaining + K _P is calculated. Therefore, the adder / subtractor 23 of the ALU 226
7 functions as an adder.

(8) Step 11: Total Gain Adjustment (see FIG. 25 (i)) In step 11, the total gain G is stored in the level table.
ain is multiplied and the total gain is adjusted.
For example, a code assignment that is not in a power of 2 relation to the D1 code (8 bits), for example, MSB side or LSB
A with 11 bits expanded by 1.5 bits each
In the case where a / D-converted signal is input, the relationship between the D1 code and the power of 2 can be corrected by multiplying this by a correction coefficient. Here, the calculation is performed based on the equations (43) and (44). Regin = RAM1out−BC (43) RAM1in = Regout * Gain + BC (44)

FIG. 31 shows the operation of the ALU 226 in the total gain adjustment of step 11. In this step 11, + of the equations (43) and (44)
Calculations other than BC are performed. Therefore, the adder / subtractor 237 of the ALU 226 functions as a subtractor. The remaining + BC operation can be omitted as described in the next section (9).

(9) Step 12: Division Processing for Obtaining Transfer Gain (See FIG. 26 (k)) In step 12, the level table created so far is converted into a table of Kw which is the dimension of gain. Divide into. FIG. 26 (j) shows this conceptual diagram. For example, in the luminance range of the vertical broken line in FIG. 26 (j), the calculation for obtaining the gain for converting the level of a into the level of b may be performed. Here, it should be noted that the gain at which the black code BC becomes the zero element must be obtained. As shown in FIG. 1, the subtracters 110R, 110G, and
At 110B, the black code BC is subtracted from the red, green, and blue color data, and the code is converted into a stimulus value. Therefore, the gain obtained by the division here must also be the gain for the stimulus value, that is, the gain with the black code as the zero element. Therefore, the gain table is converted by the calculation of the equation (45). RAM1in = (RAM1out-BC) / (x-BC) (45)

Looking at the numerator of the expression (45), the addition of the black code BC in the expression (44) is again reduced. Therefore, these operations are redundant and can be omitted. In this case, equation (44), (4
Equation (5) is as shown in equations (46) and (47), respectively. RAM1in = Regout * Gain ... (46) RAM1in = RAM1out / (x-BC) ... (47)

In this way, the conversion table for this field is created. However, if this is used as it is for the conversion of the next field, it will be affected by the flicker of the light source. Therefore, the time history calculation is performed with the conversion table in the previous field so that the table is updated with a time constant.

FIG. 32 shows the operation of the ALU 226 in the division processing for obtaining the transfer gain in step 12. The transmission of the luminance range of 0 to 60 is performed by the same division processing as in the above calculation of the peak preservation ratio. Gain is required. In step 12, the calculation of equation (47) is performed.

In this case, first, the sequencer 223 outputs the address data adr indicating the luminance range 0. As a result, the brightness data RAM1out in the brightness range 0 is read from the RAM 224, and this brightness data RAM1out is supplied to the comparator 239. In addition, the brightness data generator 227 outputs brightness data x corresponding to a section of brightness range 0.
Is output, and the black code BC is subtracted from the brightness data x by the subtractor 255 and is supplied to the multiplier 244. Then, in the multiplier 244, the output data of the subtractor 255, that is, x-BC is set in the register 238,
For example, 12-bit data b (11) to b (0) are multiplied,
The output data of the multiplier 244 is supplied to the comparator 239. In the comparator 239, the above-mentioned brightness data RAM1out is compared with the output data of the multiplier 244, and the comparison result is supplied from the comparator 239 to the division controller 240.

In this state, under the control of the sequencer 223, the division controller 240 first clears the register 238 for creating the transfer gain (b (11) -b).
(0) = 0), and then the MSB b (11) is set to “1”. Then, the division controller 240 sets b (11) to “0” when the output data of the multiplier 244 is larger than RAM1out based on the comparison result from the comparator 239.
The output data of the multiplier 244 is RAM1out.
When it is the following, b (11) is left as "1" and b (11) is confirmed. Hereinafter, the division controller 240 is b (10)-
"1" is sequentially set to b (0), and the same confirmation processing as in the case of b (11) described above is performed. The 12-bit data b (11) to b (0) thus determined are stored in the RAM 224 as the transfer gain RAM1in in the luminance range 0.

Thereafter, the sequencer 223 makes the luminance range 1 to 6
The address data adr indicating 0 is sequentially output, and the division controller 240 performs the same division processing as in the case of the luminance range 0 described above, and the transfer gains RAM1in of the luminance ranges 1 to 60 are sequentially obtained and stored in the RAM 224. Is stored.

(10) Steps 13 and 14: Time constant (L
PF) processing (see FIG. 26 (l)) In steps 13 and 14, time constant processing is performed so that the table is updated with a time constant. Further, this result is written in the RAM 205 as a final table that is actually referred to when converting the input image. For that purpose, the calculation of the expressions (48) and (49) is performed. Regin = RAM1out−RAM2out (48) RAM2in = Regout * K _T + RAM2out (49) In the formula (49), K _T is a time constant. The left side of the equation is the RAM 2in for the above reason. Here, RAM2out is output data of the RAM205, and RAM2in is input data of the RAM205.

The transfer function of this LPF calculation is expressed as shown in equation (50). The sampling frequency in this equation (50) is the field frequency. G (z) = K _T ⁻¹⁻ (1−K _T ) z ⁻¹ (50) In this way, the final table is created in the RAM 205.

FIG. 33 shows the operation of the ALU 226 in the time constant processing of step 13. This step 1
In the case of 3, + RAM2 of the expressions (48) and (49)
Calculations are performed excluding out. Therefore, the adder / subtractor 237 of the ALU 226 functions as a subtractor. Also, FIG.
Is the ALU 226 in the time constant processing of step 14.
The operation of FIG. In this step 14, the remaining +
Calculation of RAM2out is performed. Therefore, ALU22
The adder / subtractor 237 of 6 functions as an adder.

(11) Step 15: RAM clearing process In step 15, the RAM 224 is cleared in preparation for the histogram in the effective pixel period of the next field. FIG. 35 shows the operation of the ALU 226 in the clear process of the RAM 224 in step 15.

As described above, in the first embodiment shown in FIG. 1, the knee compression processing is the luminance knee processing for performing the knee compression for the luminance level and the channels that are still over the luminance knee processing. , It consists of saturation knee processing to reduce the saturation until the level of the channel falls within the standard. Therefore, the high-luminance portion can be compressed without changing the hue and without exceeding the channel level.

Further, in the first embodiment, the channel level limit value CM in the saturation knee processing is set to, for example, 110% or more to relax the channel level limit, thereby significantly increasing the high luminance range. To get colored in, D
It has a CC plus function. Therefore, there is an advantage that the coloring in the high luminance range can be obtained without changing the hue.

In the first embodiment, the white clip is applied to the luminance, and when the channel level is exceeded, the saturation knee processing is used to deal with the situation. Therefore, the white clip process can be performed without changing the hue.

Further, in the first embodiment, flare correction is performed by performing control for reducing the gain kw of the luminance in the gradation region in which black floating occurs, and flare correction that does not affect color. It can be performed. And
In the first embodiment, this flare correction is automatically performed by adaptive gradation conversion by histogram equalization, and there is an advantage that a correction operation is performed according to the occurrence of flare.

Further, in the first embodiment, the adaptive gradation conversion by the histogram equalization is performed on the brightness level, and the unused gradation range is preferentially compressed, and accordingly, the dynamic range is dynamically calculated. The range can be used effectively. In this case, the histogram equalization strength K _WC can be specified to adjust the histogram equalization (Equation (34) and FIG.
0 (d)), it is possible to perform adaptive gradation conversion by optimal histogram equalization. Further, at the stage where the histogram equalization has been adjusted, the black code offset B _OF is subtracted from the whole to perform the black code storage processing (FIG. 1).
0 (e)). Therefore, the consistency of the black code is ensured, and the brightness conversion calculation of the brightness conversion calculator 111 performed on the stimulus value can be favorably performed. Furthermore, at the stage where the black code storage processing is completed, the peak storage processing is performed so that the brightness level of the input brightness A does not change even by conversion (see FIG. 10F). Thereby, for example, when the illumination condition is good and the histogram of the section of the normal light amount is large, by storing the brightness level of the section, it is possible to prevent inconvenience such as the knee compression of the video signal of the section.

Here, in the case where the channel level is exceeded after the adaptive gradation conversion by the histogram equalization is performed on the luminance level, the saturation knee process is used to deal with it. The tonal conversion can be performed without changing the hue.

Further, in the first embodiment, the adaptive gradation conversion by the histogram equalization is performed on the brightness level, and more gradations are allocated to the brightness area actually existing in the screen. Thus, the color video signal can be compressed. If the channel level is exceeded, the saturation knee processing is used to deal with this, and adaptive gradation conversion using a histogram can be performed without changing the hue.

Further, in the first embodiment, the user can set the saturation gain kcn, and priority is given to the processing of reducing the saturation by the saturation knee processing, but in other cases, the user sets the saturation. The saturation can be freely adjusted by the gain kcn of the degree.

In the first embodiment shown in FIG. 1, the controller 1 is used to minimize the delay of the circuit involved in the calculation of the gain kw of luminance and the gain kc of saturation.
24, the luminance W and the MAX obtained from the color data R, G, B before the luminance conversion calculation is performed as shown in FIG.
The luminance gain kw1 is multiplied by (R, G, B) to obtain the data W ', MAX' for obtaining the saturation gain kc. The reason why the circuit delay is minimized in this way is that there is a main line system that passes through the color correction circuit 108 in parallel with this system and a system of the image enhancer 109, and the respective total delays must match, so kw This is because if there is a large amount of delay in the system that obtains or kc, it is necessary to insert a delay circuit into the other system described above to take timing.

FIG. 36 shows a main part of a video camera device 100A according to the second embodiment. Apart from the above-mentioned problem of delay, the luminance conversion calculation is performed and the result is used in the saturation conversion calculation. The luminance Wi and the gain kc of the saturation are obtained. 36, parts corresponding to those in FIG. 1 are designated by the same reference numerals, and detailed description thereof will be omitted.

In the video camera device 100A shown in FIG. 36, the 2fs1 rate of red, green, and blue color data output from the up converters 107R, 107G, and 107B is supplied to the luminance conversion controller 124a. Then, in the controller 124a, the gain gain kw used in the luminance conversion calculator 111 is formed by a circuit equivalent to the matrix circuit 201 and the luminance gain generator 202 in FIG. In addition, the 2fs1 rate of red, green, and blue stimulus values output from the luminance conversion calculator 111 are supplied to the saturation conversion controller 124b. In the controller 124b, the luminance Wi and the saturation gain kc used in the saturation conversion calculator 112 are calculated by a circuit equivalent to the matrix circuit 201, the maximum value circuit 210, and the saturation gain generator 216 in FIG. It is formed.

Although not described above, the following problems arise when adaptive gradation conversion by histogram equalization is adopted. When the histogram is concentrated in a specific brightness range, especially in the dark area. Lighting conditions are good, and there is almost a histogram in normal light intensity.

If the histograms are concentrated in a specific area, the differential gain is significantly increased, which may impair the S / N ratio. When the diaphragm is closed, the gain increases near black, which is the most difficult condition for a video camera, and the image quality is impaired. Therefore, the microcomputer 125 may obtain information on the luminance region in which the histogram is concentrated and narrow down the histogram equalization strength K _WC . Then, when the luminance region in which the histograms are concentrated is near black, it may be more narrowed down.

In addition, the histogram read from the RAM 224 is stored in the register of the histogram information reporting circuit (not shown) in the sequence 6 "histwr" of step 2 of the operation step of the sequencer of Fig. 16. Thereby The histogram information reporting circuit compares the values of the histogram sequentially stored in the register,
For example, the information of the luminance range corresponding to the four histograms from the largest one is obtained. Then, from this reporting circuit, the information of the luminance range corresponding to the four histograms is reported to the microcomputer 125 from the larger one.

On the other hand, when the illumination condition is good, the luminance peak storing process is required as described with reference to FIG. Therefore, it is necessary to specify the level A for fixing the brightness by software. Although not described above, the control value of the auto iris should be given to this level A. The control value of the auto iris is the brightness representative of the image extracted from the image, and the diaphragm is moved so that it becomes equal to the set value. If the control value of the auto iris is used to specify the level A, there is a merit that the luminance targeted by the auto iris system is reproduced as it is even by the histogram equalization processing.

In the above embodiment, the amplitude transfer characteristic for performing adaptive gradation conversion by histogram equalization is created based on the cumulative frequency distribution (histogram table) detected during the video period of the previous field. However, it goes without saying that it may be created based on the cumulative frequency distribution detected during the video period of the preceding plurality of fields.

[0196]

According to the present invention, the conversion information indicating the correspondence between the level of the input video signal and the level of the output video signal is generated based on the cumulative frequency distribution of the input video signal, and this conversion information is used to convert the input video signal. This is to obtain the output video signal by converting the signal level, and it is possible to obtain the output video signal in which the unused gradation area of the input video signal is preferentially compressed, and the limited dynamic range is effectively used. Can be used.

When the input video signal is at a predetermined level, the conversion information is generated so that the output video signal obtained by converting the input video signal by the conversion information has the above-mentioned predetermined level. It is possible to obtain the output video signal in which the unused gradation area of the input video signal is preferentially compressed while the signal level of the predetermined level is stored.
Thereby, for example, when the illumination condition is good and the histogram of the section of the normal light amount is large, by storing the brightness level of the section, it is possible to prevent inconvenience such as the knee compression of the video signal of the section.

[Brief description of drawings]

FIG. 1 is a block diagram showing a video camera device according to a first embodiment.

FIG. 2 is a diagram for explaining a spatial pixel shifting method.

FIG. 3 is a block diagram showing a luminance conversion calculator.

FIG. 4 is a block diagram showing a saturation conversion calculator.

FIG. 5 is a diagram showing frequency characteristics of a low pass filter (LPF) and an interpolation filter (IPF).

FIG. 6 is a diagram showing a relationship between R, G, B levels and color.

FIG. 7 is a diagram showing a relationship between R, G, B levels and colors.

FIG. 8 is a diagram showing a relationship between R, G, B levels and color.

FIG. 9 is a diagram for explaining adaptive gradation conversion by histogram equalization.

FIG. 10 is a diagram for explaining adaptive gradation conversion by histogram equalization.

FIG. 11 is a block diagram showing a detailed configuration of a controller.

FIG. 12 is a diagram showing an example of section division of a luminance region.

FIG. 13 is a diagram for explaining an interpolation calculation for obtaining a gain kw1 of luminance.

FIG. 14 is a block diagram showing a circuit portion for obtaining a brightness gain kw, a saturation gain kc, and a brightness Wi of a controller.

FIG. 15 is a block diagram showing a circuit portion related to table creation of a controller.

FIG. 16 is a diagram showing operation steps for creating a table of the sequencer of the controller.

FIG. 17: Step 0: AL in histogram acquisition
It is a figure for demonstrating operation | movement of U.

FIG. 18: Step 1: AL in accumulation and normalization
It is a figure for demonstrating operation | movement of U.

FIG. 19: Step 2: Adjusting histogram equalization (1)
FIG. 6 is a diagram for explaining the operation of the ALU in FIG.

FIG. 20: Step 3: Adjustment of histogram equalization (2)
FIG. 6 is a diagram for explaining the operation of the ALU in FIG.

FIG. 21: Step 4: AL in black code storage processing
It is a figure for demonstrating operation | movement of U.

FIG. 22: Step 4: A when calculating peak conservation ratio
It is a figure for demonstrating operation | movement of LU.

FIG. 23 is a diagram for explaining the operation of the ALU in step 5: peak saving processing (1).

FIG. 24 is a diagram for explaining the operation of the ALU in step 6: peak storage processing (2).

FIG. 25 is a diagram for explaining knee compression, white clip, and total gain adjustment.

FIG. 26 is a diagram for explaining division and the like for obtaining a transfer gain.

FIG. 27: Step 7: A in knee compression processing (1)
It is a figure for demonstrating operation | movement of LU.

FIG. 28: Step A: A in knee compression processing (1)
It is a figure for demonstrating operation | movement of LU.

FIG. 29: Step 9: A in knee compression processing (2)
It is a figure for demonstrating operation | movement of LU.

FIG. 30 is a diagram for explaining the operation of the ALU in step 10: knee compression processing (2).

FIG. 31 is a diagram for explaining the operation of the ALU in step 11: total gain adjustment.

FIG. 32 is a diagram for explaining the operation of the ALU in the division process for obtaining the transfer gain.

FIG. 33 is a diagram for explaining the operation of the ALU in step 13: time constant processing.

FIG. 34 is a diagram for explaining the operation of the ALU in step 14: time constant processing.

FIG. 35 is a diagram illustrating step 15: RAM clearing processing.

FIG. 36 is a block diagram showing a main part of a video camera device according to a second embodiment.

FIG. 37 is a diagram showing a television system.

[Explanation of symbols]

100 ... Video camera, 103R, 103G, 10
3B ... CCD solid-state image sensor, 104R, 104G,
104B ... Analog process circuit, 105R, 10
5G, 105B ... A / D converter, 106 ... Preprocess circuit, 107R, 107G, 107B ... Upconverter, 108 ... Color correction circuit, 109 ...
・ Image enhancer, 110R, 110G, 110B
... Subtractor, 111 ... Luminance conversion calculator, 112 ...
..Saturation conversion calculators 117R, 117G, 117B,
119R, 119G, 119B ... Adder, 118
R, 118G, 118B ... Gamma correction circuit, 123
R, 123B ... Low pass filter, 123G ...
Interpolation filter, 124 ... Controller, 125 ...
-Microcomputer, 201 ... Matrix circuit, 202 ... Luminance gain generator, 203, 221, 2
22 ... Up converter, 204 ... Pixel averaging circuit, 205, 224 ... RAM, 206 ... Section generator, 207 ... Address generator, 208, 212,
215, 225 ... Switch circuit, 209 ... Interpolation calculator, 210 ... Maximum value circuit, 211 ... Luminance gain multiplier, 213, 217, 218 ... Subtractor, 21
4 ... Multiplier, 219 ... Divider, 220 ... Minimum value circuit, 223 ... Sequencer, 226 ... Arithmetic logic unit, 227 ... Luminance data generator

Claims (19)

[Claims] 1. A cumulative frequency distribution detecting means for detecting a cumulative frequency distribution of an input video signal, and a first conversion indicating a correspondence between the level of the input video signal and the level of the output video signal based on the cumulative frequency distribution. First conversion information generating means for generating information, and an output video signal obtained by converting the input video signal with the first conversion information when the input video signal has a predetermined level has the predetermined level. By adjusting the first conversion information as described above, the second conversion information generating means for generating the second conversion information and the signal level of the input video signal are converted based on the second conversion information. A video camera device comprising level conversion means. 2. The video camera device according to claim 1, further comprising a high-intensity compression unit that compresses a high-intensity component in the output video signal output from the level conversion unit. 3. The cumulative frequency distribution detecting means detects the cumulative frequency distribution during a video period of the input video signal, and the first and second conversion information generating means include a vertical block of the input video signal. 2. The first and second conversion information is generated during the ranking period based on the cumulative frequency distribution detected by the cumulative frequency distribution detecting means during the preceding video period. Video camera equipment. 4. A plurality of the first and second conversion information generating means, which are detected by the cumulative frequency distribution detecting means during a plurality of preceding video periods during a vertical blanking period of the input video signal. The video camera device according to claim 1, wherein the first and second conversion information are generated based on a cumulative frequency distribution. 5. A cumulative frequency distribution detecting means for detecting a cumulative frequency distribution of an input video signal, and a first conversion indicating a correspondence between the level of the input video signal and the level of the output video signal based on the cumulative frequency distribution. First conversion information generating means for generating information, and an output video signal obtained by converting the input video signal with the first conversion information when the input video signal has a predetermined level has the predetermined level. By adjusting the first conversion information as described above, the second conversion information generating means for generating the second conversion information and the signal level of the input video signal are converted based on the second conversion information. A video signal processing device, comprising: a level converting means. 6. The video signal processing apparatus according to claim 5, further comprising high brightness compression means for compressing a high brightness component with respect to the output video signal output from the level conversion means. 7. The cumulative frequency distribution detecting means detects the cumulative frequency distribution during a video period of the input video signal, and the first and second conversion information generating means include a vertical block of the input video signal. 6. The first and second conversion information is generated during the ranking period based on the cumulative frequency distribution detected by the cumulative frequency distribution detecting means during the preceding video period. Video signal processing device. 8. The first and second conversion information generating means include a plurality of a plurality of video signals detected by the cumulative frequency distribution detecting means during a plurality of preceding video periods during a vertical blanking period of the input video signal. The video signal processing device according to claim 5, wherein the first and second conversion information are generated based on a cumulative frequency distribution. 9. A first operation for detecting signal level distribution information relating to a signal level distribution of an input video signal during a valid video period of the input video signal, and the first operation during an invalid video period of the input video signal. The second operation of detecting the conversion information for converting the gradation of the input video signal based on the signal level distribution information of the input video signal detected by the operation of Control means for controlling the operation of the detection means in accordance with the timing of the input video signal; storage means for storing information including at least the signal level distribution information and the conversion information generated during the operation of the detection means; A video camera device, comprising: a gradation conversion unit that converts the gradation of the input video signal based on conversion information. 10. The control means obtains the conversion information by causing the detection means to perform the second operation with respect to the signal level distribution information stored in the storage means. The video camera device according to claim 9. 11. The control means performs, as the second operation of the detection means, between gradation conversion and gradation non-conversion corresponding to a cumulative frequency distribution obtained by accumulating the signal level distribution information. 11. The video camera device according to claim 10, wherein the video camera device is operated to obtain conversion information such that gradation conversion of the intensity of is performed. 12. The control means causes the detection means to perform, as the second operation, an operation of obtaining conversion information such that gradation conversion for a predetermined level of the input video signal is not performed. The video camera device according to claim 10. 13. The control means causes the detection means to perform, as the second operation, an operation of obtaining conversion information such that gradation conversion for a black code of the input video signal is not performed. Claim 10
The video camera device according to. 14. A first operation of detecting signal level distribution information regarding a signal level distribution of an input video signal during a valid video period of the input video signal, and the first operation during a invalid video period of the input video signal. The second operation of detecting the conversion information for converting the gradation of the input video signal based on the signal level distribution information of the input video signal detected by the operation of Control means for controlling the operation of the detection means in accordance with the timing of the input video signal; storage means for storing information including at least the signal level distribution information and the conversion information generated during the operation of the detection means; A video signal processing device, comprising: a gradation conversion means for converting the gradation of the input video signal based on conversion information. 15. The control means obtains the conversion information by causing the detection means to perform the second operation with respect to the signal level distribution information stored in the storage means. Claim 14
2. The video signal processing device according to 1. 16. The control means performs, as the second operation of the detection means, between gradation conversion and gradation non-conversion corresponding to a cumulative frequency distribution obtained by accumulating the signal level distribution information. 16. The video signal processing device according to claim 15, wherein the video signal processing device is caused to perform an operation of obtaining conversion information such that gradation conversion of the intensity of is performed. 17. The control means causes the detection means to perform, as the second operation, an operation of obtaining conversion information such that gradation conversion for a predetermined level of the input video signal is not performed. The video signal processing device according to claim 15. 18. The control means causes the detection means to perform, as the second operation, an operation of obtaining conversion information such that gradation conversion for a black code of the input video signal is not performed. Claim 15
2. The video signal processing device according to 1. 19. A first conversion information indicating a correspondence between the level of the input video signal and the level of the output video signal is generated based on the cumulative frequency distribution of the input video signal is detected, and the first conversion information is generated. By adjusting the first conversion information so that the output video signal obtained by converting the input video signal with the first conversion information when the input video signal is at the predetermined level has the predetermined level, 2. A gradation conversion method for a video signal, comprising the step of generating conversion information No. 2 and converting the signal level of the input video signal based on the second conversion information.