JPH0468663A - Color picture processor - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a color image processing device that reads a color image and performs processing suitable for the color image.

[Prior Art] Conventionally, when an original such as a newspaper is read and recorded, the removal of the background color of a color original is performed while an operator manually adjusts the density and visually confirms the recorded image after image processing. , the adjustment work was repeated until the base layer disappeared.

[Problems to be Solved by the Invention] However, in the above-mentioned conventional example, the adjustment was performed manually, which required time.

Further, when removing the background, the density of the entire image area becomes thinner and the contrast decreases, resulting in a very unsightly image. Further, when the light amount of the document exposure lamp is adjusted to adjust the density, the color tone of the lamp changes, and the color tone of the recorded image also changes accordingly.

The present invention was made to solve the above problems, and
It is an object of the present invention to provide a color image processing device that can perform processing on a read color image according to the background of the color image.

[Means for Solving the Problems] In order to achieve the above object, a color image processing device of the present invention has the following configuration. That is, the color image processing device calculates color image background information based on a read color image, performs color image processing according to the color image background information, and controls analog gain according to the color image background information. and a setting means for setting an analog gain controlled by the control means.

[Operation] In the above configuration, when calculating color image background information based on the read color image and performing color image processing according to the color image background information, the information set according to the color image background information is calculated. Operates to control analog gain.

[Embodiment] Hereinafter, a preferred embodiment of the present invention will be described in detail with reference to the accompanying drawings.

FIG. 1 is a schematic block diagram showing the configuration of a color image processing device in an embodiment.

A color image processing device applied to a color image copying device exposes a full-color original to light from an illumination source such as a halogen lamp or fluorescent lamp, and then photoelectrically converts the reflected color image using a color image sensor such as a CCD (storage charge device). The image is captured by conversion, and the obtained analog image signal is converted into an A/D (analog/
The digitized full-color image signal is digitized using a digital converter, etc., processed, processed, and sent to various color printers.

The document is first irradiated with a light amount according to a set light amount described later by the variable light amount exposure lamp 520, and the reflected light is reflected by the CCD.
500, each image is separated into colors and read, and then amplified to a predetermined level by an analog processing circuit 501.

FIG. 2 shows the color reading sensor and drive pulses.

Figure 2 (a) shows the color reading sensor used in this example, which should be read by dividing the main scanning direction into 5.
1024 pixels, where μm (1/16 mm) is one pixel,
In other words, as shown in the figure, one pixel is divided into 3 pixels in G, B, and R in the main scanning direction.
Since it is divided, the total number of effective pixels is 1024×3=3072. On the other hand, the chips 58 to 62 are formed on the same ceramic substrate, and the 1st and 3.5th sensors (58, 60, 62) are on the same line LA, and the 2nd and 4th sensors (59, 61) are on the same line LA. 4 lines (62.5 μm
x 4 = 250 μm) on the line LB, and scans in the force direction of the arrow AL when reading the document.

The 1st and 3.5th of each S CCDs are drive pulse group 0
The second and fourth are driven by the DRVI 18 and the EDRV 119 independently and synchronously. 0DRV1
18i: Includes 6001A.

EOIA included in 002A, ORS and EDRV119
, EO2A, and ERS are charge transfer lock and charge reset pulses within each sensor, respectively, and due to mutual interference between the 1st, 3.5th and 2nd, and 4th and noise limitations, they are designed to avoid jitter from each other. Generated completely synchronously. These pulses are therefore generated from a single reference oscillator source 05C558 (FIG. 1).

FIG. 3(a) shows the driving pulse 0 of the color reading sensor.
DRVI 18. The circuit block of the pulse generator 534 that generates the EDRVI 19, FIG. 3(b) is a timing chart thereof, and is included in the system control generator 534 of FIG. single 03c
0135 is a clock that generates reference signals 5YNC2 and 5YNC3 that determine the generation timing of 0DRV and EDRV (7). 5YNC2 and 5YNC3 are CPUs
The output timing is determined according to the set value of the presettable counter 64.65 set by the signal line 139 connected to the 22 bus, and the 5YNC2 and 5YNC3 operate the frequency divider 66.67 and the drive pulse generator 68.69. initialize. That is, H3YNC115 input to this block
0DRV118 and EDR.
Each pulse group of VII9 is obtained as a synchronous signal with absolutely no jitter, and signal disturbances due to interference between sensors can be prevented.

Here, the sensor drive bars/1. which are obtained in synchronization with each other.
0DRV118 and EDRVII are CCD driver 5
1,3.5th sensor 58.60.62 through 33
Then, the dregs are supplied to the second and fourth sensors 59 and 61, respectively. Eight video signals 1 to v5 are independently outputted from each sensor 58, 59, 60, 61', and 62 in synchronization with the drive pulse.

The voltage is then amplified to a predetermined voltage value by independent analog processing circuits 501-1 to 501-5 for each channel shown in FIG.
VI V3. at the timing of O3129. V5 is EO5
At the timing of 134, V2. The V4 signal is sent out and input to the sample and hold circuit S/H502.

The color image signals obtained by dividing the original input into the S/H circuit 502 into 5 parts and reading them are G (green), G (green),
It is separated into three colors: B (blue) and R (red). Therefore, after S/H, 3×5=15 signals are processed.

The analog color image signals sampled and held for each color R, G, and B by the S/H circuit 502 are sent to the next stage A/D.
Each channel from 1 to 5 is digitized by a conversion circuit,
Each of channels 1 to 5 is independently outputted in parallel to the next stage.

Now, in this embodiment, as mentioned above, 4 lines (62
, 5 μm x 4 = 250 μm) in the sub-scanning direction and divided into 5 areas in the main scanning direction. , 3.5, the reading position is off.

Therefore, in order to connect these lines correctly, a shift correction circuit 505 having memory for a plurality of lines corrects the shift.

Next, the blue signal black level correction circuit 77 in FIG. 4(a)
The black correction operation will be explained with reference to the block diagram B. Fourth
As shown in Figure (b), the black level outputs of channels 1 to 5 have large variations between chips and pixels when the amount of light input to the sensor is small. If you output this as is and output it as an image,
Streaks or unevenness appear in the data part of the image. Therefore, it is necessary to correct the output variation of this black part, as shown in Figure 4 (a).
Correction is performed using a circuit like this. Prior to the original reading operation, C
The original scanning unit including the CD 500 is moved to the position of the black plate having uniform density, which is placed in the non-image area at the tip of the original table, the halogen 520 is turned on, and a black level image signal is input to this circuit.

Regarding the blue signal Bin, when reading the black board,
One line of this image data is stored in the black level RAM 78.

On the other hand, when reading an image, the above-mentioned RAM 78 is in a data read mode, and the data is read line by line and pixel by pixel and inputted to the B input of the subtracter 79 via the data line 153→157 path. That is, at this time, the gate 81 is closed (■) 80
opens (■). Further, the selector 86 becomes an A output.

Therefore, the black level output 156 is the black level data DK
For (i), for example, in the case of a blue signal, Bin (i)
−DK (i)=Bout (i) (referred to as black correction mode). Similarly, Green Gin, Red R
Similar control is also performed for in by 77G and 77R.

In addition, the control lines of each selector gate for this control
◎, ■, ■ are performed under CPU control by the latch 85 assigned as Ilo of the CPU 22 (FIG. 1).

Note that by selecting B on the selectors 82, 83, and 86, the RAM 78 can be accessed by the CPU 22.

Next, white level correction (shading correction) in the black correction/white correction circuit 506 will be explained with reference to FIG.

White level correction is based on the white data when the document scanning unit is moved to the position of a uniform white plate and the illumination system,
Corrects sensitivity variations in the optical system and sensors. The basic circuit configuration is shown in FIG. 5(a). This basic circuit configuration is the same as that in FIG. 4(a), but for black correction, the subtracter 79
In contrast, in color correction, the multiplier 79
The only difference is that ′ is used, and the explanation of the same parts will be omitted (.

First, during color correction, when the CCD 500 for reading the original is at the reading position (home position) of the uniform white plate, that is, before the copying or reading operation, the exposure lamp 520 is turned on, and the image data at the uniform white level is is stored in the correction RAM 78' for one line. For example, if it has an A4 longitudinal width in the main scanning direction, 16pe
1/mm = 16 x 297 mm = 4752 pixels, which is small (the capacity of the RAM 78' is 4752 bytes).

As shown in Figure 5(b), the i-th pixel white board data is
(i=1 to 4752), the fifth
As shown in Figure (c), data for the white board is stored for each pixel.

On the other hand, for Wi, the corrected data Da=Dix FF H/ at the read value Di of the normal image of the i-th pixel
It should be W 1. Therefore, the CPU 22 sends the gate 80' to the latch 85'■', ■', ■', ■'.
, 81', and selectors 82', 83',
B is selected at 86' and output, and RAM 78' is set to C.
Enable PU access. Next, according to the flowchart shown in FIG. 5(d), the CPU 22 performs data replacement by sequentially calculating FF for the first pixel W0, 4/W, FF, /W for W, etc. . When the blue component of the color component image is completed (step S1), the green component (step S2) and the red component (step S3) are similarly sequentially processed. From now on, the input original image data D
D o ” D I XF F o / W for i
Gate 80' is open so that i is output (■1 and 81 are closed (■'), selector 83'86' selects A, and R
Coefficient data F F H/ read from AM78'
W i passes through signal lines 153→157, is multiplied by the original image data 151 input from one side, and is output.

Before explaining the color correction circuit (1) 507, document background removal will be explained. The image reading unit starts bliscanning in order to detect the original by the copy start operation. That is, during the initial movement in the forward direction of Bliscan, the document edge detection circuit 21 detects the document edge coordinates based on the signal from the image reading unit based on the reflected brightness of the document and the scanning position, thereby detecting the document position and Identify shapes. and,
When the briscan returns to the home position side, the CPO 22 and the like calculate the reflected luminance of the document based on the data from the color correction RAM 78' and stores it in the RAM 24 as a histogram (FIG. 10). Here, based on the detected edge coordinates of the document, 50 points in the main scanning direction and 50 lines in the sub-scanning direction from within the document range, a total of 25
From the data of pixel 00, the B, G, and H data of the n-th pixel taken into the calculation unit 220 are set as Bn, Gn, and Rn, and the document reflection brightness Dn is determined by the following equation.

Dn=a-Bn'+b-Gn'c-Rn' However, a+ b+ C+ l+ J+ k is a predetermined constant, for example, a=b=c=1/
3, using values such as i = j = k = 1.

Then, for the original background reflection brightness, if the number of original reflection brightness data existing within the range of the preset reflection brightness variation width from the maximum original reflection brightness in 2500 pixel data is, for example, 30% of the total (arbitrary value ), the maximum original reflected brightness is determined to be the original background reflected brightness, while if it is less than 30%, the reflected brightness of the white plate is determined to be the original original background reflected brightness.

Therefore, the background removal process is not performed on photographs and full-page images (30% or less), while the background removal process can be performed on originals containing two images of text (30% or more).

In this manner, in the above-described embodiment, the CPU 22 stores the original reflected brightness in the RAM 24, and in the case of a full-page image, the maximum original reflected brightness is set as the original background reflected brightness.

In addition, by substituting the minimum value of each of B, G, and R of 2,500 pixels into the formula for calculating Dn described above and using the original reflection brightness as the original background reflection brightness, it is possible to eliminate errors in background color removal due to variations in color components. can be prevented. For this purpose, the CPU 22 uses the RAM 7 for original reflection brightness and white correction.
The output data of 8' is stored in the RAM 24. If it is a full-page image, Bn corresponding to the original reflected brightness that is within the range of the reflected brightness variation width from the maximum original reflected brightness,
The CPU 22 or the like retrieves the data of Gn and Rn from the RAM 24, and uses the minimum value of each Bn, Gn, and Rn as the original reflection brightness obtained based on the above-described formula for calculating the original reflection brightness Dn as the original background reflection brightness.

Next, an algorithm for background removal aimed at high-speed automation will be explained with reference to FIG. The base level of the document is detected using the return operation of the document detection scan (step 511) that detects the size of the document (step 512).
The background level is removed by switching the gain of the analog circuit system according to the background level (step 513), switching the logarithmic conversion characteristic (step 514), and performing image processing (step 515).

The analog circuit system described above will now be described in detail with reference to FIG. Since the image signal from the CCD 500 is a minute analog signal, it is amplified to a predetermined value by the analog processing circuit 501.

After that, the S/H, A/D conversion circuit 502 converts the 15th
Perform the processing shown in the figure. First, a signal is input as B, G, R serial data.

Since the force ratio of R is not 1:1:1 when irradiating blank paper, etc., the serial attenuator 1001 is used to
, R are changed to a predetermined ratio. After that, A
The signal is passed through the S/H circuit 1002 as a pre-stage to the /D conversion, and here also serves to clamp the signal at the same time.

Next, in order to make it possible to change the signal to any amplitude, the signal is passed through a variable gain analog amplifier 1003, passed through a clamp circuit 1004 again, and passed through an offset variable circuit 1005 that adjusts the offset of the signal, after which it is A/D converted. The signal is input to the deviation correction circuit 505.

Here, the gain of the analog system can be easily set by the CPU 22 controlling the variable gain circuit 1003.
The gain is set so that the background level is the maximum digital value, that is, FF. The light amount setting changes continuously with respect to the background level data, and the relationship between the setting value calculated by the CPU 22 and the gain changes linearly.

Note that since the analog system has five independent channels, the gain settings for each channel are calculated separately. Furthermore, for a full-page image original, the reflected brightness of the white plate is used as the original reflected brightness, so the light amount remains the same as when the white level was corrected.

At this time, as the gain increases, the color tone of the image also changes in a certain density range, so in order to correct this phenomenon, in this embodiment, adjustments are made by switching the LUT (look-up table) for logarithmic conversion, which will be described later. shall correspond.

As mentioned above, the black level sensitivity of the image input system.

CCD dark current variation, sensitivity variation between each sensor,
By correcting the black level and white level based on various factors such as optical system light intensity variations and white level sensitivity, the black level and white level are made uniform in the main scanning direction, and color image data proportional to the input light intensity is easily understood by the human eye. The signal is input to a logarithmic conversion circuit 508 (FIG. 1) in order to match the luminosity characteristic.

Here, white = OOH, black = FFH (the image source that is converted and further input to the image reading sensor 500, for example, a normal reflective original, a transparent original such as a film projector, or even the same transparent original) , negative film, positive film, the sensitivity of the film, the exposure condition, etc. Since the input gamma characteristics differ, as shown in Fig. 6(a),
As shown in (b), there are multiple LUTs (lookup tables) for logarithmic conversion, which are used depending on the purpose.This switching is performed by the CPU 22 on the signal line βgO2.
This is done by controlling βgl and ρg2 (160-162). Here, B, G.

Each data added to R corresponds to the density value of the output image, and is represented by B (Blue', G (Green), R (
Each signal (red) corresponds to the amount of yellow, magenta, and cyan toner, respectively, so the subsequent color image data is Y and M.

Correspond to C.

The CPU 22 selects an appropriate LOT so that the tone of the recorded pixel does not change as the gain for background removal increases. This makes it possible to cope with the phenomenon in which the hue of an image changes due to an increase in gain.

Note that the color correction circuit (1) 507 is a circuit that detects a specific color from input color image data R, B, and G and replaces it with another color. For example, it realizes the function of converting a red part of a document into blue or any other arbitrary color.

Next, the color correction circuit (2) 509 performs color correction on each color component image data from the original image obtained by logarithmic change, that is, yellow component, magenta component, and cyan component, as described later. As shown in Fig. 7, the spectral characteristics of the color separation filters arranged for each pixel in the color reading sensor have unnecessary transmission areas such as the shaded areas, and on the other hand, for example, the color transferred to the transfer paper It is well known that toner (Y, M, C) also has unnecessary absorption components as shown in Fig. 8. Therefore, color component image data Yi.

Masking correction is known in which color correction is performed by calculating a linear equation for each color for Mi and Ci.Furthermore, for Yi and Mi.

Min by Ci (Yi, Mi, C1) (Yi.

The minimum value of ML and Ci) is calculated, and this is used as a smear (black), and then black toner is added (smear removal) and the amount of each coloring material added is reduced according to the added black component (undercolor removal). UCR) operations are also well performed. Figure 9(a) shows masking and inking.

The circuit configuration of a color correction circuit (2) 509 that performs LICR is shown. The features of this configuration are as follows.

■It has two masking matrices and can be switched at high speed with one signal line "Ilo".

■The presence or absence of OCR can be switched at high speed with a single signal line "Ilo".

■It has two circuits that determine the amount of smear, and can be switched at high speed with °゛■10''.

First, prior to image reading, a desired first matrix system M3. A second matrix coefficient M2 is set via a bus connected to the CPU 22. In this embodiment, M is set in registers 87-95, and M2 is set in registers 96-104.

Further, 111 to 112, 131, 135, and 136 are selectors, respectively, which select A when the S terminal is “1” and select “
When matrix M is selected, B is selected. Therefore, when matrix M is selected, switching signal MAREA364 is set to "1", and when matrix M2 is selected, switch signal MAREA364 is set to "0".

Further, 123 is a selector, and the selection signal C8°C, (3
36, 367), the equations a, b, and c are obtained based on the truth table of FIG. 9(b). Selection signals C6, C1 and C
2 corresponds to the color signal to be output, for example Y, M,
In the order of C and B, (C,,C,,C,)=(0,0,
O), (0°0.1), (0,1,O), (1,O,O
) Further, by setting (0, L 1) as a monochrome signal, a desired color-corrected color signal is obtained. Now (C,,
C,,C,)=(0,O,O) and MAREA=”
1'', the outputs (a, b, c) of the selector 123 contain the contents of the registers 87, 88°89, and therefore (ay
+ + -by, -Cc+) are output.

On the other hand, M i n(Yi
, Mi, C1)=black component signal 374 calculated as
At 137, Y=ak-b (a.

b is a constant) and undergoes a -time conversion (passes through selector 135) and is input to the B inputs of subtracters 124, 125, and 126. Each subtractor 124 to 126 uses Y as undercolor removal.
= Y i (ak − b), M = M 1
(ak-b), C=C1-(ak-b) is calculated,
The signals are input via signal lines 337, 378, and 379 to computing units 127, 128, and 129 for masking operations. The selector 135 is controlled by the signal UAREA 365, and the UAREA 365 has a configuration that enables high-speed switching between the presence and absence of UCR (undercolor removal) at "I10".

Multipliers 127, 128, and 129 each have (ay+ + -by+ + -Ccl) t at their A inputs.
The B input has the above-mentioned [Yi - (ak-b), Mi (
ak-b) Ci-(ak-b)] = [Yi,
Mi, C1] is input, so as is clear from the figure, the output Dout has the condition of C2=0 (Y.

(selection of M, C), You u t=Y i X (ay
+) +MiX (-bv+) 10cix (-C
cl) is obtained, and yellow image data that has been subjected to masking color correction and undercolor removal processing is obtained. Similarly, Mout=Yi×(-ay□)+MiX (bvz)
+CiX (-ccz)Cout=YiX (-ay
3) + MiX (b+va) + cix (C
CV) is output to Dout. Color selection is controlled by the CPU 22 according to the table of FIG. 9(b) from (Co, CC2) in accordance with the order of output to the color printer. Registers 105 to 107 and 108 to 110 are registers for monochrome image formation, and based on the same principle as the masking color correction described above, MONO=+Yi+ff,
Weighting of each color by Mi+m+Ci is obtained by addition.

As mentioned above, the switching signal MAREA364 is for high-speed switching between the masking color correction coefficient matrices M1 and M2, and the UAREFA365 is for high-speed switching between the presence and absence of UCR.
KAREA 387 switches the primary conversion of the black component signal (outputted to Dout through the signal line 369 and selector 131), that is, K=Min (for Yi, Mi, and Ci, Y=ek-d or Y= ek-f (c, d, e,
f is a signal that rapidly switches the characteristics of a constant parameter).

As explained above, according to this embodiment, the background level of a document containing both images and text is removed quickly and automatically, the image is the same as the original, the text is processed clearly, and the quality is high. It becomes possible to obtain the output of

Therefore, by removing the base, color bleeding at the edges of black characters and thin black lines, as well as black characters and thin black lines, caused by a black character processing circuit, which will be described later, is improved. For example, the background color of newspapers, etc. is removed, the characters are clearly expressed, and the image can be reproduced as it was before the background color was removed.

Next, the above-mentioned black character processing circuit will be explained with reference to FIGS. 11 and 12.

As explained in FIGS. 4 and 5, R, B, and G (red) have their black and white levels corrected.

green, blue) color signals 113, 114° 115
After undergoing logarithmic transformation 508, masking, and undercolor removal 509, the color signals to be output to the printer are selected and output to the signal line 401. Parallel to this, signals R, G, and B are applied to the achromatic portion of the document and the edge portion (
In other words, black text.

In order to detect the black thin line), luminance signal Y1 color difference signal 1. Q is calculated and compared by a Y, I, and Q calculation circuit 535.

The luminance signal ¥392 is for extracting the edge signal (
In order to calculate a 5×5 matrix using a known digital second-order differential circuit 538, a line buffer circuit 5 for 5 lines is used.
37, and as described above, the Laplacian calculation is performed in the calculation circuit 538. That is, when the human-powered luminance signal Y is a step-like input (for example, a character part) as shown in (i) shown in FIG. 12(d), the output 406 after the Laplacian calculation is as shown in FIG. (hereinafter referred to as edge signal).

Lookup table LUTA539. LUTB540
is the printing amount (
For example, the table is a lookup table for determining the amount of toner, and is composed of lookup tables each having characteristics as shown in FIGS. 12(a) and 12(b). That is, when LUTA acts on the edge signal 406, the amplitude increases as shown in (1ii) of FIG. 12(d).

This determines the amount of black toner at the black edge portion, as will be described later. Furthermore, when LUTE is applied to the edge signal 406, the absolute value appears as negative, and this is due to the Y, M
, C (yellow, magenta.

Determine the amount of toner (cyan). This is shown in Figure 12 (d
), and by passing through the smoothing (averaging) circuit 543, it becomes a signal as shown in (V) of the figure.

On the other hand, the achromatic color detection circuit 536 outputs, for example, the 12th
This circuit applies pressure to a signal 393 according to the characteristics shown in the figure. This signal 393 is selected by the selector 553 by the signal 407 which becomes 1" when printing black toner, and the signal 398 is selected by the selector 553.
becomes.

Then, after being multiplied by the aforementioned signal 394 (FIG. 12(d) (1ii)) which determines the amount of black toner in a multiplier 548, it is added to the original image signal in an adder 547.

Also, when printing with Y, M, C (yellow, magenta, cyan) toner, Y is printed on black characters and black lines.

Since it is desirable that M and C toners not be printed, the selector 55 selected by the color selection signal 407
3, "l" is output to the multiplier 548, while the selector 544 outputs a signal obtained by smoothing the output of the LUTB 540 (FIG. 12(d) (V)), and multiplication is performed. As a result, in the adder 547, as shown in FIG.
The same signal as (V) is input, and only the signal at the black edge portion is subtracted from the original signal.

In other words, this means that the signal that determines the amount of black toner for the black edge portion is strong, which means that the amount of black toner is increased.
By reducing the amount of Y, M, and C toner for the same area, black areas can be rendered blacker.

The signal 397 obtained by binarizing the achromatic color signal 393 by the binarizing circuit 546 becomes "1" when the color is achromatic, and "O" when the color is chromatic. That is, as described above, the selector 544 selects black toner printing. (when 407 = "1"), S input = "1" and six inputs, ie.

A signal 394 (FIG. 12(d) (1ii)) is output to emphasize the black edges. However, when printing with Y, M, and C toners (when 407="0"), the signal 397="1"
, Therefore, if it is an achromatic color, S manual power is selected to reduce the amount of toner of Y, M, and C as described above (Fig. 12(d)).
(v) is output, but in the case of chromatic color, signal 397
="0", therefore 397="1", i.e. selector 54
The S human power of 4 becomes “1” and A is selected, and as shown in Fig. 12 (
d) The signal of (1ii) is output to the adder 547 and becomes the usual well-known edge enhancement.

The LUTA539 includes a LUT that becomes zero when the edge signal value is less than or equal to ±n, as shown in FIG.
There are two types of LUTs that are set to zero below, and the value to be clamped to zero is selected depending on the level of the original signal 401, that is, the density of the document at this time. When the density level of the original is higher than the value set by the CPU 22 via the bus 422, that is, when it is dark, the output of the comparator 590 becomes "1", as shown in FIG. 12(a).
The LUT clamped to zero at A' and B' shown in
Furthermore, when the concentration is below a certain level, that is, when the output of the comparator 590 is "O", the LUT is clamped to zero at A and B.
The effect of noise removal is changed depending on the concentration range by selecting .

Furthermore, the output 400 of the AND gate 552 has been further improved for the edges of black characters; 403 (A input) is used when printing Y, M, and C for the edges of black characters, and 403 (A input) is used for other cases. This is the signal to select (switch) 402 (S manual power).

The signal 396 input to the AND gate 552 is a signal obtained by applying the characteristics of LUTC (FIG. 12(C)) to the edge signal described above, and is binarized by the binarization circuit 545. “1” when the absolute value of the signal is greater than the specified value
, becomes "0" when:

Therefore, 396="l"397="1"399="L
” occurs when the edge signal is large for achromatic colors, that is, at the edge portion of the black signal, and only when printing with Y, M, and C toners. As shown above, the signal that determines the amount of Y, M, and C toner is subtracted from the original signal only at the portion corresponding to the black edge, and the remaining signal is smoothed by the averaging circuit 549, and the signal ER=
When it is "1"', it passes through the selector 551 and is output to 405. Otherwise, the normal edge-enhanced signal 40
2 is output.

The above-mentioned signal ER is controlled by the CPU 22, and ER=“
1”, the output of the averaging circuit 549 is output to the output 405,
When R="0", "0" is output to the output 405. This completely sets the signals of the color toners (Y, M, C) around the edges of the black characters to "0" to further eliminate the color blur, and these are selectable.

Next, 511 shown in FIG. 1 is a density conversion circuit, which is configured to be able to change the density gradation for each color as shown in FIGS. 14(a) and 14(b). It is made up of etc.

By performing the image processing described above, it is possible to quickly and automatically remove the background color of the original, and it is also possible to reproduce and process images with clear text and images that are identical to the original. .

As described above, in this embodiment, the CPU 22 prescans a color original, reads it into the CCD 500, and stores it in the RAM 24.
Take a brightness histogram.

Then, the CPU 22 determines whether the document is a full-page image document or a text image mixed document based on the variation around the maximum value of the histogram.

That is, if the data near the maximum luminance value is small, it is determined that the document is a full-page image document, and if there is a large amount of data, it is determined that the document is a mixed document. If it is a mixed document, the gain of the analog signal system is controlled using the maximum value of luminance as the background level. on the other hand,
If it is a full-page image, the standard white original read before pre-scanning is used as the background level, and the gain is not changed.

In other words, in the case of a mixed document, the gain is increased so that the output of the reading unit 500 matches the brightness of the standard white document at the time of prescanning when the background level image is read.

Therefore, the base can be removed.

In addition, in the case of mixed originals, that is, when the light intensity is increased,
The lookup table of the logarithmic conversion unit 508 is changed so that the hue of the output image is corrected in accordance with the increase in gain. Therefore, it is possible to compensate for a change in the hue of an output image due to an increase in gain, and output an image with an appropriate hue.

[Other Embodiments] Next, other embodiments according to the present invention will be described below with reference to the related drawings.

As shown in Figure 1, the image data read by the CCD 500 is digitally converted, then subjected to misalignment correction, black correction, and white correction, color conversion, logarithmic conversion, one-color correction, and black character processing. In this embodiment, as a countermeasure against the phenomenon in which the density of the entire image area becomes low due to the increase in analog gain accompanying background color removal,
A corresponding configuration was obtained by switching the LUT 511 for density conversion (step 816 shown in FIG. 16). As mentioned above, the contents of the LUT 511 for density changes store data for correcting one density gradation, but this content is changed for each color, that is, the contents of the LUT for B signal, G, and R are switched respectively. By doing so, the density of each color signal can be changed.

For example, when the gain is increased to remove the background, the image signals of E, G, and R also increase accordingly. Similarly, after logarithmic transformation, the density close to the base density changes slightly. Here, by changing the contents of the LUT 511 according to the values corresponding to the increased signal data, it is possible to suppress changes in image density.

[Effects of the Invention] As described above, according to the present invention, a read color image can be processed in accordance with the background of the color image.

[Brief explanation of the drawing]

Figure 1 is a schematic block diagram of an image processing device to which the present invention is applied. Figures 2 (a) and (b) are diagrams showing a color reading sensor and drive pulses. Figure 3 (a) and (b) are diagrams showing drive pulses. Figure 4 (a) and (b) are diagrams showing the circuit block for generating pulses and their output, Figures 4 (a) and (b) are diagrams showing the circuit block for black correction and its operation, and Figures 5 (a) to (d) are diagrams showing white correction. 6(a) and 6(b) are diagrams showing the circuit blocks for logarithmic conversion and their operations. FIG. 7 is a diagram showing the characteristics of the color separation filter. Figure 9 shows the characteristics of color toner, Figures 9(a) and 9(b) show the circuit blocks for color correction and their operations, Figure 10 shows the process of generating document reflection luminance data, and Figure 11 The figure shows a circuit block for black character processing, Figures 12(a) to (d) are diagrams showing the operation of black character processing, Figure 13 is a flowchart showing the algorithm for background removal in this embodiment, and Figure 14. (a) and (b) are diagrams showing the density conversion circuit 511 and its characteristics, FIG. 15 is a diagram showing the S/H & A/D conversion circuit 502, and FIG. 16 is an algorithm for background removal in another embodiment. It is a flowchart which shows. In the figure, 22... CPU, 24... RAM, 500...
...CCD, 520...Halogen lamp, 507...
・Color correction circuit (1), 508...logarithmic conversion section, 509
...color correction circuit (2). Agent Patent Attorney Yasunori Otsuka (and 1 other person) Figure (b) Figure 41 (6) Figure (c) Figure (d) Figure (CI) Figure (b) Figure (b) Figure Figure LU Ding (Rusiqua, Putebuno L) A LUT (Rusiqua Noah Table) 8 LUT () L Sogua Ibu 1-Full) C Figure (CI) Figure (b) Figure (C) Figure (d) ) Figure Figure ((1) Figure (b)

Claims (1)

[Claims] (1) A color image processing device that calculates color image background information based on a read color image and performs color image processing according to the color image background information, which calculates analog gain according to the color image background information. A color image processing apparatus comprising: a control means for controlling; and a setting means for setting an analog gain controlled by the control means.