JPH0224074B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a method for processing image signals such as high-definition television signals, and in particular to a method for displaying high-definition images by improving image quality without increasing the required frequency bandwidth. This is what I did.

In conventional image signal processing methods of this type, so-called line interlace scanning is used as image signal processing to reduce the required transmission frequency bandwidth without degrading the image quality of the image reproduced and displayed on the receiving side of the image signal transmission system. It was common to use However, in images displayed using normal line interlaced scanning, the coarse scanning line structure in each field is easily noticeable and is a major cause of image quality deterioration. There were various drawbacks, such as deterioration of image quality due to loss of image quality. In other words, an image using a line interlaced scanning method with a structure of alternating scanning lines for each field requires 70% of the vertical resolution expected from the number of scanning lines.
Although it has the inherent drawback that only a certain degree of vertical resolution can be obtained, it has the characteristic that flicker can be reduced without increasing the number of images per second and therefore the required frequency bandwidth. is carried out exclusively by the line interlace scanning method, so even if a high-definition television image is constructed using the line sequential scanning method, which doubles the number of images per second, it will be adapted to the line interlace scanning method during playback and display. In addition, in order to avoid an increase in the transmission frequency bandwidth, it was necessary to intermittently sample the line-sequential scanning image signal with twice the number of images per second, perform time axis expansion, and convert the image scanning method.

However, when sampling is applied to an image signal for time axis conversion as described above, in addition to the frequency components associated with image scanning, the sampling frequency and its sideband frequency components are newly generated, and the original image signal is The essential problem of image quality deterioration that accompanies sampling is that the reproduced image quality of high-definition image signals is significantly degraded due to mixing in the distribution of components, especially the distribution of high frequencies. This is a particularly serious problem for high-quality image signal processing systems.

Conventionally, to remove unnecessary signal components that mix into image signals through such sampling, a wave generator with a simple configuration that simply handles the arrangement of each signal component on the frequency axis has been used, but it is difficult to remove high-quality images. In order to reproduce and display, it is necessary to arrange each signal component not only on both the horizontal and vertical frequency axes, but also on the time axis, that is, to arrange two signals corresponding to the image signal spectrum.
It is necessary to perform image signal processing in which unnecessary signal components mixed in are removed based on sampling of image signals from all angles using a wave generator configured in multiple dimensions, from 3D to 3D.

In addition, when converting the scanning method mentioned above,
Since scanning line order conversion and time axis conversion are performed exclusively in the vertical direction, except for cases where scanning method conversion between sequential scanning and dot interlacing is performed, two-dimensional scanning is mainly performed on the vertical frequency axis and time axis. It is common to use a wave device configured as follows.

SUMMARY OF THE INVENTION An object of the present invention is to provide an image signal processing method that does not cause image quality deterioration inherent in interlaced scanning when reproduced and displayed using interlaced scanning.

Another object of the present invention is to provide an image signal processing method using a wave generator that removes high-frequency unnecessary signal components corresponding to an image signal spectrum on a multidimensional coordinate plane consisting of at least a vertical frequency axis and a time axis. It is about providing.

Still another object of the present invention is to provide an image signal processing method that further improves image quality by emphasizing high-frequency signal components in the passband of a multidimensional waveform device used.

Still another object of the present invention is to remove unnecessary signal components caused by sampling of an image signal by a multi-dimensional wave generator on at least one of a transmitting side and a receiving side of an image signal transmission system. The objective is to provide a signal processing method.

That is, in the image signal processing method of the present invention, when converting between sequential scanning and interlaced scanning performed by sampling an image signal, the horizontal frequency axis expressed by the sampling frequency and the horizontal scanning frequency, the vertical frequency axis expressed by the vertical scanning frequency, and the per second At least one sampling frequency point generated based on sampling of the image signal in a multidimensional coordinate consisting of a coordinate plane formed by at least a vertical frequency axis and a time axis among the time axis expressed by the number of images, and the origin of the multidimensional coordinate. Sampling the image signal using at least one multi-dimensional low-pass filter whose desired passband is a region corresponding to the origin side region of the multi-dimensional coordinates of a line connecting points that are approximately equidistant from each other. In order to suppress high-frequency unnecessary signal components on the coordinate plane formed by at least the vertical frequency axis and the time axis, which are generated by The interlace ratio to be converted is determined between each intersection of the vertical frequency axis and the time axis and the origin of the multidimensional coordinates with the line connecting the points that are approximately equidistant from each other. a vertical frequency-axis low-pass filter and a time-axis low-pass filter each having upper limits of their passbands set at successive frequency points divided into equal parts, respectively; The input and low-pass filter outputs are each guided to a subtracter, and the subtracted outputs obtained by subtracting each other are used as the high-pass filter output. The low-pass filtered output of the time-axis low-pass filter is guided to the adder via a filter, and the low-pass filtered output of the time-axis low-pass filter is directly or indirectly guided to the adder, and at least the summed outputs that are mutually added are subjected to two-dimensional low-pass filtering. The present invention is characterized in that a two-dimensional low-pass filter for output is used as at least a part of the multi-dimensional low-pass filter.

The invention will be explained in detail below by way of example embodiments with reference to the drawings.

First, an example of the basic configuration of an image signal processing apparatus according to the present invention is shown in FIG. Although the basic configuration shown in the figure will be described in detail later, what is in line with the features of the present invention described above are the multi-dimensional prefilter 2 and interpolation filter 7b inserted on the transmitting side and the receiving side, respectively. Among these, the multi-dimensional pre-filter 2 processes the image signals intermittently extracted by the sub-sampling circuit 3, and before the time-axis conversion circuit 4 expands the time axis and reduces the signal frequency band, the multi-dimensional pre-filter 2 detects newly generated image signals by sampling This is to prevent the sampling frequency component and its sideband component from becoming a so-called folded signal component with the sampling frequency point as a virtual origin and being mixed into the original image signal spectrum.

That is, for example, the number of scanning lines is 525, which can be obtained from a line-sequential scanning high-definition television camera 1.
A wideband image signal a of 60 frames per second is generally
On the vertical frequency/time coordinate plane in three-dimensional coordinates as shown in Figure 2, which consists of horizontal and vertical frequency axes and time axes, we can calculate the spread of the image in the vertical direction based on the origin O and its spread according to the movement. A point ( _vp , O) determined by the signal spectrum distribution of the subject itself within the area surrounded by the dotted line and the vertical scanning frequency _vp on the vertical frequency axis _v when the subject having this signal spectrum is captured by camera 1. , consists of a signal spectrum generated in a region having the same spectral spread as described above centered on the points (O, _F ) and ( _vp , _F ) determined by the number of images per second _F on the time axis _t , respectively. It exhibits a spatial spectrum. In FIG. 2, the frequency point _VP on the vertical frequency axis and the frequency point _F on the time axis are both normalized and shown equidistant from the origin.

As shown in FIG. 1, the wideband image signal a exhibiting such a spatial spectrum is directly guided to the sub-sampling circuit 3 and the time axis conversion circuit 4 without passing through the multi-dimensional pre-filter 2 according to the present invention, and is, for example, scanned. When converting to a normal narrowband image signal c using an interlace method of 525 lines, 60 fields, and 30 frames, the sub-sampling circuit 3 converts only odd and even lines from the wideband image signal a every frame. Assuming that the time axis of the intermittent image signal b extracted from the sub-sample circuit 3 is converted to double the time axis to form a normal narrowband image signal, the spatial spectrum of the intermittent image signal b extracted from the sub-sampling circuit 3 is as shown in Fig. 2. In addition to the signal spectrum distribution in the above region, a new vertical scanning frequency based on intermittent extraction with half the number of scanning lines.
The area shown surrounded by a solid line in Figure 2 has the same extent as above and is centered on a new coordinate point ( _{vp/2 , F/2} ) determined by vp/2 and the number of images per second _F/ 2. At least new signal frequency spectral components are generated, and as shown in FIG. 2, overlap occurs between each of them and the original signal spectral distribution region surrounded by the dotted line. Among them, it occurred between the original signal spectrum distribution region centered on the coordinate origin O.
The overlap is mixed into the original signal spectrum distribution as a so-called sampling aliasing component with the point ( _vp/2 , _F/2 ) as the virtual origin _OS , causing image quality deterioration due to aliasing distortion in the image signal.

In the conversion process of this type of image signal, the present invention provides the following features:
By removing the above-mentioned aliasing distortion components that occur based on signal sampling, at least those that occur on a two-dimensional coordinate plane regarding the vertical frequency axis and the time axis, it is possible to reproduce an image of sufficiently high quality. For example, the sampling aliasing component mentioned above with reference to FIG. 2 spreads around the virtual origin O _S and invades the original signal spectrum distribution region.
In order to avoid overlap between the original signal spectral distribution region centered at the origin O and the folded component region centered at the virtual origin O _S , the origin O and the virtual origin O _S
, i.e., the point on the vertical frequency axis ( _vp/2 , O) and the point on the time axis (O,
After restricting the original signal spectrum distribution region to only the region closer to the origin O than the straight line connecting _F/2 ), intermittent extraction and time axis conversion with half the number of scanning lines as described above are performed. This is applied to the wideband image signal a, and the original signal spectrum distribution area is limited using a two-dimensional or three-dimensional low-pass (pass) wave device.

Therefore, according to the present invention, although the original signal spectral distribution area of the broadband image signal itself is somewhat reduced, the signal components in the reduced area are mixed with sampling aliasing distortion components. Compared to the reproduced image quality when used in its original state, the image quality obtained when sampling aliasing distortion components are removed is much better, although some high-frequency signal components are lost. It is possible to play back and display high-quality images from the period.

That is, in the image signal processing apparatus according to the present invention shown in FIG. After removing the signal components in the shaded area outside the straight line passing through the above-mentioned points ( _vp/2 , O) and (O, _F/2 ) from the original signal spectral distribution area, A normal narrow band image signal c is formed by performing the above-described intermittent extraction of scanning lines and time axis expansion by the circuit 3 and the time axis conversion circuit 4, and is transmitted through the modulation device 5 with the required transmission frequency band reduced. On the transmitting and receiving side,
The narrowband image signal extracted by the demodulator 6 is guided to the time axis conversion circuit 7a to form a line-sequential scanning pseudo-wideband image signal in which the time axis is restored to the original state of the wideband image signal a, which is then displayed on a display. The device 8 reproduces and displays high-quality images. However, since the original signal spectral distribution area of the narrowband image signal extracted at the receiving side is limited by the transmitting side pre-filter 2, the above-mentioned aliasing distortion component does not occur. , by intermittent extraction of scanning lines in the sub-sample circuit 3,
In the figure, unnecessary signal components based on sampling remain on the outer periphery of the straight line connecting points (F _vp/2 , O) and (O, _F/2 ), touching the limited original signal spectrum distribution area. In addition, since unnecessary signal components are newly generated in the time axis conversion circuit 7a,
These unnecessary signal components are also displayed as noise on the display device 8, which has broadband characteristics for displaying high-quality images, deteriorating the display image quality. Therefore, in the image signal processing apparatus according to the present invention shown in FIG. The image signal is supplied to the display device 8 to prevent noise components due to sampling or the like from being displayed.

Next, the above-mentioned image signal processing process carried out using the multi-dimensional low frequency filter according to the method of the present invention will be explained with reference to FIGS. 3 and 4, which are shown on the usual frequency axis. For example, if the line-sequential scanning broadband image signal a with 525 scanning lines and 60 frames has the signal waveform shown in waveform a in Fig. 3, in the case of a still image with no image movement, , the signal frequency spectrum is, as shown in waveform a in Figure 4, when the horizontal scanning frequency is _2H , horizontal frequency axis spectrum components are arranged at frequency positions that are integral multiples of 2H, and the horizontal frequency axis spectrum components are arranged at frequency positions that are integral multiples of 2H, and the integer frame frequency _v before and after that. The vertical frequency axis spectrum components are arranged at double frequency positions, and when a part of the signal frequency spectrum is enlarged, it becomes waveform b in FIG. 4.
That is, the horizontal scanning frequency of the wideband image signal a in the illustrated example is _2H , which is twice the horizontal scanning frequency _H in a normal television system, and any horizontal frequency axis spectral component _2nH (n=1, 2,
), vertical frequency axis spectral components as sideband components are arranged at frequency intervals of frame frequency _v = 60 Hz. A wideband image signal a exhibiting such a frequency spectrum is converted into a two-dimensional to three-dimensional
When the signal is supplied to the prefilter ₂ consisting of a low-pass filter of
0), (O, _F/2 ), so as shown in waveform c in Figure 4, the horizontal frequency axis component 2n in the frequency spectrum shown in waveform b is In the arrangement of vertical frequency axis components centered on _H , a part of the high frequency band is removed, and each exhibits a comb-shaped passband with the center frequency as the center frequency. The broadband image signal a that has passed through the pre-filter 2 having such a pass characteristic is guided to the sub-sampling circuit 3, and the image signal is output every other horizontal scanning period of the sequential scan, as shown in waveform b in FIG. When extracted and subjected to signal thinning, that is, so-called subsampling, the entire band remains almost unchanged, but the horizontal fundamental frequency is reduced to 1/2. Next, in the time axis conversion circuit 4, as shown in waveform c in Fig. 3, when the interaxis axis of the extracted signal component is expanded by a factor of two, the signal band is halved and the total band is also halved. This corresponds to an interlaced narrowband image signal with 525 scanning lines, 60 fields, and 30 frames, which corresponds to the standard television system. In other words, the narrowband image signal spectrum obtained by halving the frequency of the wideband image signal spectrum shown in waveform _b in FIG. As shown in waveform d in the figure, horizontal frequency axis components are arranged at frequency intervals of horizontal scanning frequency _H , and vertical frequency axis components are arranged at frequency intervals of field frequency 60Hz relative to both sides of those horizontal frequency axis components. It extends in the opposite direction. Therefore, if the broadband image signal spectrum of waveform b is sampled and time-axis converted as it is, as shown by the thin line in waveform d, the horizontal frequency axis components 2n _H and (2n+1) _H The vertical frequency axis components extending as shown in FIG. Noise due to a distortion component corresponding to the sampling aliasing distortion component occurs in the image signal, resulting in deterioration of image quality. However, in the pre-filter 2 having the passband characteristic shown in waveform c, the spectral components in the center portion between adjacent horizontal frequency axis components in the wideband image signal frequency spectrum shown in waveform b are preliminarily filtered. If this is removed, as shown by the thick line in waveform d, there will be no contamination due to the intersection of the vertical frequency axis components, and therefore a high-quality image signal with the sampling aliasing distortion component described above equivalently removed. This corresponds to limiting the vertical spatial frequency component to below _vp/2 in FIG. 2.

In the above explanation, the wideband image signal a is assumed to be a still image signal, but if the wideband image signal a represents a moving image, the image is Because the position shifts and the time interval required to reproduce the same pixel differs from the vertical scanning period length, etc., the frequency of the vertical frequency axis component at 60 Hz intervals in the signal frequency spectrum shown in waveform b in Figure 4. When a spread occurs in the position and the image moves rapidly, the spread increases, and cross-mixing similar to the above-mentioned cross-mixing of vertical frequency axis components between adjacent horizontal frequency axis components occurs between adjacent vertical frequencies that have a mutual spread. This also occurs between the axis components, and the movement of the reproduced and displayed image loses its naturalness, resulting in deterioration of the moving image quality.

However, if the wideband video signal a having a frequency spectrum exhibiting the spread of vertical frequency axis components is supplied to the pre-filter 2 which has a comb-shaped passband characteristic with a frequency interval of 60 Hz as shown in waveform c as described above. , the high-frequency components that cross-mix with each other in the spread of the vertical frequency axis components are removed, so that when image signal processing is performed using the multi-dimensional low-frequency filter according to the present invention, even for wideband video signals. , it is possible to effectively prevent the conventional image quality deterioration, and corresponds to limiting the spatial frequency component in the time axis direction to _F/2 or less in FIG. 2. Therefore, the effect of a waveform exhibiting a comb-shaped passband characteristic as shown in waveform c in FIG. 4 of the frame frequency interval on the frequency axis is the original signal spectrum of the image signal described above with reference to FIG. Points in the distribution area ( _vp/2 , O), (O, _F/2 )
This corresponds to the effect of the restriction on the straight line passing through .

In the above explanation, for example, a line-sequential scanning wideband image signal with 525 scanning lines and 60 frames per second is used as a wideband image signal with 525 scanning lines and 60 fields per second.
When converting to an interlaced narrowband image signal of a frame, the image signal of a horizontal scanning period in a line sequential scanning image signal is extracted only once every two times and converted to a 2:1 interlaced image signal. The aspect of image signal processing according to the method of the present invention has been described, but when similarly converting to a high-order interlaced image signal such as 3:1 or 5:1, it is necessary to Image signals in the horizontal scanning period are extracted only once every three times or once every five times, and subjected to image processing similar to that described above. In addition, when converting to a dot interlaced image signal, sampling is performed on the line-sequential scanning broadband image signal at a much higher sampling frequency, for example, several hundred times the horizontal scanning frequency, and the sample image signal is converted into a dot-interlaced image signal. , as described above with reference to FIG. 3, the image signal is intermittently extracted once every two samples or once every three samples, and the image signal processing similar to that described above is applied to the intermittently extracted image signal. . Note that, as described above, the multi-dimensional low-frequency wave device used in such a case is a three-dimensional low-frequency wave device based on three-dimensional coordinate axes including the horizontal frequency axis.

Next, FIG. 5 shows an example of a specific configuration of the time axis conversion circuit 4 in the basic configuration shown in FIG. 1 of the image signal processing apparatus according to the present invention. In the time axis conversion circuit with the illustrated configuration, the input terminal 9
As shown in the waveform b of FIG. 3, the intermittently extracted broadband image signal b is supplied to the
1 is supplied with a sampling clock signal for digitizing the image signal. The intermittently extracted image signal b from the input terminal 9 is guided to the low-pass wave generator 10 to remove unnecessary signal components outside the required band, and then sent to the input terminal 21.
The signal is guided to an analog-to-digital converter 11 driven by a clock signal from the clock signal, and converted into an intermittent digital image signal. It is alternately supplied to lines 13 and 14 and driven by a clock signal applied from input terminal 21 via changeover switch 19 to write alternately. A low-speed clock signal of 1/2 clock frequency, which is formed by guiding the clock signal from the input terminal 21 to the 1/2 frequency divider 22, is also applied to these 1H delay lines 13 and 14 via the changeover switch 20. The image signals for one horizontal scanning period, written alternately as described above, are read out alternately with the time axis doubled, and transmitted through the changeover switch 15 as shown in the waveform c in FIG. The digital-to-analog converter 16 converts the state into a narrowband image signal c.
The converted output analog narrowband image signal is extracted from an output terminal 18 as a time-base converted output after unnecessary signal components outside the required band are removed by a low-pass wave filter 17. The changeover switches 12, 15, 19, and 20 are switched in conjunction with each other, and when an image signal is written at high speed to one of the 1H delay lines 13 and 14 in accordance with the input clock signal, the 1H delay line is switched. It is assumed that the image signal is read out at a low speed from the other of the lines 13 and 14 in response to a low speed clock signal.

In the configuration shown in FIG. 5, if the wideband image signal supplied to the input terminal 9 has already been digitized, the analog-to-digital converter 11 in the configuration shown can be omitted, and the subsequent Of course, if the signal processing is performed in digital form, the digital-to-analog converter 16 in the illustrated configuration can be omitted.

The waveform c in FIG. 3 obtained as described above from the time base conversion circuit 4 in the basic configuration shown in FIG.
The converted output image signal c shown in FIG. It is significantly different from conventional interlaced image signals in that a three-dimensional wave is applied to prevent the generation of so-called sampling aliasing distortion components that are mixed into the original signal spectrum distribution during signal conversion. are doing.

Next, a description will be given of the wave characteristics of the three-dimensional low-pass (pass) wave device that constitutes the singularity of the image signal processing method of the present invention, and the specific configuration for realizing the wave characteristics.

First, in order to define three-dimensional wave characteristics, an image such as a television is spatially divided into two parts in both the horizontal and vertical directions by horizontal and vertical screen scanning, as shown in Figure 6. It is analyzed by approximating that it is sampled dimensionally and also sampled temporally. Therefore, the sampling frequency in spatial two-dimensional sampling corresponds to the horizontal and vertical scanning frequencies in screen scanning. For example, in the line sequential scanning method,
The vertical sampling frequency _vp is the number of scanning lines
This corresponds to the frame frequency that is the reciprocal of L _o ,
_vp = 1/L _o , and the horizontal sampling frequency _up is based on the sampling of the image signal in the horizontal direction when converting an analog image signal to a digital image signal. The value obtained by dividing the frequency by the horizontal scanning frequency _h is multiplied by the aspect ratio of the screen, which is 3:4 in standard television, that is, 3/4, and is expressed in the form of a spatial frequency. For example, ,
Sampling frequency during analog-to-digital conversion
If _S , then _up = _S / _H × 3/4. Furthermore, in the line sequential scanning method, the sampling frequency _tp in the time axis direction is the number of frames per second F _N because temporal changes due to image movement appear on a frame-by-frame basis. When the sampling frequency of each axis in the three-dimensional coordinate axes for defining three-dimensional wave characteristics is set in this way, the spatial frequency/time domain represented by the three-dimensional coordinates of the image signal of the line sequential scanning method is The signal spectral distribution exhibited by the image signal can be expressed using a discrete Fourier transform, or so-called DFT, and also for interlaced image signals, the interlacing itself is a type of sample different from that in the line-sequential scanning method. , it can be similarly expressed using the DFT method,
At that time, the sampling frequency in each direction is set to 1/1/2 for the 2:1 interlace method, for example.
2, the image signal spectral distribution in the three-dimensional coordinate area described above can be obtained in the same way for any scanning method, and
It is possible to estimate what kind of aliasing distortion components occur in relation to such a sampling frequency, and furthermore, based on the estimation, it is also possible to clarify the image signal spectral distribution characteristics necessary to prevent the mixing of aliasing distortion components. It turns out.

_As shown in FIG _. , _tp and their respective harmonic frequency points 2 _up , 2 _vp , and 2 _tp are set at the positions of the black dots shown in the figure, respectively.

However, as described above with reference to FIG. 2, in general, image signals are not sampled in the horizontal spatial frequency axis direction, and between the line-sequential scanning method and the interlaced method, sampling is not performed in the horizontal spatial frequency axis direction. Since there is no change in Spectral distribution can be considered.

As shown in FIG. 8, each sampling frequency of the interlaced image signal expressed in two-dimensional coordinates is a frequency point 1/2 _vp , 1/2 on each axis.
_tp and each of their harmonic frequency points _vp and _tp are set at the positions of the black points shown in the figure, which are determined by the combination of the points.

As is clear from a comparison between Fig. 7 and Fig. 8, in the interlaced image signal, the frequency points 1/2 _vp and 1/2 _tp on each axis are lower than in the line sequential scanning method. A new sampling frequency has been generated, and therefore, when converting a line-sequential scanning wideband image signal into an interlaced narrowband image signal by performing signal processing including sampling on the line-sequential scanning wideband image signal, as described above with reference to FIG. , as mentioned above, the new sampling frequency point (1/2 _vp , 1/2 _tp ) is set as the virtual origin O _S , and the boundary line that is equidistant from the coordinate origin O, that is, on each axis If the signal spectrum distribution of the line-sequential scanning wideband image signal is restricted to the region on the origin O side from the straight line passing through the new sampling frequency points 1/2 _vp and 1/2 _tp , it becomes a 2:1 interlaced narrowband image signal. When converted, image quality deterioration due to mixing of sampling aliased components into the original signal spectrum distribution region can be prevented from occurring, and such signal spectrum distribution limited regions are shown with diagonal lines in FIG. 8.

Therefore, FIG. 8 shows the signal spectral distribution restriction area when converting a line-sequential scanning wideband image signal to a 2:1 interlaced narrowband image signal, and in the case of other interlace ratios described later. The virtual origin O _S is shown as O _S2 to distinguish it from the above. In addition, FIG. 8 shows only the first quadrant of the coordinate plane regarding the vertical frequency axis and the time axis, and when the other quadrants are combined, the signal spectrum distribution restricted area in this case is indicated by the diagonal lines shown in the figure. It becomes a square made up of four triangles with . Therefore, the distribution area of the folded component spectrum centered on the virtual origin O _S2 is also approximately the same as the first one.
The illustrated visual diamond area can be schematically represented.

Therefore, there are two directions on the vertical frequency axis and the time axis in which the pass band is the signal spectrum distribution limited region.
In order to realize a two-dimensional low-pass filter, a two-dimensional low-pass filter whose pass band is equivalently shown in the shaded area in FIG. An example is shown in Figure 9b. The two-dimensional wave device with the illustrated configuration is a time axis direction low frequency wave device 2.
4 and a vertical frequency axis direction low-pass filter 25 are combined in adders ADD1 and ADD2, and the time-axis direction low-pass filter 24 has an upper limit of point α on the time axis _t in FIG. 9a. The vertical frequency axial low-pass filter 25, which has a passband as the shaded area with diagonal lines, and subtracts and supplies the passband component from the input image signal supplied to the input terminal 23, is configured as described above in FIG. 9a. The hatched area with the lower limit of point α as the passband is defined as the passband, and
If the wave output components are added and taken out from the output terminal 26, a two-dimensional wave output image signal whose pass band is the entire shaded region in FIG. 9B is obtained.

In addition, in FIG. 9a, the pass band shown in FIG. 8 is approximated in a relatively simple form, but in order to approximate this more accurately and precisely, it is necessary to pass through the diagonally shaded area shown in FIG. 10 a. A two-dimensional wave device for the band can be constructed as shown in FIG. 10b. In the two-dimensional wave device having _the configuration shown in the figure, in the same manner as in the two-dimensional wave device having the configuration shown in _FIG . Time-axis direction low-pass filters 28 and 29 whose passbands are the vertical frequency axis direction low-pass filters, and vertical frequency-axis direction low-pass filters whose respective passbands are hatched areas whose upper limits are points β ₁ and β ₂ on the vertical frequency axis. are combined by adders ADD3 to ADD5.

Next, FIG. 11 shows an example of a specific configuration of a low-frequency wave device in the time axis direction or vertical frequency axis direction used in the two-dimensional wave device configuration shown in FIG. 9b or FIG. 10b. The low frequency amplifier with the illustrated configuration includes a plurality of delay elements 34 connected in series in series,
It forms a so-called transversal filter consisting of a plurality of weighting coefficient units 35 connected to each input and output terminal of these delay elements 34 and an adder ADD6 that adds the output signals of these coefficient units 35, and each delay element 34 is one frame memory in the time-axis direction low-pass filter, and one line memory in the vertical frequency-axis direction low-pass filter.

Next, the line sequential scanning wideband image signal is
In the case of converting to a 2:1 interlaced narrowband image signal, in the same manner as shown in Fig. 8,
FIGS. 12A and 12B show how the signal spectrum distribution region of the two-dimensional wave device is restricted when converting to a 3:1 interlaced narrowband image signal. As is clear from the figure, the sampling frequency points newly generated when converting to a 3:1 interlaced image signal are frequency points 1/3 _vp , 2/3 _vp on each axis determined by the interlace ratio of 3:1. and 1/3 _tp and 2/3 _tp appear at the positions of black points O _S3 and O' _S3 shown in the figure. Therefore, for the signal spectral distribution limited region when converting to a 3:1 interlaced image signal, the above-mentioned sampling frequency point is
Setting O _S3 and O' _S3 as virtual origins, we can find boundary lines that are equidistant from the virtual origins O _S3 , O' _S3 and the coordinate origin O, as shown by diagonal lines in Figure 12. is the first quadrant of a regular octagon centered on the coordinate origin O, and the distribution area of the sampling aliasing distortion component spectrum centered on each sampling frequency point O _S3 and O' _S3 is a regular octagon with almost the same shape. Become.

Next, the line sequential scanning wideband image signal is 4:
The manner of limiting the signal spectral distribution area of the two-dimensional wave device when converting to a 3:1 interlaced image signal is similar to that shown in FIG. 12 when converting to the 3:1 interlaced image signal described above. It is shown in FIG. As is clear from the figure, the newly generated sampling frequency points when converting to a 4:1 interlaced image signal are frequency points 1/4 _vp , 1 on each axis determined by the interlace ratio of 4:1. /2 _vp ,3
/4 _vp and 1/4 _tp , 1/2 _tp , 3/4 _tp appear at the positions of black points O _S4 , O' _S4 , O'' _S4 shown in the figure. Therefore, the 4:1 interlaced image For the signal spectrum distribution limited region when converting to a signal, the above sampling frequency points O _S4 , O' _S4 ,
With O'' _S4 as the virtual origin, if we find the boundary lines that are equidistant from the virtual origins O _S4 , O' _S4 , O'' _S4 and the coordinate origin O, we can find them as shown with diagonal lines in Figure 13. is the first quadrant of an irregular dodecagon centered on the coordinate origin O, and the distribution area of the sampling aliasing distortion component spectrum centered on each sampling frequency point O _S4 , O′ _S4 , O″ _S4 is as follows. The result is an irregular dodecagon with substantially the same shape.Therefore, the desired two-dimensional low-frequency filter can be realized in the same manner as described above with reference to FIGS. 9 to 11.

As is clear from the above deduction, even when the interlace ratio is further increased, the new sampling frequency point close to the desired band will be the frequency on the vertical frequency axis, in the same way as described above. It occurs on a straight line passing through the point 1/2 _vp and the frequency point 1/2 _tp on the time axis, so in the same manner as in FIGS. By configuring the two-dimensional low-pass filter whose passband is the distribution restricted area as described above, even when converting to interlaced image signals having various interlace ratios, there is no problem due to the mixing of sampling aliasing distortion components. Image quality deterioration can be prevented.

The signal spectral distribution restriction area for preventing sampling aliasing components according to the present invention as described above is applicable not only to the above-mentioned line interlacing, but also to converting a line sequential scanning wideband image signal to a dot interlacing image signal. can also be applied in the same way, and an example of the aspect of limiting the signal spectral distribution region in such a case is shown in the first example.
This is shown in FIGS. 4 and 15. That is, as in the case of converting to the line interlaced image described above, the sequential pixels of the image constructed by line sequential scanning are divided into every other group 〇 and × as shown in Fig. 14. The newly generated sampling frequency points when performing 2:1 dot interlacing are also sampled in the horizontal frequency axis direction, so as shown in Figure 15, the sampling frequency points newly generated are The frequency points 1/2 _up , 1/2 vp, and 1/2 on each axis of the horizontal frequency axis, vertical frequency axis, and _time axis are determined by
In relation to _tp , the black points h shown in the diagram, that is, the above-mentioned frequency points 1/2 _up , 1/2 _vp , _{and tp} are respectively converted to c, a, and e.
The signal spectral distribution restricted region that appears at the diagonal vertex h opposite the coordinate origin O in the cubic abcdefho and is based on the same idea as when converting to a line interlaced image signal is:
This is an area surrounded by line segments connecting each of the vertices a, b, c, d, e, f, and o of the cube and the intersection point g of the diagonal lines. However, in order to actually construct a three-dimensional low-pass waveform whose required pass band is a signal spectrum distribution restricted region having such a complicated shape, it is necessary to convert such a complicated shape into a relatively complex shape as shown in FIG. 16a. By approximating with a simple shape, substantially the same wave effect can be obtained. Therefore, in order to realize a three-dimensional low-pass wave filter having a passband shape shown in FIG. 16a, a two-dimensional low-pass waveform generator having a passband shape as shown in FIG. In the same way as the pass band of 16 is divided into a band in the time axis direction and a band in the vertical frequency axis direction and constructed as shown in Figure 16b, the low frequency band in the time axis direction is It can be easily constructed by combining the frequency range filter 38, the vertical frequency axial low-pass filter 39, and the horizontal frequency axial low-pass filter 40 in the adders ADD7, ADD8, and ADD9.

Note that in order to improve the accuracy of approximation of the passband of the three-dimensional low-pass filter having the configuration shown in FIG. 16b to the required passband shown in FIG.
As shown in FIG. 3B, the passbands of the three-dimensional low-pass filters 43 and 44, each having roughly approximated passbands, are slightly different from each other as shown in FIG. 43 and 44 are combined by adders ADD10 and ADD11,
The composite passband can be made to closely approximate the required passband shown in FIG.

Note that the configuration of the three-dimensional low-pass filter described above for converting to a 2:1 dot trace image signal is similar to that described above for the line interlace method, and converts the dot interlace image signal with an increased interlace ratio. The same can be applied when converting to .

Therefore, in the image signal processing apparatus according to the present invention shown in FIG. The narrowband image signal c converted from the line-sequential scanning wideband image signal a by the axis conversion circuit 4 as described above with reference to FIG. On the side, the signal is demodulated to its original form by a demodulator 6. Note that the transmission modulation/demodulation devices 5 and 6 can be of either an analog type or a digital type, and conventionally used devices can be used as they are.

On the receiving side, the narrowband image signal c restored as described above is guided to the time axis conversion circuit 7a, where it undergoes time axis conversion that is opposite to the time axis conversion performed by the time axis conversion circuit 4 on the transmission side. As shown in FIG. 3b, the intermittent image signal b is reconverted by high-speed scanning, and the transmission side time axis conversion circuit 4 has a circuit configuration similar to that shown in FIG. is applied to convert it into a line-sequential scanning pseudo-wideband image signal. Although this pseudo-wideband image signal has been subjected to preventive measures against deterioration in image quality due to the mixing of sampling aliasing distortion components by the pre-filter 2 on the transmitting side,
After the preventive measures are taken, sampling is actually performed by the sub-sampling circuit 3 to generate a new sampling frequency component, and further, when subjected to time-base conversion by the time-base conversion circuit 7a on the receiving side. Since there is a risk that unnecessary signal components adjacent to the signal spectrum distribution restricted area may be mixed in,
The reproduced pseudo-wideband image signal is supplied to the three-dimensional interpolation filter 7b, which applies a three-dimensional low frequency wave having exactly the same pass band as the three-dimensional prefilter 2 on the transmitting side. A high-quality line-sequential scanning broadband image signal from which unnecessary signal components have been removed is supplied to the display device 8 so that a high-quality image can be displayed.

Note that in order to simplify the configuration and reduce the cost of the receiving side device, the configuration of the three-dimensional interpolation filter 7b can be simplified, but even in such a case, the three-dimensional spatial frequency generated by sampling the image signal Low sampling frequency components in the region, i.e., for example, the folded component region centered on the sampling frequency point O _S shown in FIG. 8, or the sampling frequency point O _S3 , shown in FIG. 12,
By sufficiently suppressing the sampling aliasing components distributed in the aliasing component regions centered on O′ _S3 ,
It is necessary to make it difficult to visually detect it in the displayed image.

Furthermore, the order of the time base conversion circuit 7a and the interpolation filter 7b on the receiving side of the image signal processing apparatus according to the present invention shown in FIG. It is also possible to apply a three-dimensional low frequency wave to remove unnecessary sampling frequency components associated with the signal, and then perform time axis transformation. An example of the configuration of the main part in such a case is shown in FIG. 17 in the case of converting a 2:1 line interlaced image signal to a line sequential scanning wideband image signal. In the configuration shown, FIG.
A low-speed scanning interlaced narrowband image signal 46 as shown in FIG.
The 1 field memories 47, 48 and the 1/2 line memory 49 produce a 1 frame delayed image signal 53 and a 1 field + 1/2 line delayed image signal 54.
are similarly guided to the three-dimensional interpolation filter 50, the signal spectral distribution area is restricted as described above, and the 1 field + 1/2 line delayed image signal 54 is used as a reference.
As shown in FIG. 18, the subsequent time axis conversion circuit 5
An interpolation signal for supplementing signal interpolation to the intermittent image signal as shown in FIG.

That is, as shown in FIG. 18, each line image signal with scanning line numbers i ₁ +1, i ₁ +3, i _{1 +5, ... in the 1} field + 1/2 line delayed image signal 54 is converted into a signal after time axis conversion. The scanning line numbers i ₁ +1, i ₁ +3, i ₁ +5, . ₀ , i ₀ +2, i ₀ +4, ... and each line image signal of scanning line numbers i ₂ , i ₂ +2, i ₂ +4, ... in the 1-frame delayed image signal 53 are loaded respectively. The added and synthesized result is removed by the sub-sampling circuit 3 in the configuration shown in the first diagram, and the scanning line numbers i ₁ , i ₁ + that are not transmitted are removed.
2, i ₁ +4, . . . are supplied to the time axis conversion circuit 51 as an interpolated image signal 56 for reconstructing each line image signal. Note that these interpolated output image signals 55
and 56, the former is added to a low-pass waveform in the vertical spatial frequency axis direction to extract its low-frequency components, and the latter is also added to a high-pass waveform in the vertical spatial frequency axis direction to extract its high-frequency components. Take out the
The scanning line number is determined by combining these high and low band components.
It is also possible to reconstruct each line image signal of i ₁ , i ₁ +2, i ₁ +4, . . . . However,
According to the configuration shown in FIG. 17, each input image signal of the three-dimensional interpolation filter 50 is in the form of a low-speed scanning narrowband image signal as shown in FIG. The signal processing speed of the interpolation filter 50 can be made 1/2 of the signal processing speed of the three-dimensional interpolation filter 7b in the configuration shown in FIG. .

After or before performing the interpolation wave processing as described above, the low-speed scanning interlaced narrowband image signal is processed by the time axis conversion circuit 51.
Alternatively, the signal is guided to 7b and subjected to time axis conversion, and interpolation of scanning lines is performed to convert it into a high-speed line sequential scanning image signal. An example of the configuration of a time base conversion circuit for performing such signal conversion is shown in FIG. 19a, and the mode of scanning line compression and interpolation is shown in FIG. 19b. In the illustrated configuration, interpolated output image signals 55 and 5 taken out from the interpolation filter 55 described above are used.
1-line memories 63-1, 63-2 and 63-3, 63 in which two odd-line image signals 57 and even-line image signals 58 respectively corresponding to 6 are arranged via a double-pole double-throw switch circuit 59. −
4 alternately at low speed, read out alternately at high speed via the 2-pole, double-throw switch circuit 60, and then alternately switched by the double-throw switch circuit 61, as shown in FIG. , high-speed scanning odd line image signals a, b, ... and c, d, ...
are taken out alternately to obtain each line image signal a, as a high-speed line sequential scanning broadband image signal.
c, b, d,... are constructed.

If the high-speed line-sequential scanning broadband image signal that has been subjected to the interpolation wave processing and time axis conversion processing as described above is supplied to the display device 8 made of an ordinary cathode ray tube, etc., the sampling aliasing that occurs due to signal conversion can be avoided. It is possible to display a high-quality image that sufficiently prevents image quality deterioration due to contamination of components.

Next, although image quality deterioration due to the mixing of sampling aliasing components is prevented by applying the signal processing according to the present invention, an image signal processing apparatus having a configuration that greatly simplifies the configuration shown in FIG. Each of the transmitting components 1, 2, 3 and 4 is replaced with a normal 2:1 interlaced television camera, and the standard television image signal obtained from the television camera, i.e., the deviation as described above, is obtained. For example, a slow scanning interlaced narrowband image signal is used in place of the slow scanning interlaced narrowband image signal as shown in FIG. 3c in the configuration shown in FIG. 1 and is supplied to the modulation device 5. Therefore, with such a simplified configuration, the three-dimensional low frequency processing and time axis conversion processing according to the method of the present invention are performed only on the receiving side. Note that in the image signal processing of the present invention method with such a simplified configuration, the image signal supplied from the television camera on the transmitting side to the modulation device 5 is the standard television image signal itself. It also has the advantage of being able to be used separately as a standard television image signal.

Furthermore, the image signal processing according to the method of the present invention described above is as follows:
It can be applied not only to black and white image signals but also to color image signals. Basically, each primary color image signal R, G, B or luminance signal Y and each color difference signal R- Y, B-Y
can be applied in parallel to each of the component image signals. However, such 3
If the three-dimensional wave processing to be applied to each component image signal is performed in the following manner, the number of frame memory devices required on the receiving side can be reduced.

That is, for example, an example will be described in which a high-speed line sequential scanning color image signal is converted into a high-speed line sequential scanning wideband color image signal through a process of converting a low-speed scanning 2:1 interlaced color image signal. Then, as described above with reference to FIG. 8, the sampling frequency point for three-dimensional wave processing when converting a line-sequential scanning image signal to a 2:1 interlace image signal is the frequency point 1/1 on the vertical frequency axis. The virtual origin O _S2 is related to 2 _vp and the frequency point 1/2 _tp on the time axis. Therefore, compared to the luminance signal Y constituting the color image signal, each color difference signal B−Y,
As is well known, the required resolution of R-Y is extremely low.
Since the standard color television image signal actively utilizes this characteristic, the image signal processing method of the present invention also actively utilizes this characteristic. In particular, the prefilter 2 on the transmitting side and the interpolation filter 7b on the receiving side in the configuration shown in the first figure
The passband of the two-dimensional low-pass filter used for Although a two-dimensional low frequency filter using a half-reduced rectangular area and having such a passband is constructed from a frame memory and a line memory and used as the prefilter 2 on the transmitting side, as the interpolating filter 7b on the receiving side, the vertical Only in the spatial frequency axis direction, we use only a cut-off frequency of 1/2 _vp of the normal low-frequency wave generator on the frequency axis, and in the time axis direction, we use a normal color filter without any low-frequency wave generator. Similar to the display of television images, the sampling aliasing component is visually suppressed in the display image so that it does not cause any disturbance to the color image quality. be able to. Note that for the color difference signals B-Y and R-Y, which require significantly lower resolution, even if the passband width in the vertical spatial frequency axis direction is significantly reduced as shown by the dashed line in FIG. There is virtually no deterioration in the quality of the displayed color image. Also,
With this configuration, the frame memory for the color signal can be omitted on the receiving side, but for the luminance signal Y, as mentioned above, the frame memory can also be used in the time axis direction. Since the processing is actually performed, the luminance signal Y will be delayed in frame time units with respect to each color difference signal R-Y, B-Y.
It is necessary to add a delay corresponding to the delay time in units of frame time to each color difference signal R-Y, BY in advance on the transmitting side before transmitting it to the receiving side.

Next, an example of the configuration of an image signal processing device in the case where the image signal processing of the present invention method is applied to a coded image signal of the interframe coding method is shown in FIG.
This will be explained with reference to b. Figure a shows the basic configuration of an image signal encoding device using an interframe coding method, and if the present invention is applied to an image signal encoding device with such a basic configuration, Figure b The configuration will be as shown. That is, in the image signal encoding device using the interframe encoding method shown in FIG. 20a, the input analog image signal 64 is
The quantized image signal obtained from the quantizer 65 is supplied to the quantizer 65 via the adder ADD12, and the quantized image signal obtained from the quantizer 65 is written to the frame memory 66 via the adder ADD13. The quantized image signal is guided to the adder ADD13 and added to the quantized image signal from the quantizer 65, and the adder ADD12
and subtracts the interframe difference image signal from the input image signal to form an interframe encoded image signal and transmit it during reception. On the other hand, on the receiving side, the received interframe encoded image signal is supplied to the adder ADD14, and the added output image signal is supplied to the frame memory 67. In the frame memory 67,
The image signal of one frame before is written as an initial condition, and the image signal is read out as a predicted image signal and supplied to the adder ADD14, and added to the input interframe image signal to generate a new image signal of one frame. By writing the image signal of one frame into the frame memory 67, the adder ADD
Output frame image signals 68 are sequentially extracted from 14. On the other hand, in the interframe image signal encoding device to which the image signal processing method of the present invention shown in FIG. A low-speed scanning interlaced narrowband image signal as shown in FIG. In this case, the frame memory 67 for extracting the predicted image signal is not only used for forming the predicted signal, but also used as a frame memory for the three-dimensional interpolation filter 7b in the configuration shown in the first figure. That is, if we take the case of converting a line-sequential scanning image signal to an interlaced image signal as an example, the frame memory 67 is divided into two to create two field memories, and these two field memories are arranged as shown in FIG. It is used as one field memory 47, 48 in the receiving side signal processing device of the present invention having the configuration shown in FIG. 20a.
By configuring the receiving side circuit device in the image signal interframe encoding device having the basic configuration shown in FIG. Also on the receiving side of the image signal encoding transmission system using the image signal processing method of the present invention, three-dimensional interpolation filter processing is performed using the image signal processing method of the present invention to prevent the reproduction image from being mixed with sampling aliased components caused by quantization during encoding. Image quality deterioration can be prevented.

Further, the image signal processing method of the present invention can be applied to a receiving side circuit device in so-called still image broadcasting in the same manner as described above for the interframe image signal encoding device, and the same effects as described above can be obtained. can. In other words, in a transmission system for still image broadcasting, a frame memory is provided on the receiving side, and a desired still image signal among various still image signals transmitted frame by frame is stored in the frame memory and repeatedly reproduced and displayed. do. Therefore,
Even in still image broadcasting, the frame memory provided on the receiving side is divided into two into two field memories, and these two field memories are used in the receiving side processing device of the present invention having the configuration shown in FIG. If the structure shown in FIG. 17 is used as one field memory 47, 48 and similar signal processing is performed, three-dimensional interpolation filter processing can be easily performed and a high-quality image can be reproduced and displayed.

Next, an example will be explained in which the image signal processing of the present invention is applied to contour compensation for improving the sharpness of television images, and contour compensation is performed for three-dimensional frequency components including the time axis. do.

Conventionally, in order to improve the sharpness of television images, a contour compensation circuit having a configuration as shown in FIG. 21 has been used. In the configuration input shown in the figure, a one-line delay line 72 in which two image signals 71 are connected in series,
72' and the input image signal are added to the adder ADD18 and the 6 dB attenuator 73 to form an average image signal, and the average image signal and the 1 line delayed image signal are combined. is added to the adder ADD19 and the variable attenuator 74 and the obtained difference image signal is added to the one-line delayed image signal by the adder ADD20, thereby producing an output image signal 7 that has undergone contour compensation in the vertical spatial frequency axis direction.
I got a 5. The spatial frequency characteristics of such a contour compensation circuit are as shown in FIG. 22 for a standard television image, and because the standard television image performs 2:1 line interlacing Basically, by using a one-line delay line, a vertical spatial frequency characteristic related to the number N of scanning lines is obtained as shown in the figure. In addition, contour compensation in the horizontal spatial frequency axis direction is
One line delay line 72, 72' in the configuration shown
Instead, a delay line having a delay amount of about several hundred nanoseconds is used to make the horizontal spatial frequency characteristics match the vertical spatial frequency characteristics described above. Therefore, its vertical spatial frequency characteristic is
As shown in FIG. 2, the center frequency of the boost region for contour compensation is located at N/4 with respect to the number N of scanning lines. In addition, in FIG. 22, the spatial frequency of the television image is normalized according to the height of the screen, and for an image with the number of scanning lines N,
It is assumed that up to N/2 spatial frequencies can be transmitted and displayed. Therefore, the center frequency N/4 of the spatial frequency characteristic boost region for sharpness improvement obtained by the conventional contour compensation circuit with the configuration shown in FIG. 21 is equal to the spatial frequency N/4 that can be transmitted and displayed. The disadvantage was that it was only 1/2 of 2, and sufficient contour compensation could not be performed.

In order to eliminate such conventional drawbacks and increase the center frequency of the vertical spatial frequency characteristic boost region, it is possible to use a 1-field delay line instead of a 1-line delay line. When contour compensation is performed in the vertical spatial frequency axis direction using the configuration described above, extremely strong flicker disturbance occurs in the details of the image in a standard television image that performs 2:1 line interlacing, and the displayed image quality is significantly degraded. It will cause deterioration. In addition, since conventional television cameras use the accumulation effect of the image pickup tube to increase sensitivity, the movement of the subject causes severe blurring in the image, which also affects the display image quality. This will cause significant deterioration.

In order to solve the above-mentioned problems regarding contour compensation of television images by applying the image signal processing concept of the present invention, and to perform contour compensation in a three-dimensional coordinate space including the time axis, as shown in FIG. As such, a circuit device having exactly the same configuration as that shown in the first figure is used.

To briefly explain the configuration shown in FIG. 23, a line-sequential scanning image signal with the number of scanning lines N from the camera 1 is supplied to the three-dimensional low-pass filter 2, the signal spectrum distribution region is limited, and then Axis conversion circuit 4
The signal is supplied to the modulator 5, where it is converted into a 2:1 line interlaced image signal, and further supplied to the modulator 5, where it is converted into a carrier signal. Therefore, the required passband of the three-dimensional low-pass filter 2 is a hatched triangular area, as described in detail above, as shown in FIG. 24 again. That is, in the frequency domain distribution shown in FIG.
The signal spectrum distribution in the frequency domain surrounded by the dotted line is caused by converting the line sequential scanning camera output image signal into a 2:1 line interlaced image signal. , has unnecessary sideband signal component distribution regions surrounding each of the points C, D, E, etc. in the figure, and these unnecessary signal components cause so-called aliasing distortion components in the original signal spectrum distribution region including the coordinate origin B. In order to prevent the mixture from occurring, a three-dimensional low-frequency wave filter 2 is provided. As described above, when converting to a 2:1 line interlaced image signal, it essentially functions as a two-dimensional low frequency filter.

Next, on the receiving side, the 2:1 line interlaced image signal demodulated by the demodulator 6 is reconverted into a line sequential scanning image signal by the time axis conversion circuit 7a, but the converted output image signal includes: Sideband components in the frequency domain surrounding each point A, C, and D in the frequency domain distribution shown in Figure 24 are included, and when those sideband components adjacent to the frequency domain surrounding point B are displayed, the reproduction Since this significantly degrades the image quality, a three-dimensional low-frequency filter is provided as an interpolation filter 7b on the receiving side as well, to sufficiently suppress such unnecessary signal components. Note that, in general, sideband components in the frequency domain surrounding point C in the frequency domain distribution shown in FIG. Also, since the entire transmission system is designed so that the sideband components in the frequency domain surrounding point D are naturally suppressed by the decrease in the visual MTF characteristics in the time axis direction, The sideband component at point A mainly affects the reproduced image quality since it does not significantly impair the quality of the reproduced image.

Next, the case where the contour compensation according to the method of the present invention in the aspect described above is applied to a television image transmission system will be explained. Contour compensation is also performed in the axial direction, 2:1
For line interlaced image signals,
The signal spectrum distribution limited region shown in FIG. 24 does not have a simple low-pass characteristic, but contour compensation is performed by boosting the vicinity of the upper limit frequency of the passband. Examples of such contour compensation are sequentially shown in FIGS. 25a, b, and c. That is,
By expanding the upper limit frequency region near the diagonal line in the above-mentioned triangular required passband shown in FIG. 25a,
Boosting is performed as indicated by diagonal lines, and Figure b shows the boost characteristic in the vertical spatial frequency axis direction, and Figure c shows the boost characteristic in the time axis direction.

Although the boost mode of the three-dimensional band-limited waveform generator shown in FIG. It is also necessary to configure a band-limiting waveformer with this in mind. Therefore, it is preferable to approximate the three-dimensional band-limited wave characteristics shown in FIG. 25a by a combination of simple wave characteristics that are easy to realize as shown in FIG. In order to realize the combination of wave characteristics, a circuit configuration as shown in FIG. 27 is used. That is, in the configuration shown in FIG.
Reference numeral 8 is a band-limiting filter in the time axis direction, and its passband characteristic is as shown in the amplitude/frequency characteristic curve a shown in FIG. 28, and the area a in the passband characteristic shown in FIG. 26 is boosted. do. Also,
Reference numeral 79 in the configuration shown in Figure 27 is a band-limiting transducer in the vertical spatial frequency axis direction, and its passband characteristics are as shown in the amplitude/frequency characteristic curve a shown in Figure 29, and the passband characteristics shown in Figure 26 are as follows. Boost region b in the band characteristics. Furthermore, 80 in the configuration shown in FIG. 27 is a band-limiting wave filter in the time axis direction, and its pass band characteristics are as shown in the amplitude/frequency characteristic curve b shown in FIG. Boost region c in the passband characteristics. Further, 81 in the configuration shown in FIG. 27 is a band-limiting filter in the vertical spatial frequency axis direction, and its pass band characteristics are as shown in FIG.
It has a frequency characteristic curve b, and boosts the region d in the passband characteristic shown in FIG. Each of these band-limiting wavers can be easily realized, for example, by a so-called transversal filter having a configuration similar to that shown in FIG. can use a memory element for one field of a line-sequential scanning image signal as a unit delay element of the transversal filter, and the band limiters 79 and 81 in the vertical spatial frequency axis direction can be used as a unit delay element of the transversal filter. A memory element for one line of a line sequential scanning image signal can be used as the unit delay element.

Furthermore, in order to approximate the required passband characteristics shown in FIG. 25a more precisely than the approximate characteristics shown in FIG. 26, for example, a combination of approximated passband characteristics as shown in FIG. is suitable.
That is, as shown with diagonal lines in FIG.
In the same way as shown in FIG. 27, by boosting the region approximated in the step shape of minute steps,
Furthermore, it is constructed by combining a large number of band-limiting wave devices in the time axis direction and the vertical spatial frequency axis direction, each having a wide variety of amplitude and frequency characteristics.

Regarding contour compensation in the horizontal spatial frequency axis direction, a contour compensation circuit in the horizontal spatial frequency axis direction is performed using a delay line having a delay amount of about several hundred nanoseconds in exactly the same manner as the conventional method described above. is delayed and used in a form in which it is cascaded to the contour compensation circuit in the time axis direction and the vertical spatial frequency axis direction described above. Further, the boost frequency region at this time is preferably a frequency region corresponding to region b in the passband characteristic shown in FIG. 26 among the boost frequency regions in the vertical spatial frequency axis direction.

Note that in order to simplify the configuration of the contour compensation circuit according to the present invention described above, both or one of the band-limiting waveforms in the time axis direction corresponding to regions a and c in the passband characteristic shown in FIG. 26 may be omitted. You can also do it. That is, in the configuration shown in FIG. 27, the band-limiting waver 78 in the time axis direction corresponding to region a in the passband characteristic shown in FIG. 26 is relatively important and difficult to omit. Although the increase in the number of circuit elements due to the provision of the
By omitting the band-limiting waveform generator 80 in the time axis direction corresponding to , the number of circuit elements can be reduced considerably.

In addition, in the above explanation, contour compensation using the image signal processing according to the present invention has been described only in the case of converting a line sequential scanning image signal to a 2:1 line interlaced image signal. Contour compensation according to the method of the present invention is applicable not only to an interlace ratio of 2:1, but also when converting to a line interlace image signal with a higher interlace ratio, or to a dot interlace image signal, as described above. The required passband of the three-dimensional band-limiting wave filter is set as described in detail earlier, and the region near the upper limit of the required passband is shown in Figure 25a. Contour compensation can be performed by boosting in a similar manner.

Therefore, image contour compensation to which the image signal processing method of the present invention is applied only needs to be performed on the transmitting side, as in the case of contour compensation for conventional standard television images, and there is no need to perform it individually on the receiving side. However, if it is also performed on the receiving side, contour compensation can be performed even better. In that case, the interpolation filter 7 on the receiving side in the configuration shown in FIG.
A required boost characteristic is imparted to the pass band of the three-dimensional band-limiting waveform filter constituting b. Note that when three-dimensional contour compensation is performed by providing a three-dimensional interpolation filter on the receiving side and boosting its passband in this way, the three-dimensional contour compensation on the transmitting side may be omitted. Therefore, even when using a 2:1 line interlaced imaging camera on the transmitting side as in the past, the effects of applying the image signal processing of the present invention to contour compensation can be fully demonstrated. can.

As is clear from the above description, according to the present invention, on the transmitting side of a television image transmission system, the contamination of the three-dimensional sampling aliasing distortion component that occurs due to the scanning of the imaging camera into the original signal spectrum distribution region is prevented. Therefore, it is possible to sufficiently eliminate the image quality deterioration of the reproduced display image due to the mixing of the sampling aliasing distortion component, and it is possible to perform high-quality image transmission. In addition, on the receiving side, display is performed by removing three-dimensional sampling frequency points and their sideband components included in the received signal or generated by signal processing on the receiving side using a three-dimensional interpolation filter. The screen can display a high-quality image consisting only of the original baseband frequency component, and eliminates image quality deterioration due to the contamination of three-dimensional sampling frequency components and their sideband components, which were previously difficult to remove. It is possible to display high-quality images with significant improvements. If the removal of unnecessary signal components by such a three-dimensional wave device is applied to both the transmitting side and the receiving side of a television image transmission system, the most desirable image transmission system can be constructed in principle.

Furthermore, if a three-dimensional interpolation filter with the minimum performance required to sufficiently suppress unnecessary signal components is used on the receiving side, the image quality of the reproduced display image can be improved, and the three-dimensional bandwidth of the receiving side can be improved. Since the limiter can be simplified and made smaller, the cost of the receiving circuit device can be reduced.

Furthermore, if the receiving side performs image signal processing using a 3-dimensional interpolation filter and then performs time axis conversion of the image signal, the required speed of image signal processing using the 3-dimensional interpolation filter can be reduced. , it is possible to easily reduce the cost of the receiving side circuit device.

In addition, the transmitting side does not perform any special image signal processing such as three-dimensional band limiting, subsampling, or time axis conversion, and transmits the normal standard television image signal as is, and only the receiving side performs the above-mentioned processing. By performing three-dimensional interpolation filter processing and time axis conversion, the image signal processing of the present invention can be applied to normal television image signal transmission.

In addition, 3
An image signal subjected to dimensional band limiting, subsampling, and time axis conversion originally has a signal form that is almost the same as an interlaced image signal, so the image signal processed by the present invention can be used as it is as a normal image signal. It has the advantage of being compatible with standard television images. That is, the transmitting side output image signal processed by the configuration shown in the first figure can be used as a standard format television image signal as it is by converting it to the 2:1 line interlaced format. Even if the structure shown in the figure is played back and displayed in the conventional manner without performing three-dimensional interpolation filter processing or time axis conversion, there is little deterioration in image quality due to the mixing of sampling aliasing distortion components, and the image quality is much superior to that of conventional methods. Images can be displayed.

Furthermore, when applying the image signal processing of the present invention to color image signals, the frame memory required for three-dimensional wave processing on the receiving side can be simplified for color signals with a narrow frequency band, so that black and white images can be processed easily. The capacity of the frame memory is not required as much as can be inferred from the amount of frame memory required for signal processing, and the receiving side circuit device can be made simpler and cheaper.

In addition, when applying the image signal processing of the present invention to interframe encoding of an image signal, three-dimensional band limitation is applied according to the signal format before performing interframe encoding processing on the image signal on the transmitting side. Therefore, the amount of transmitted information of the inter-frame difference signal can be reduced, and on the receiving side, the frame memory for forming the predicted frame image signal can be replaced with the frame memory for three-dimensional wave processing using the method of the present invention. Therefore, the number of required frame memories can be reduced, and the decoded output image signal, which has significantly improved image quality by using the image signal processing method of the present invention, can be reproduced and displayed using a relatively inexpensive circuit device. Can be done.

Furthermore, when the image signal processing of the present invention is applied to still image broadcasting, the frame memory for repeatedly reproducing and displaying the desired still image signal frame by frame on the receiving side can be replaced with the 3D waveform of the present invention. Since it can be used as a frame memory for processing, it is possible to reproduce and display still images whose image quality has been significantly improved by the image signal processing method of the present invention without significantly increasing the manufacturing cost of the receiving side circuit device. This can have a significant effect on improving frame memory usage efficiency and popularizing still image broadcasting services.

Furthermore, when applying the image signal processing of the present invention to contour compensation of an image, conventionally, the frequency range in which boosting is applied for contour compensation in the vertical spatial frequency axis direction of the image is limited to the number N of scanning lines. While it remained in a low frequency region with a peak of N/4,
By using the image signal processing method of the present invention in combination, it is possible to increase the boost region for contour compensation in the vertical spatial frequency axis direction to a frequency region near N/2 for images with little temporal change. In addition, since it is possible to secure the boost frequency region around N/4 for images with large movements as before, the image quality improvement effect by contour compensation can be significantly increased compared to the conventional method, and the sampling aliasing distortion By removing the components, it is possible to reproduce and display an image of extremely high quality.

Furthermore, if the image signal processing of the present invention is used in conjunction with image contour compensation, contour compensation will also be applied to the frequency characteristics in the time axis direction, so that when the subject moves, the contour portion will be emphasized. This makes it possible to significantly improve image quality deterioration due to blurring of moving images, which conventionally occurred in such cases due to the accumulation effect of the image pickup tube.

In addition, since the image contour compensation using the image signal processing of the present invention need only be applied on the camera side, there is no need to individually perform contour compensation on the receiving side, and the entire system can be configured economically. However, if a three-dimensional band-limiting waveform generator is used on the receiving side, contour compensation using the above-mentioned image signal processing according to the present invention may be required, which makes the receiving side circuit device particularly expensive. Since this can be done individually on the receiving side, the sending side can continue as before, and high-quality images can be reproduced and displayed by image signal processing only on the receiving side. .

[Brief explanation of drawings]

FIG. 1 is a block diagram showing the basic configuration of an image signal processing device according to the present invention, FIG. 2 is a diagram showing the operating principle of image signal processing, and FIG. 3 is a block diagram showing the signal waveforms of each part shown in FIG. 1. FIG. 4 is a characteristic curve diagram showing an example of the frequency characteristics of each part of the signal, and FIG. 5 is a block diagram showing an example of the detailed configuration of the time axis conversion circuit in the configuration shown in FIG. 1. , FIG. 6 is a diagram showing the aspect of the three-dimensional configuration of an image to which the image signal processing according to the present invention is applied, and FIG. 7 is a diagram showing the aspect of the frequency component distribution in the three-dimensional configuration of the image. , FIG. 8 is a diagram showing an example of the required passband of the three-dimensional low-pass filter in the configuration shown in FIG. 1, and FIGS. 9 a and b are examples of the actual passband of the three-dimensional low-pass filter. 10a and 10b are diagrams and block diagrams respectively showing an example of the circuit configuration, and FIGS. Figure 11 is a block diagram showing an example of the detailed configuration of an actual low-frequency device that realizes the three-dimensional low-frequency device, and FIG. 12 is the same three-dimensional low-frequency device. FIG. 13 is a diagram showing another example of the required passband of the three-dimensional low-pass filter, and FIG. 14 is a diagram showing another example of the required passband of the three-dimensional low-pass filter. Figure 15 is a diagram showing another aspect of the frequency component distribution in the three-dimensional low frequency filter; Figure 16 a, b, c
17 is a diagram and a block diagram respectively showing still another example of the required pass band and another example of the circuit configuration of the three-dimensional low-pass filter, and FIG. A block diagram showing another example of the side configuration, Fig. 18 is similar to Fig. 17.
A signal waveform diagram showing a mode of image signal processing with the illustrated configuration; FIGS. 19a and 19b are block diagrams and signal waveforms showing still another example of the receiving side configuration and a mode of image signal processing with that configuration, respectively. 20a and 20b are block diagrams respectively showing the configuration of an image signal encoding device applying image signal processing according to the present invention and an example of its application, and FIG. 21 is a block diagram showing a conventional image contour compensation circuit. A block diagram showing an example of a configuration to which the invention method is applied, FIG. 22 is a characteristic curve diagram showing the conventional compensation characteristics of an image signal compensation circuit, and FIG. 23 is an image contour compensation using the image signal processing according to the invention method. FIG. 24 is a block diagram showing an example of the configuration of the circuit, and FIG. 25 is a diagram showing the operating principle of the image contour compensation circuit.
Figures a, b, and c are diagrams and characteristic curve diagrams respectively showing examples of the required passband and boost characteristics of the three-dimensional low-pass filter used in the image contour compensation circuit, and Figure 26 is the same three-dimensional Figure 27 is a diagram showing an example of the actual passband of a low-pass filter, Figure 27 is a block diagram showing an example of the detailed configuration of the three-dimensional low-pass filter, and Figure 28 is a diagram showing an example of the three-dimensional low-pass filter. FIG. 29 is a characteristic curve diagram showing another example of the boost characteristic of the device, FIG. 30 is a characteristic curve diagram showing still another example of the boost characteristic, and FIG. It is a line diagram showing other examples of. 1...Imaging camera, 2...3-dimensional prefilter, 3
... Subsample circuit, 4, 7a... Time axis conversion circuit, 5... Modulator, 6... Demodulator, 7b... Three-dimensional interpolation filter, 8... Display device, 9... Input image signal, 10, 17... Low pass wave Vessel, 11...
Analog to digital converter, 12, 15, 1
9, 20... Selector switch, 13, 14... 1-line delay line, 16... Digital-to-analog converter,
18... Output image signal, 21... Input clock signal,
22... 1/2 frequency divider, 23... Input image signal, 24... Time axis low pass wave generator, 25... Vertical frequency axis direction low pass wave generator, 26... Output image signal, 27... Input image signal,
28, 29... time axis direction low pass wave generator, 30,
31...Vertical frequency axial low pass wave generator, 32...
Output image signal, 33... Input image signal, 34... 1 frame or 1 line delay line, 35... Weighting factor unit, 36... Output image signal, 37, 42... Input image signal, 38... Time axis direction low pass wave generator ,39...
Vertical frequency axial low-pass transducer, 40...Horizontal frequency axial low-pass transducer, 41, 45...Output image signal, 43, 44...Three-dimensional low-pass transducer, 46,
52... Input image signal, 47, 48... 1 field memory, 49... 1/2 line memory, 50... Three-dimensional interpolation filter, 51... Time axis conversion circuit, 53... 1
Frame delay signal, 54...1 field + 1/2 line delay signal, 55, 56...Interpolation output signal, 5
7, 58...Interpolation output image signal, 59, 60, 61
...Selector switch, 62...Output image signal, 63-
1 to 63-4...1 line memory, 64...input image signal, 65...quantizer, 66, 67...frame memory, 68...output image signal, 71...input image signal, 72, 72'...1 line delay line , 73...attenuator, 74...variable attenuator, 75...output image signal, 7
7... Input image signal, 78, 80... Time axis direction low pass wave generator, 79, 81... Vertical frequency axis direction low pass wave generator, 82... Output image signal, ADD1~
ADD22...adder.

Claims (1)

[Claims] 1. When converting between sequential scanning and interlaced scanning performed by sampling an image signal, a horizontal frequency axis expressed by a sampling frequency and a horizontal scanning frequency, a vertical frequency axis expressed by a vertical scanning frequency, and a number of images per second are used. At least one sampling frequency point generated based on the sampling of the image signal in a multidimensional coordinate consisting of a coordinate plane formed by at least a vertical frequency axis and a time axis among the time axes represented is approximately equal to the origin of the multidimensional coordinate, respectively. By sampling the image signal using at least one multi-dimensional low-pass filter whose desired pass band is a region corresponding to the origin side region of the multi-dimensional coordinates of a line connecting points at a distance. In order to suppress the generated high-frequency unnecessary signal components on the coordinate plane defined by the vertical frequency axis and the time axis, a plurality of delay elements sequentially connected in cascade are used, each with one line memory and one frame memory as delay elements. a transversal filter having a transversal filter having a transversal filter having a transversal filter having a transversal filter having a transversal filter having a transversal filter having a transversal filter having a transversal filter having a transversal filter having a transversal filter having a transversal filter having a transversal filter having a transversal filter having a transversal filter having a transversal filter having a transversal filter having a transversal filter having a transversal filter having a transversal filter having a transversal filter having a transversal filter having A vertical frequency-axis low-pass filter and a time-axis low-pass filter each having an upper limit of a pass band set at successive frequency points divided equally, respectively, and the input of the time-axis low-pass filter and The high-pass filter outputs of the time-axis high-pass filters are guided to at least the vertical frequency-axis low-pass filter, and the low-pass filter outputs are respectively guided to a subtracter and the subtracted outputs obtained by subtracting each other are used as high-pass filter outputs. and directly or indirectly guide the low-pass filtered output of the time-axis direction low-pass filter to the adder to make at least the summed outputs mutually added as two-dimensional low-pass filtered outputs. An image signal processing method characterized in that a two-dimensional low-pass filter is used as at least a part of the multi-dimensional low-pass filter. 2. In the image signal processing method according to claim 1, the transfer function of the high frequency portion of the passband on the coordinate plane formed by at least the vertical frequency axis and the time axis of the multidimensional low-pass filter is An image signal processing method characterized in that the signal component in the high frequency range is increased by making the transfer function larger than the transfer function in the low frequency range. 3. The image signal processing method according to claim 1 or 2, characterized in that the multi-dimensional low-pass filter is provided on both the transmitting side and the receiving side of the image signal transmission system, respectively. Image signal processing method. 4. The image signal processing method according to claim 1 or 2, characterized in that the multidimensional low-pass filter is provided only on the transmitting side of the image signal transmission system. method. 5. The image signal processing method according to claim 1 or 2, wherein the multidimensional low-pass filter is provided only on the receiving side of the image signal transmission system. method.