JPS6037680B2 - Encoder/decoder for narrowband color television signals - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an encoding/decoding apparatus for composite color television signals such as NTSC and PAL systems.

In color television signals such as the NTSC system, a carrier color signal (hereinafter referred to as a C signal) obtained by modulating a subcarrier with a color signal is frequency multiplexed so that the frequency spectrum is interleaved with a luminance signal (hereinafter referred to as a Y signal). signal format is used. As a method for efficiently encoding such a composite color television signal, the composite color television signal is directly encoded without demodulating into base band signals (for example, luminance signal Y, chrominance signal 1, and chrominance signal Q in the NTSC system). As methods for doing this, there are direct assisted coding methods such as high-order predictive DPCM and orthogonal transform coding such as Hadamard transform. The direct encoding method directly encodes a composite color television signal containing a C signal with a frequency spectrum concentrated in the high frequency region, so when sampling the signal, the sampling frequency fs is 2 It is necessary to make the value more than double,
Conventionally, the value (N
In the case of TSC color television signals, it was selected as 9-11M ratio). For this reason, the transmission bit rate required to transmit the encoded signal tends to be large because the sampling frequency is high. Furthermore, if sampling is performed at a frequency lower than twice the frequency of the signal band (hereinafter referred to as Nyquist frequency), the frequency spectrum will be folded and overlapped, resulting in distortion of the waveform and the color signal may not be reproduced correctly. Waveform distortion caused by folding of the spectrum is referred to as aliasing distortion.) The configuration of the filter circuit to remove aliasing distortion (in the frequency domain, the folded frequency spectrum, or aliasing spectrum) was complicated. The purpose of the present invention is to set the sampling frequency $ to a special condition (approximately 2h or 11 times the value of Sagi c) in an encoding method that performs sampling at a frequency lower than the Nyquist frequency (hereinafter referred to as sub-Nyquist sampling). , where m and n are positive integers) to easily configure a filter circuit that removes aliasing distortion, and to provide an encoding/decoding device that realizes highly efficient band compression encoding with a simple device. It is in.

The encoding/decoding device of the present invention has f in the frequency domain.
In order to prevent the folding of the C signal near sc from overlapping the C signal itself, the sampling frequency fs should be lower than twice the subcarrier frequency c, and the subcarrier frequency &C reservation (m
A coding device controlled to sample a composite color television signal at a frequency that is a ``quasi-positive integer'' and a received signal sampled by the coding device at a sampling frequency $. ■ A decoding device having a function of removing aliasing distortion and reproducing a composite color television signal using a filter that includes a function of S (where m is the same value as m above). It consists of

According to the present invention, it is possible to directly sample a composite color television signal at a sampling frequency lower than twice the subcarrier frequency fsc, that is, at a frequency lower than the Nyquist frequency, thereby eliminating aliasing distortion of the decoding device. The structure of the filter circuit to be removed becomes simple, and an efficient encoding/decoding device can be obtained.

Hereinafter, a case where the signal format is an NTSC composite color television signal will be explained using the drawings as an example.

In the frequency multiplexing color television signal system of NTSC variation, the frequency range of the frequency spectrum of the C signal is set to Y in order to reduce image quality deterioration due to mutual interference between the Y signal and C signal.
The frequency spectrum of the Y signal (hereinafter abbreviated as frequency and referred to as the spectrum of the Y signal) is reflected in the high frequency portion of the signal frequency spectrum where the amplitude is small.

) The subcarrier frequency fsc is in a frequency interleap relationship with the horizontal scanning frequency fH so that the frequency spectrum of the C signal (hereinafter abbreviated as frequency and referred to as the spectrum of the C signal) is included between the subcarrier frequency fsc and the horizontal scanning frequency fH. That is, in the case of the NTSC system, Tsuru c = thing 5 · fH sardine wetness is small. In general, the distribution of the frequency spectrum of a TV signal is such that in a normal video signal that spreads over the entire frequency band, the spectrum of the Y signal is centered around a frequency that is an integer multiple of fH, and in the vicinity thereof, and in the low frequency range. The amplitude of the frequency spectrum in the Takagi part is very small. Furthermore, the spectrum of the C signal can be considered to be concentrated in the frequency range near $c.

Therefore, fs is selected so that the center of the spectrum of the C signal has an appropriate relationship between c and the center of the folded spectrum of the C signal where 1' is folded at s (degrees - fsc), assuming the sampling frequency is . .

That is, $ is made to satisfy the relationship between fsc and m, '2} below. fs<2sc [1
)$=2;lfSC {21 However, m and n are positive integers. As the first example, {1},
A case where m=2 and n=2 as fs that satisfies '2), that is, a case where the sampling frequency fs is selected to a value of fs=5'4fsc will be explained.

If a multi-sign color television signal with an amplitude distribution of the entangled frequency spectrum shown in Figure 1a is sampled using the sampling frequency heron, the result will be as shown by the broken line in Figure 1b. A folded spectrum results. $c (i.e. 4
The folding of the C signal near fs-fsc (i.e., 1/8s) occurs concentratedly near fs-fsc (ie, 1/8s). On the receiving side, if the value of 0 is interpolated between each sent signal and resampled to a PCM signal with a sampling frequency of s, the spectral distribution of the resampled signal will be as shown in Figure 1c. It becomes like this. In order to obtain a composite color television signal with less deterioration in image quality, it is necessary to remove at least the folded spectrum of the C signal near fs-fsc and the folded spectrum of the Y signal near fs from among the folded spectra. Here, if the sampling frequency $ is selected as shown in the formula '2}, the common fs
c is shown as follows.

fS - fSC = × Wellness; lfS '3'$C; Public h
Kajulfs {4) Rescue MS
C'Magazara's strange vertical frequency Sagi Shikufu has a frequency several times that of Mifune's fox voice.

Therefore, if m=2, n=2, K-fsc
and fsc are frequencies that are odd multiples and even multiples of 1/as, respectively. The frequency characteristics are better than this (Sora Nikesa)
In other words, if a filter including the function of ■S(5/Fuki) is used, the amplitude of the frequency f in the vicinity of −$c=1/gs, that is, its insulation value, is much smaller than 1, and the frequency f becomes fsc=4. Obtain a frequency characteristic in which the absolute value is almost equal to 1 near /5$, the absolute value is almost equal to 1 near the frequency 0, and the absolute value is much smaller than 1 near the long frequency. be able to. For frequencies above fs, the frequency characteristics of the filter are symmetrical with respect to fs, but the spectrum and distribution of the signal sampled at s are also symmetrical with respect to fs. ■If the digital filter showing the frequency characteristics of S (5'2 poisoning) is represented by Z transformation with the sampling frequency as s, then the transfer function (Z) will be (Z)! It can be given by a simple raw function.

Therefore, for example, the transfer characteristic day (Z) shown by Z transformation is day (
The frequency characteristic of the digital filter given by Z〉=(Tsuchi Kozo)2 is 〈f〉=(Cucumber 5'2) and is approximately as shown in Fig. 1d. Therefore, this filter can be used. For example, the aliasing spectrum, that is, aliasing distortion, can be removed from a signal with the spectral distribution shown in Figure 1c, and a composite color television signal with a sampling frequency of as can be obtained. In order to prevent the envelope of the spectrum of the color signal and the envelope of the folded spectrum of the carrier color signal, which is folded by subnyquist sampling and is generated around fs-$c, to overlap, fs-fsc should be compared to fsc. The sampling frequency $
You need to choose.

■S to remove the aliased spectrum
The value of m cannot be made unnecessarily large; the value of m cannot be made unnecessarily large; It is within the range. As is clear from the above explanation, it is generally easy to use a digital filter to remove aliasing distortion by using a filter that includes the sampling frequency as a function of $ = Ikui SC (similar to synthetic fibers). It can be configured as follows.

A digital filter with cos (Miyoshi's poisoning) frequency characteristics has a transfer function of 10 Z-
≦2 books 1) (5, Can be structured in the form of .

In the explanation example, the case where m=2, n=2, that is, $=5'4fsc was explained, but in other cases, fs=3/2$c,
In the case where fs=7/6$c, fs=9' is sc...etc., a digital filter with a simple configuration, for example, transfer function day (0 is day (Z) = (10 Z-(2 books) 1))2
The digital filter is given by (2), where m' is given by m=1, 3, 4, and so on.

) can remove aliasing distortion. In addition, the sampling frequency fs does not need to be selected exactly to the value of fSC, but can be selected close to that value, and even if there is a deviation of about the horizontal scanning frequency fH, it will have almost no effect on the frequency characteristics of the filter. .

Therefore, the conditions for the sampling frequency fs in order to be able to construct a filter that removes the aliasing distortion expressed by the equations '1} and (2) using a simple digital filter can be restated as follows. $ neck sc
m $〒2h or 10lfSC (2') However, m, n
is a positive integer, and m is within a number of 10 at most.

In the example shown above, the folded spectrum is not completely removed (for example, the spectrum of the Y signal folded near fsc), so the way to remove them is to remove the Y signal before sampling at the sampling frequency fs. It is conceivable to limit the coverage of signals and C signals. For example, a composite color television signal sampled at sampling frequency s is band-limited using a digital filter with the same characteristics as the transfer function used on the receiving side, and then the sampling frequency is fs.
to re-target the signal. In this way, a composite color television signal with less deterioration can be decoded. Furthermore, as a method to reduce image quality deterioration, focusing on the fact that the frequency spectra of the Y signal and C signal of a composite color television signal are interleaved, band limiting and aliasing distortion can be applied to the Y signal and C signal separately. Try to remove it. In this case, a comb-shaped filter is required to separate the composite color television signal into Y signal and C signal, but in order to limit band width and remove aliasing distortion, it is necessary to use a comb-shaped filter suitable for each signal distribution. It is possible to obtain a composite color television signal with almost no deterioration in image quality. As an example of using a comb-shaped filter, when the composite color television signal is in the NTSC format, the folding of the Y signal is made to overlap the spectrum of the Y signal (sampling frequency fs is set to an integral multiple of fH). cases) and cases in which the folding of the Y signal is made to fall between the spectrum of the Y signal and the spectrum of the C signal (in cases where sampling is performed with the sampling frequency fs set to an odd multiple of 1/4 fH), etc. . When the composite color television signal is in the PAL system (in the PAL system, the spectrum of the C signal is concentrated at frequencies that are odd multiples of 1'4fH), the folding of the Y signal is made to overlap the spectrum of the Y signal ( When $ is an integer multiple of fH), and when the folded spectrum of the Y signal is made to overlap at a frequency that is an odd multiple of fH (when $ is a frequency that is an odd multiple of fH), etc. Next, as an example of the above-mentioned method as a second embodiment, the sampling frequency fs is a frequency in the vicinity of 5'4fsc (
(2') when m=2, n=2) and $=
We will explain the case where the value is selected to be fH '71 (note that it is an integer positive number).

When fs is near 5/4fsc, the value is
There are 284, 284±1, etc. When an NTSC color television signal with an entangled distribution of frequency spectra as shown in Figure 1a is sampled at a sampling frequency of $ that satisfies the '71 formula, the folded spectrum is obtained by sampling as shown in Figure 1b. occurs. At this time, the folded spectrum of the C signal falls between the spectra of the Y signal, and the folded spectrum of the Y signal falls between the spectra of the C signal. Therefore, if a comb filter is used, the aliasing distortion of the C signal and the Y
It is possible to separate the aliasing distortion of the Y signal and the C signal. That is, if we look closely at the frequency spectrum distribution, the relationship between the aliasing distortion of the Y signal and the C signal is as shown in FIG. 2a. For the Y signal, which has a frequency spectrum concentrated near frequencies that are integral multiples of fH, as shown by the solid line, the folded spectrum of the C signal is concentrated near frequencies that are odd multiples of 1/2fH, as shown by the broken line. and then turned back. From this, in order to remove the folded spectrum of the C signal, as shown in Figure 2b, at frequencies that are integral multiples of fH, its amplitude, that is, its absolute value, becomes approximately 1, and at frequencies that are odd multiples of 1'2fH, its amplitude, or its absolute value, becomes approximately 1. Its size is almost 0
It is sufficient to pass through a filter having a frequency characteristic such that . An example of a 'filter having the characteristics shown in FIG. 2b is an IH type comb-shaped filter that passes the Y signal. This filter value can be expressed using Z-transform as ZI=e-j2/2rS.The transfer function is given by Z = 2 (where 1 day indicates the sample point one horizontal scanning time ago). The frequency characteristic (f) of this is calculated as (f) = l cosmf
/fHI, and its magnitude is 1 for integral multiples of fH and 0 for odd multiples of 1/2fH.

In other words, by using such a filter, it is possible to eliminate the aliasing distortion of the C signal, and it is possible to eliminate the aliasing distortion of the C signal.In other words, it is possible to eliminate the aliasing distortion of the C signal. Y component signal) can be obtained. In addition, the folded spectrum of the Y signal with respect to the C signal passes through a signal mainly consisting of the frequency components of the C signal, that is, the C component signal (in the case of the NTSC system, a signal whose frequency spectrum is only in the vicinity of odd multiples of 1/2 fH). A comb-shaped filter (for example, the transfer function is
It can be removed by using the IH type comb-shaped filter provided by Koji Kami. Therefore, if a signal sampled at $ is transmitted, and the received signal is fs sampled and has a frequency spectrum distribution as shown in Fig. 2 on the receiving side, (fs
- Heron c) 1/2 of the folded spectrum of the C signal and the folded spectrum of the Y signal occurring near the frequency
By removing the frequency spectrum of the Y signal that has been folded back into the frequency range of s or more, it is possible to decode a signal having almost the same frequency spectrum distribution as the input signal, that is, a composite color television signal. Although the frequency spectrum of the Y signal of the input signal is explained here as up to the frequency of 1/2 $, if the amplitude of the spectrum of the Y signal where 1' is greater than or equal to s is large and its influence cannot be ignored, the sampling frequency common A band limit is applied in advance to the signal sampled at s (for example, by using a comb filter and a bandpass filter with a passing frequency characteristic of 1/as to 3' up to s, (The frequency components of the Y signal up to s are blocked and the other frequency components are passed.) After that, subsampling is performed to convert the signal to a signal with a sampling frequency of $, and send it to the transmission line. On the receiving side, if the value of 0 is interpolated between each sent signal and resampled to a PCM signal with a common sampling frequency, the spectral distribution of the resampled signal will be as shown in Figure 1C. Become. From this, for the Y signal with a spectral distribution concentrated near frequencies that are integral multiples of fH and the aliasing distortion of the Y signal, that is, the Y component signal, after extracting the Y component signal by filtering out the comb-shaped filter, 0 to 1' By passing the signal through a low-pass filter having a pass frequency band of up to 2 degrees, it becomes possible to remove the folded spectrum of the Y signal having a frequency of 1' equal to or higher than s. For the C signal and folded component of the C signal, that is, the C component signal, which has a spectral distribution concentrated near frequencies that are odd multiples of the reference frequency, after extracting the C component signal using a comb filter, It becomes a very small value of 1 near the frequency, and it becomes almost 1 near the frequency of
It is possible to remove the folded spectrum of the C signal generated near the frequency of ($-fsc) by passing through a filter having frequency characteristics such that the value is equal to , and other frequency bands are passed through as much as possible. 0~
For example, as a low-pass filter with a pass frequency band of up to 1 to 28, the sampling frequency is phantom s, and the transfer function (
Z) is day 5 (Z) = ni - 2 - (2 sango) 2 (upper 3)]
(Formula 2) There is a digital filter given by 2.

For example, as a filter for removing the folded spectrum of the C signal so that the sampling frequency $ satisfies the equation (2'), the transfer function Z (Z) shown in the first embodiment is
It can be easily configured by using a digital filter of (Yo3)2 (where ZI=e-j2〆/2fS). By using these filters to perform filtering appropriate to each signal distribution to remove aliasing distortion, it is possible to decode a color television signal with less deterioration in image quality.
Note that $ may be selected so that the folded spectrum of the C signal falls approximately in the middle of the spectrum of the Y signal, and therefore it is sufficient that it approximately satisfies the formula '7}.

As is clear from the above explanation, the principle of the present invention is to select the sampling frequency fs to a value smaller than twice the subcarrier frequency c, which is below the Nyquist frequency, and encode and decode the Enoki color television signal. In this method, by selecting the sampling frequency $ to be at or near Mitsubishi's -fSC (m, n are positive integers), it is possible to easily construct a filter that removes the folded spectrum caused by the sampling of $ on the receiving side. This is the way to do it.

By using such a method, it is possible to directly sample and transmit a band-limited composite color television signal at a frequency below the Nyquist frequency without demodulating it to a base band signal with a simple device configuration. Signal system is NTSC
In the case of a PAL system, for example, the PAL system can be processed in the same way as the NTSC, if it is taken into account that the C signal is concentrated near the fabric of the smooth steering system.

In other words, in the case of the first embodiment, the NTSC system is also PAL.
The method is also the same. In the case of the second embodiment, since the spectral distribution of the C signal is different, it is necessary to change the configuration of the comb filter that separates the Y component signal and the C component signal.

The configuration of an embodiment in the case where signal formation is based on the NTSC system will be described below.

FIG. 3 is a block diagram showing the configuration of the first and second embodiments of the narrowband color television signal encoding/decoding apparatus of the present invention.

In this embodiment, the sampling frequency $ is the subcarrier frequency f
A case where a frequency approximately 5/4 times sc (approximately 4. mz) is selected will be explained. In equation (2'), m=2, n=
This is case 2. The oscillation circuit 4 is a circuit that generates a clock with a frequency of s, and in the case of the first embodiment, it is free oscillation, and in the case of the second embodiment, it is a circuit that generates a clock with a frequency of s. It is necessary to control the oscillation frequency as to =284fH, that is, 2s = 58 yen days. Obtaining an oscillation frequency synchronized with the scanning frequency of a television signal can be easily achieved using known techniques. A clock with a frequency of s generated by the oscillation circuit 4 is sent to the A/D converter 1 and the band limiting circuit 2.
At the same time, it is supplied to the sub-sampling circuit 3 where the frequency of the common s is divided by 1/2 to the frequency of $ by the frequency dividing circuit 11. Further, the clock generation circuit 8 is a circuit that generates a clock having a frequency as in synchronization with the clock frequency of the transmission line, and can be easily realized using already known technology. The generated aS clock is supplied to each section. The analog NTSC color television signal supplied to the A/○ converter 1 is converted by the A/D converter 1 into a PCM (pulse code modulation) signal with a sampling frequency of s, and is supplied to the bandwidth limiting circuit 2. The bandwidth limiting circuit 2 limits the frequency band of signal components that are unnecessary for signal reproduction, and supplies the PCM signal with the limited sampling frequency as to the sub-sampling circuit 3. The sub-sampling circuit 3 performs resampling every other sampling value from the signal with the sampling frequency 2s, and the composite color television signal is converted into a signal with the sampling frequency fs and is supplied to the transmission path. On the receiving side, a PCM signal with a sampling frequency fs received from a transmission line is supplied to an interpolation circuit 5.

The interpolation circuit interpolates sample values with a value of 0 between each sample value from the PCM signal of sampling frequency fs, so that the sampling frequency is s.
is the PCM signal of s, is doubled so that the average amplitudes are equal when $ and s are the same frequency, and then supplied to the removal filter circuit 6. The removal filter circuit 6 uses a filter to remove aliasing distortion unnecessary for signal reproduction, thereby obtaining a PCM signal of a band-limited composite color television signal with a sampling frequency of 2 seconds. The PCM signal with a sampling frequency of s obtained by the removal filter circuit 6 is converted into an analog composite color television signal by the D/A converter 7. FIG. 4 shows an example of a specific configuration of the band limiter circuit 2 and removal filter circuit 6 of FIG. 3 in the case of the first embodiment.

In this embodiment, the Obishiro limiting circuit 2 and the removal filter 6 both have a transfer function (Z>=(Mikozo)).
It consists of a digital filter with the characteristics given by 2. (However, ZI=e-j2mf/trillion) 401~41
The shift register operates at a clock frequency of 2 s, and each input signal is outputted to the output of each shift register with a delay of one clock period.

411 and 412 are adders, and 413 is a multiplier with a magnification of 1/4.

The multiplier 413 can be easily configured by shifting each input signal downward by two digits and outputting the shifted signal. The operation of each component is a well-known configuration of a digital filter and can be clearly understood from the diagram, so a description thereof will be omitted. The output signal of this digital filter is output with a delay of 5 clock cycles with respect to the input signal. FIG. 5 shows an example of a specific configuration of the sub-sample circuit 3 and interpolation circuit 5 shown in FIG. 3.

The sub-sample circuit 3 is composed of a switch 51. The switch 51 is a switch that opens and closes according to the clock frequency, and by opening and closing the switch, the PCM signal of the sampling frequency 2s supplied to the sub-sampling circuit 2 is passed every other sampling value, thereby changing the sampling frequency fs. PCM
A signal is output to the output of switch 51. That is, the sub-sampling circuit 2 outputs a PCM signal of sampling frequency &. The interpolation circuit 5 includes a switch 52 and a multiplier 53
It consists of The PCM signal of sampling frequency fs supplied to the interpolator 5 is sent to one input of a switch 52 whose clock frequency changes at s. 0 for the other input
A level generator 54 supplies a PCM signal with a value of 0.
By switching the switch 52 at a frequency of 2s in synchronization with the input signal, a PCM signal with a sampling frequency of s, with a value of 0 interpolated, is obtained at the output of the switch 52. This signal is sent to the multiplier 53 and outputted so that the average value of the signal is equal when the sampling frequency is fs and when the sampling frequency is s. A multiplier with a magnification of 2 can be easily constructed by shifting the input signal one digit higher and outputting it. Figure 6 shows the second
A specific example of the configuration of the band limiting circuit 2 shown in FIG. 3 in the case of the embodiment shown in FIG. The PCM signal with a sampling frequency of s is supplied to a delay circuit 61 and a signal separation circuit 62 that compensate for delays in the signal separation circuit 62 and filter circuit 63, respectively.

The signal separation circuit 62 separates the Y component signal using a comb filter, and the obtained Y component signal is sent to the filter circuit 6.
3, and frequency components from 1/2$ to 3'2$ are passed. The high frequency component of the Y signal obtained at the output of the filter circuit 63 is sent to a subtracter 64, where it passes through a delay circuit 61 and the phase compensated sampling frequency is subtracted from the PCM input signal of s. As a result, the output of the subtracter 64 is Y
A signal whose sampling frequency is s is obtained by band-limiting the Takagi frequency component of the signal. Figure 7 shows the signal separation circuit 6 in Figure 6.
This is an example of a specific configuration of No. 2.

In this embodiment, a case is shown in which a so-called IH type comb filter is used to separate the Y component signal using the correlation between the current sample value and the sample value one scanning line before. A/
○ PCM with sampling frequency s sent from converter 1
The signals are supplied to line memory 71 and adder 72, respectively. The line memory 71 is a circuit that provides a delay of one horizontal scanning period, and can be composed of a storage element such as a shift register. The input signal and the signal obtained by delaying the input signal by one horizontal scanning period are added by an adder 72 and then multiplied by 1/2 by a multiplier 73 to output a Y component signal to the output of the multiplier 73. This output signal has no delay with respect to the sampling period with respect to the input signal. The multiplier 73 is simply configured by shifting each bit of the input signal one digit downward and outputting the resultant signal. FIG. 8 shows an example of a specific configuration of the filter circuit 63 shown in FIG.

The filter characteristics shown in this example are Z-transformed (ZI=e-j2
mf'2fS), it is given by day (Z) = - [Z-2- (Ushizo) 2 (bank and 2)] (Kamita thousand) 2 legs.

This frequency characteristic is calculated as follows:
)2・COS2mf/ga s] (sinmf/phantom s)2
(9}.

As is clear from equation '9}, this filter value is approximately 1
/ is a filter having a pass characteristic that passes signal components with frequencies from s to 3/Xs. In this embodiment, the transfer function (Z) given by the equation (2) is realized by a non-recursive digital filter.

Reference numerals 801 to 808 are shift registers operating at a clock frequency of 2 s, and each input signal is output to the output of each shift register with a delay of one clock period.

809, 810 and 811 are adders, 812, 813
and 814 is a subtracter, 815 is a multiplier with a magnification of 1'8,
816 is a multiplier with a magnification of 1/4.

The multiplier 815 can be easily constructed by shifting the input signal three digits lower and outputting the same, and the multiplier 816 can be easily configured by shifting the input signal two digits lower and outputting the same. The operation of each component is a well-known configuration of a digital filter and can be clearly understood from the diagram, so a description thereof will be omitted. In the case of using the type 2 digital filter shown in FIG. 8, the output signal of the filter circuit 63 is delayed by three clock cycles with respect to the input signal. Therefore, the delay circuit 61
is a shift register that operates at the public S clock frequency.
The output of the delay circuit 61 is delayed by three clock cycles depending on the input, and the delay is compensated for. FIG. 9 shows a specific configuration example of the removal filter circuit 6 of FIG. 3 in the case of the second embodiment. The PCM signal with the sampling frequency as supplied from the interpolation circuit 5 is sent to the signal separation circuit 92, and is separated into a Y component signal and a C component signal containing aliasing distortion using a comb-shaped filter that utilizes line correlation. . The separated Y component signal is sent to the Y filter circuit 93, which passes only the frequency components with a frequency of 0 to 1/2 seconds, and removes the folded spectrum of the Y signal of 1/2 $ or more caused by folding. A Y signal with a sampling frequency of 2s is obtained, which has almost the same frequency spectrum as the band-limited original Y signal. Further, the separated C component signal is sent to a C filter circuit 94, where the folded spectrum of the C signal near the frequency of fs-fsc, which is generated by folding, is removed, and the folded C signal is made up of almost the same frequency spectrum as the original C signal. A C signal with a sampling frequency of 2s is obtained. The Y signal and C signal, which have been compensated for the difference in delay time due to filtering, are supplied to an adder 95 and added to obtain a composite color television signal of sampling frequency as. 10th
The figure shows an example of a specific configuration of the signal separation circuit 92 shown in FIG.

In this embodiment, a case is shown in which an IH type comb-shaped filter is used. A PCM signal with a sampling frequency of s supplied from the inter-tub circuit 5 is supplied to a line memory 101, an adder 102, and a subtracter 103, respectively. The line memory 101 is a circuit that provides a delay of one horizontal scanning period, similar to the one shown in FIG. 7, and can be constituted by a storage element such as a shift register. The input signal and the signal delayed by one horizontal scanning period are added by an adder 102, and a multiplier 10
The signal is multiplied by 1'2 by 4 to output a Y component signal. Further, the subtracter 103 subtracts the input signal delayed by one horizontal scanning period from the input signal, and the multiplier 105 subtracts the input signal delayed by one horizontal scanning period.
By multiplying by 1'2, a C component signal is output. FIG. 11 shows an example of a specific configuration of the Y filter circuit 93 shown in FIG. 9.

The filter characteristics shown in this example are Z-transformed (ZI=e-j2
〆'2fS) is H.
It is given by

In the working equation, the final variable Z-2 is added to compensate for the delay with the output of the C filter circuit 94. At this time, the frequency characteristic day (f) is day (f) 2
・COS f/2s] (cos f/2s) 2 (
11).

As is clear from equation (11), the Y filter circuit 93 serves as a filter that passes low frequency components. In this embodiment, the function (Z) given by equation (10') is realized by a non-recursive digital filter.1
Reference numerals 101 to 1110 are shift registers operating at a clock frequency of 2 s, and each input signal is output to the output of each shift register with a delay of one clock period. 1111-11
13 is a subtracter, 1114 to 1116 are adders, 1117
is a multiplier with a magnification of 1'8, and 1118 is a multiplier with a magnification of 1'4.

Each multiplier 1117, 1118 can be easily configured by shifting each input signal downward by three or two digits and outputting the resultant signal. The operation of each component is a well-known configuration of a digital filter and can be clearly understood from the drawings, so a description thereof will be omitted. The output signal of this digital filter is output with a delay of 5 clock cycles with respect to the input signal.
As an example of a specific configuration of the C filter circuit 94 shown in FIG. 9, a filter having the same function as the removal filter circuit shown in FIG. 4 is used.

Although the first embodiment and the second embodiment have been described above, the belt-limiting circuit shown in FIG. There is also a method of adding band limitation to the C signal as well.

Although the signal separation circuits of the embodiments shown in FIGS. 7 and 10 have a so-called IH type configuration, they are not limited to this configuration method, and other methods may be used, such as a so-called spot-on pattern using front and rear scanning lines. It is also possible to configure it with a type.

The embodiments of the C filter circuit 94 shown in FIGS. 4, 8, 11, and 9 are not limited to these embodiments, and each embodiment is configured to satisfy the required frequency characteristics. Any filter circuit may be used.

The embodiment of the interpolation circuit shown in FIG. 5 and the removal filter circuit shown in FIG. 52 can be arranged between the signal separation circuit 92 and the Y filter circuit 93 and between the signal separation circuit 92 and the C filter circuit 94, respectively. In this way, the number of delay elements required for the signal separation circuit can be reduced. Furthermore, in the first and second embodiments, the sampling frequency fs is 5'4$c (m=2, n
= 2), but in the case of other values, the frequency characteristics of each filter circuit in the band limiting circuit 2 and the removal filter circuit 6 may be changed to the desired characteristics. FIG. 12 is a block diagram showing the configuration of a third embodiment of the present invention.

This embodiment is a combination of the first or second embodiment and the band compression encoding method, and includes an A/D conversion 1, a band limiting circuit 2, a sub-sampling circuit 3, an oscillation circuit 4, and an interpolation circuit 5. , removal filter circuit 6, D/A converter 7, clock generation circuit 8, and frequency dividing circuits 11 and 12 are the same as the A/D converter 1 shown in FIG.
, Obijo limit circuit 2, sub-sample circuit 3, oscillation circuit 4,
It has the same functions as the interpolation circuit 5, removal filter circuit 6, D/A converter 7, clock generation circuit 8, and frequency division circuit 11. The Obijo compression encoding circuit 9 converts the PCM signal of sampling frequency fs supplied from the sub-sampling circuit 3 into DP.
This circuit performs band compression encoding such as CM (differential PCM) and Hadamard transform.

The signal encoded by the band compression encoding circuit 9 is sent out to a transmission path. This is a circuit that decodes a compressed signal sent from a transmission line into a PCM signal having a sampling frequency fs. The decoded signal is interpolated by an interpolation circuit 5, an aliasing component is removed by a removal filter circuit 6, and then converted into an analog signal by an O/A converter 7. The oscillation circuit 4 and clock generation circuit 8 supply necessary clocks to each section.

The encoding device of this embodiment is a PC sampled at the sub-Nyquist sampling frequency fs in the first or second embodiment.
By combining a band compression encoding method suitable for the PCM signal to transmit the M color television signal, a larger compression of the number of transmission bits is achieved than in the first or second embodiment. .

FIG. 13 shows an example of a specific configuration of the band compression encoding circuit 9 and the band compression encoding circuit 10 shown in FIG. 12.

In this embodiment, a case will be described in which high-order DPCM using a high-order prediction function adapted to a color television signal including subcarriers is adopted as the band compression encoding method. If the prediction function is P(Z) and the transfer function of the prediction error signal is day(Z), it is expressed as day(Z)=(1-P(Z)), and the higher-order DP
In CM, the transfer function (Z) of the prediction error signal may be such that the RMS value of the prediction error output is small with respect to the Y signal and the carrier color signal in the low frequency part. As such a function, 1-Z-Shoe (Z-2 days indicates the sampling time two horizontal scanning cycles ago) is used, which uses the value of the sampling time two lines ago, and as a correction function, (1-QZ -1). (Here, if fs=evening/fH as shown in formula '7}, day Z-2 is expressed as Z-2H=Z-2 sleeves.)
That is, day (Z) = (11QZ-1) (11Z-2H)
The prediction function P(Z) = 1-day(Z
) as an example. From this, the prediction function P(Z) is P
(Z)=QZ-10Z-2 days 1QZ-1. Z-2 days (1
4).

However, Q is a constant, for example, Q=0.5. The high-order DPCM encoder has a subtracter 141 and a quantizer 142.
, an adder 143, a prediction filter 144, and a code converter 145.

The input signal to the higher order DPCM encoder is supplied to a subtractor 141.

The subtracter 141 extracts the prediction filter 14 from the input signal.
The prediction signal output from step 4 is subtracted and a prediction error signal is output. The prediction error signal, which is the output of the subtracter 141, is sent to the quantity predictor 142, where it is converted into a star using a predetermined quantization characteristic, and the star-ized signal is sent to the adder 123.
and is supplied to the code converter 145. In adder 143 , a locally decoded signal is obtained by adding the output of quantizer 142 and the output signal of prediction filter 144 , and is supplied to the input of prediction filter 144 . Prediction filter 144
has the transfer characteristic shown in equation (14), calculates the predicted value of the next sampling time from the locally decoded signal, and outputs it as a predicted signal. The code converter 145 quantizes the signals quantized by the quantizer 142, converts them into codes corresponding to the signals, and sends the signals to the transmission path. In the high-order DPCM decoder, the code sent from the transmission path is quantized by the quantizer 142 in the code inversion circuit 146, converted into a signal level corresponding to the bell, and sent to the adder 147.

The adder 147 adds the signal sent from the sign inverse conversion circuit 146 and the prediction signal sent from the prediction filter 148 to obtain a decoded signal. The decoded signal is predicted filter 1
48, which calculates and outputs a predicted signal for the next sampling time. Note that the predicted filter 148 has the same transfer characteristics as the predicted filter 144. FIG. 14 shows an example of a specific configuration of the prediction filters 144 and 148 in FIG. 13.

In this example, the function P(Z
) consists of a non-recursive digital filter. 151 is a shift register that operates at a clock frequency of $; 152 is a multiplier with a multiplier of Q;

When the magnification Q is selected to be Q=0.51, the multiplier 152 can be simply configured by simply shifting the input signal one digit lower and outputting it. 153 is a subtracter, I55 is an adder, and 154 is a line memory. The line memory 154 is a circuit that provides a delay of two horizontal scanning periods, and can be constructed from a storage element such as a shift register. The operation of each component is a well-known component of a digital filter and can be clearly understood from the diagram, so a description thereof will be omitted. As is clear from the above explanation, by coarsely adapting the band compression encoding method to the encoding/decoding device capable of encoding at a sampling frequency lower than the Nyquist frequency described in the first or second embodiment, the bandwidth can be significantly increased. An encoding/decoding device capable of compression encoding can be realized.

Moreover, since the composite color television signal is not demodulated into a base band signal and all post-encoding processing can be performed using digital logic circuits, highly accurate calculations can be performed.

Furthermore, the transfer characteristics of the prediction filters 144 and 148 in FIG. 13 are not limited to the prediction function P(Z) shown by equation (14), and can efficiently predict both the luminance signal component and the carrier color signal. Any function is fine.

Furthermore, the band reduction longitudinal encoding method in the third embodiment is also a high-order D
The present invention is not limited to PCM, and may be an orthogonal transform coding method such as Hadamard transform or an interframe coding method.

Alternatively, a method combining these with variable length encoding may be used.
In the first, second, and third embodiments, if no band limit is applied, the sampling frequency is s, and there is no need for A/D conversion, and the sampling frequency can be $.

Therefore, in this case, the band limiter circuit 2 and the sub-sample circuit 3 are unnecessary, and the device configuration is simplified. In each embodiment, signal lines other than signal lines indicating the flow of information of television signals, clock signal lines, etc. are omitted except for important parts.

[Brief explanation of the drawing]

FIG. 1a is a diagram showing the envelope of the frequency spectral distribution of the Y signal and C signal of the input decoded color television signal. Figure 1b shows the frequency spectrum distribution of an input signal sampled at a sampling frequency of $, and Figure 1c shows a signal with a sampling frequency of FIG. 1d is a diagram showing the frequency characteristics of a filter for removing aliasing distortion of the C signal. Figure 2 a is Y
FIG. 2B is a diagram showing the relationship between the frequency spectra of the signal and the reflected C signal, and FIG. 2B is a diagram showing the frequency characteristics of the comb-shaped filter that passes the Y component signal. FIG. 3 shows the first and second coding and decoding devices for narrowband color television signals according to the present invention.
FIG. 4 is a block diagram showing an example of a specific configuration of the band limiter circuit 2 or removal filter circuit 6 in the first embodiment, and FIG. 3 and an interpolation circuit 5, and FIG. FIG. 7 is a block diagram showing an example of a specific configuration of the signal separation circuit 62, FIG. 8 is a block diagram showing an example of a specific configuration of the filter circuit 63, and FIG. 9 is a block diagram showing an example of a specific configuration of the filter circuit 63. FIG. 10 is a block diagram showing an example of a specific configuration of the removal filter circuit 6 in the case of FIG.
FIG. 1 is a block diagram showing an example of a specific configuration of the Y filter circuit 93. FIG. 12 is a block diagram showing the configuration of the third embodiment of the present invention, FIG. 13 is a block diagram showing the configuration of the high-order DPCM encoder 9 and the decoder 10, and FIG.
The figure is a block diagram showing an example of a specific configuration of the prediction filters 144 and 148. In the figure, 1 is an A/D converter, 2 is a band limiting circuit, 3 is a supsampling circuit, 4 is an oscillation circuit, 5 is an interpolation circuit, 6 is a removal filter circuit, and 7 is a D
/A converter, 8 is a clock generation circuit, 9 is a band compression encoding circuit, 1 is a compression encoding circuit, 11 and 12 are frequency dividing circuits, 401 is a shift register, 411 is an adder,
413 is a multiplier, 51 is a switch, 54 is a 0 level generator, 61 is a delay circuit, 62 is a signal separation circuit, 63 is a filter circuit, 64 is a subtracter, 71 is a line memory, 92 is a signal separation circuit, 93 is a Y filter circuit, 94 C filter circuit, 101 line memory, 142 quantizer, 1
44 and 148 are prediction filters, 145 is a code converter, 146 is a code inverse converter, 151 is a shift register, 1
54 each indicates a line memory. O figure juice 2 figure O 3 figure age 4 figure O 5 figure O 6 figure age 7 figure O 9 figure age 8 figure O ○ figure O l figure O 2 figure O'3 figure O'4 figure

Claims (1)

[Claims] 1. In an encoding/decoding device that encodes and transmits a composite color television signal obtained by frequency multiplexing color signals, the sampling frequency fs is a value lower than 2fsc, where the subcarrier frequency is fsc, and (2m+1)/(2n)・fsc (m, n
is a positive integer and m is within several 10 at most), or a value close to it.
os((2m+1)/(2n)πf/(fs)), and a decoding device equipped with means for removing aliasing distortion using a filter containing a function of An encoding/decoding device for a narrowband color television signal, characterized in that the encoding/decoding device performs decoding with a narrowband color television signal.