JPH0352718B2 - - Google Patents

Description

[Detailed description of the invention] [Industrial application field]

The present invention relates to a predictive decoding device used for data compression encoding of PAL color television signals.

[Conventional technology]

Conventionally, various methods for data compression encoding (bandwidth compression encoding) of color television signals have been studied. Compared to a separation method that restores the carrier color signal to a baseband signal and encodes the luminance signal and color signal separately, the method that directly predictively encodes a composite color television signal requires processing of separation, demodulation, and modulation of the carrier color signal. Since there is no code, the hardware is simple and high-quality encoding can be performed. However, since color signals are frequency multiplexed, efficient encoding cannot be performed using the prediction function used for monochrome signals. On the other hand, as a method for efficiently direct predictive coding of composite color television signals, there is a method proposed by IEEE Transactions on communications.
Vol.COM−22 No.8 August 1974, 1106~
Page 1113 Vol.COM−25No.11 November 1977
1349-1385, the function of sampling frequency f _S and subcarrier frequency f _SC is expressed as f _S =l/
It is known that efficient predictive coding can be performed by making predictions using sampling points where the phases of subcarriers are in the same phase, where m·f _SC (l and m are positive integers). On the other hand, in the case of an NTSC television signal, the number of samples per line is 1, since the relationship between the subcarrier f _sc and the horizontal synchronization frequency f _H is f _SC = 455/2f _H , so if l = 4 and m = 1, the number of samples per line is 1. The number of line samples is an integer, 910. However, in the case of a PAL television signal, f _SC = (1135/4 + 1/625) f _H , so the number of samples per several lines could not be easily set to an integer value. For this reason,
When encoding a PAL television signal in units synchronized with a horizontal synchronization signal, special processing must be performed on irregular samples, which has the disadvantage of complicating the hardware. Also, f _SC
Since the relationship _{between f} Sampling frequency oscillator (VCO), if any
The disadvantage is that it becomes complicated. Also, when connecting to other encoding devices in the form of digital television signals, it is convenient for the number of samples per several lines to be an integer value.

[Purpose of the invention]

The purpose of the present invention is to perform PAL sampling at a sampling frequency such that the number of samples per several lines is an integer.
It is an object of the present invention to provide a decoding device capable of predictively decoding a signal obtained by efficiently predictively encoding a television signal.

[Structure of the invention]

According to the invention, approximately 3 of the subcarrier frequency
and N/2 times the horizontal sync frequency (N is an integer)
A device that receives ^and decodes a signal obtained by predictively encoding a PAL television signal sampled at a sampling frequency ^of /3Z} and a means for performing predictive decoding using a predictor having the same prediction characteristics as used in predictive encoding.

[Principle and action]

By performing sampling by setting the sampling frequency f _S to approximately three times the subcarrier frequency f _SC and N/2 times the horizontal synchronization frequency f _H , it is possible to , among the sample points two lines before the sample points are in phase, efficient prediction can be performed for PAL television signals with strong vertical correlation using three consecutive samples including the sample point directly above. Ru. The arrangement of sample points at this time is shown in Figure 2, f _S = (N/2) f _H
(N is a positive integer), the arrangement of sample points on the (l-2)th line, which is two lines before the lth line, is vertically in phase. Figure 2 is N
When is an even number, the (l-1)th line is also vertically in phase. For images with strong vertical correlation,
For a sample point whose sample value is X of a PAL television signal, prediction is better achieved by using sample points whose sample values are S ₁ , S ₂ , and S ₃ above than by using sample points in front of it on the same line. Since f _S ≒ 3f _SC , the sum of the components of the carrier color signal of the three samples becomes almost zero in the sample values S ₁ , S ₂ , S ₃ of three consecutive samples, and there is no large change in the luminance signal. Assuming that the sum of sample values S ₁ , S ₂ , and S ₃ is approximately the luminance signal Y
Only. S ₁ + S ₂ + S ₃ ≒ 3Y ... (1) If the signal at the current sample point ₂ is expressed as follows using a luminance signal Y and a carrier color signal C.
The error subcarrier of a PAL television signal has a frequency relationship of f _SC = (1135/4 + 1/625) f _H , that is, one period T _H of the horizontal synchronization signal and a period T _SC of the color subcarrier, T _H = (2833 /4+1/625) There is a T _SC relationship. The subcarrier phase shift two lines before is (3/4+1/625)×2. The phase shift of the subcarrier two lines before is (3/4 + 1/625) × 2, but 2/625T _SC is negligible, and it is 1/ (2T _SC ), that is, the subcarrier has a phase shift of π. . Therefore, the sample value X (X=Y+
The signal S ₂ two lines before the signal C) can be expressed as S ₂ ≒Y+C·e ^j =Y−C (2). Therefore, X using S ₁ , S ₂ , S ₃
The predicted value X^ is as follows from equations ( ₁ ) and _{( 2} ): From this, even if f _s is set so that the number of samples per two lines is an integer value N, an image with strong vertical correlation can be efficiently predicted using three sample points two lines before. Generally, in order to correctly decode a predictively encoded signal, the prediction function (expressing the prediction characteristics in the form of a function) of the predictor used for predictive decoding must be The same prediction function is used.

【Example】

Embodiments of the present invention will be described below with reference to the drawings. FIG. 6 is a diagram showing the configuration of a conventional predictive encoding device and predictive decoding device for PAL television signals. The sampling frequency is generally f _S = (l/m) f _SC
However, in this example, l=3 and m=1. That is, f _S =3f _SC , and the arrangement of sample points at this time is shown in FIG. The locations indicated by marks are locations where the subcarriers are in phase, but since the V signal is inverted for each line, the phase of the carrier color signal differs between even and odd lines. The predictor 66 of the predictive encoding device and the predictor 73 of the predictive decoding device perform two-dimensional prediction using sample points including two lines before the phase matching.
Furthermore, if one of the sampling points is fixed on the U axis, prediction using the previous line becomes possible. Further, by correcting the color signal by performing phase conversion and filtering processing on the color signal, it becomes possible to perform prediction using the previous line. However, with these methods, it is not easy to make efficient predictions by setting the number of samples per several lines to an integer number. FIG. 1 is a block diagram showing an embodiment of the present invention. applied to input terminal 1 of the predictive encoding device
The PAL television signal is sent to an A/D converter 2 and a pulse signal generator 3. The pulse signal generator 3 uses the horizontal synchronizing signal of frequency f _H obtained by synchronous separation to generate a signal with a value close to three times the subcarrier frequency f _SC and f _S = N/2f _H ...(4) ( N is an integer), and _A /
Supplied to D converter 2. Note that N=1702 is taken as a specific example of the value of N. At A/D converter 2, f _S =
The PAL television signal converted into a digital signal at a sampling frequency of N/ _2fH is supplied to a subtracter 4.
A prediction signal is output from the predictor 6 and supplied to the subtracter 4 and adder 7. The prediction error signal output from the subtracter 4 is supplied to the quantizer 5, where it is quantized according to predetermined quantization characteristics, and outputs a quantized prediction error signal, in other words, a quantized output signal. Adder 7 and code conversion circuit 8
supply to In the adder 7, the predicted signal and the quantized output signal are added, and a (locally) decoded signal is obtained as an output. The (local) decoded signal is supplied to a predictor 6, where a predicted signal is determined according to the prediction characteristics and output as a predicted value for the next sample time. The code conversion circuit 8 converts the quantized prediction error signal into a transmission line code signal according to predetermined conversion characteristics, outputs the signal, and supplies the signal to the output terminal 9 of the predictive coding device. On the receiving side, the transmission line code signal input from the input terminal 10 of the predictive decoding device is sent to the code inverse conversion circuit 1.
Inverse code conversion is performed in step 1, and a prediction error signal (quantized output signal) quantized according to the transmitted code is reproduced and output, and is supplied to the adder 12. The adder 12 adds the reproduced quantized output signal and the prediction signal from the predictor 13 to obtain a decoded signal, outputs it, and supplies it to the D/A converter 14 and the predictor 13 . In order to correctly predictively decode a predictively coded signal, the predictor 13 of the predictive decoding device has the same prediction characteristics as the predictor 6 of the predictive coding device, and the next sample is determined according to the prediction characteristics. Determine and output the time prediction signal. The D/A converter 14 converts the decoded digital PAL television signal into an analog signal, outputs it, and supplies it to the output terminal 15 of the predictive decoding device. FIG. 3 is a diagram showing a first specific example of the predictors 6 and 13. The predictor in this example calculates a predicted value according to equation (3). Since the sampling frequency f _S is defined as f _S ≒ 3f _SC and f _S = N/2f _H (N is an integer), the Z function (Z ^-1 =
The prediction function P _{2 ( Z} ) expressed _as It is configured as a characteristic digital filter prediction circuit. P ₂ (Z)=Z _-N (2Z ^-1 -1+2Z)/3 ...(5) 21 is a shift register that delays the input signal by (N-1) sampling clock cycles and outputs the delayed signal.
22 and 23 are registers for delaying one sampling period, and the coefficients of multipliers 24, 25 and 26 are 2/3, -1/3 and 2/3, respectively. FIG. 4 is a diagram showing a second specific example of the predictors 6 and 13. When f _S ≈3f _SC , there is a one-dimensional prediction function P ₁ (Z) that allows efficient predictive coding to be expressed by the following equation. P ₁ (Z)=0.5Z ^-1 +Z ^-3 -0.5Z ^-4 (6) The prediction circuit 31 is composed of a digital filter having a prediction function P ₁ (Z) of equation (6). The prediction circuit 32 is composed of a digital filter with a prediction function P ₂ (Z) expressed by equation (5). The coefficients A ₁ and A ₂ of the multipliers 33 and 34 are:
Normally, it is determined that A ₁ +A ₂ ≒1, but for example, A ₁ =0.5 and A ₂ =0.5. The output terminal 36 provides a predicted signal determined by the predictive filter characteristic such that the predictive function P(Z)=A ₁ ·P ₁ (Z)+A ₂ ·P ₂ (Z). FIG. 5 is a diagram showing a third specific example of the predictors 6 and 13. An applied prediction is used in which two prediction characteristics are switched for each sample. The switching method is to determine which prediction function had a closer value to the decoded signal at the previous sample time, and then use the closer prediction function to change the prediction function at the next sample time. demand. Prediction circuit 4
1 and 42 are constructed of digital filters shown by equations (6) and (5), respectively, and output prediction signals P^ ₁ and P^ ₂ , respectively. The comparison circuit 43 determines the magnitude of |X-P^ ₁ | and |X-P^ ₂ | from the input signal X from the input terminal 40 and the predicted signals P^ ₁ and P^ ₂ ,
The determination signal is delayed by one sampling clock period in the register 44 and outputted, and is supplied to the switching circuit 45.
In the switching circuit 45, if |X-P^ ₁ | in the previous sample is small, the prediction signal P^ ₁ for the current sample from the prediction circuit 41 is selected and output, and |X-P^ ₂ | in the previous sample is selected and output. is small, the prediction signal P^ ₂ for the current sample from the prediction circuit 42 is selected and output and supplied to the output terminal 46.

【Effect of the invention】

As is clear from the above explanation, according to the predictive decoding device of the present invention, in order to simplify the encoding/decoding process, f _S =N/ Even with 2f _H , it is possible to receive and predictively decode signals that have been subjected to efficient secondary predictive coding using a simple prediction function, and it is possible to perform high quality even for images with strong correlation in the vertical direction. Can play PAL TV signals.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an embodiment of a predictive coding device and a predictive decoding device of the present invention that decodes signals encoded by the device, and FIG. 2 shows an arrangement of sample points according to the present invention. 3, 4, and 5 are diagrams showing first, second, and third specific examples of the predictors 6 and 13, and FIG. 6 is a conventional predictive encoding device and predictive decoding device. FIG. 7 is a diagram showing an example of a converting device, and FIG. 7 is a diagram showing an arrangement of sample points in a conventional embodiment. 1...Input terminal, 2...A/D converter, 3...
Pulse signal generator, 4... Subtractor, 5... Quantizer, 6... Predictor, 7... Adder, 8... Code conversion circuit, 9... Output terminal, 10... Input terminal, 1
1... Sign inverse conversion circuit, 12... Adder, 13...
...Predictor, 14...D/A converter, 15...Output terminal, 20...Input terminal, 21...Shift register, 22, 23...Register, 24, 25, 26
... Multiplier, 27 ... Adder, 28 ... Output terminal, 30 ... Input terminal, 31, 32 ... Prediction circuit, 33, 34 ... Multiplier, 35 ... Adder, 3
6...Output terminal, 40...Input terminal, 41, 42
...Prediction circuit, 43...Comparison circuit, 44...Register, 45...Switching circuit, 46...Output terminal.

Claims (1)

[Claims] 1. Receiving a signal obtained by predictively encoding a PAL television signal sampled at a sampling frequency that is approximately three times the subcarrier frequency and N/2 times the horizontal synchronization frequency (N is an integer). A decoding device that includes a prediction circuit whose Z-domain prediction function is Z ^-N {2/3Z ^-1 -1/3 + 2/3Z} and that predicts the same prediction characteristics as used in predictive encoding. A predictive decoding device includes means for performing predictive decoding using a device.