JPS628055B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a predictive coding device, and more particularly to a predictive coding device that directly codes a signal obtained by multiplexing a plurality of signals, such as a color television signal.

A color television signal is a signal obtained by frequency multiplexing the luminance signal and chromaticity signal of an image.As a predictive encoding device that efficiently encodes this signal, the luminance signal and chromaticity signal are separated and each signal is processed. There are devices that use a separate coding method that performs predictive coding, and devices that use a direct coding method that directly predictively codes multiplexed signals without separating them. A device using a separate coding method must be equipped with a signal separation circuit, a signal synthesis circuit, and a plurality of predictive coding circuits, resulting in a large scale, whereas a device using a direct coding method requires a signal separation circuit. Also, since a signal synthesis circuit is not required and only one predictive coding circuit is provided, the scale of the apparatus can be reduced.

FIG. 1 is a block diagram showing a conventional predictive coding apparatus using a direct coding method. In the figure, when a color television signal is input from an input terminal 1, the input signal passes through a subtraction circuit 2, and a prediction signal from a prediction circuit 3 is subtracted therefrom to form an error signal, which is then sent to a quantization circuit 4. The error signal is sampled and quantized with a period T in the quantization circuit 4, and is outputted from the output terminal 5 as a differential pulse modulation code (hereinafter abbreviated as DPCM code) and sent to the prediction circuit 3. FIG. 2 is a characteristic diagram illustrating the quantization characteristics of the quantization circuit 4 of FIG. 1. As illustrated in the figure, a non-linear quantization characteristic 30 is generally used as the characteristic of the quantization circuit 4. By using these characteristics, parts of a flat image signal where the error signal is small but quantization noise is easily visible can be finely quantized, and image signals where the error signal is large but quantization noise is difficult to visually see can be finely quantized. It is possible to coarsely quantize the edge portions to achieve quantization that is suitable for visual perception. The signal sent to the prediction circuit 3 in FIG. Transfer function P of predictor 7
(Z) is given by the following formula.

P(Z)=0.5Z ^-1 +bZ ^-2 -0.5bZ ^-3 However, in the above formula, b=1-(1/2) ⁴ , Z ^-1 =
e ^−j 〓 ^T , or T indicates the sampling period of the quantization circuit 4. The output signal of the predictor 7 passes through a digital-to-analog converter 8, becomes a predicted signal converted into an analog signal, and is sent to a subtraction circuit 2, where the error signal, which is the difference from the input signal from the input terminal 1, is obtained.

A predictive coding device using a conventional direct coding method performs predictive coding without separating a luminance signal and a chrominance signal, which have different error signal level generation distributions. Since optimal quantization cannot be performed for each of This has the disadvantage that the speed reduction rate is small. Furthermore, since the quantization characteristics of the quantization circuit are not variable, there is a drawback that the signal-to-noise ratio deteriorates if the generation distribution of the level of the error signal changes significantly over time.

It is an object of the present invention to eliminate the above-mentioned drawbacks and to provide a signal-to-noise ratio and a coding system that adaptively operates as the generation distribution of the levels of the luminance signal component and the chrominance signal component of the error signal changes over time. An object of the present invention is to provide a predictive encoding device that can perform direct encoding with good efficiency.

The apparatus of the present invention includes a subtraction circuit that sends out an error signal representing the difference between an input signal including a plurality of signal components and a predicted signal, and a subtraction circuit that separates the plurality of signal components included in the error signal to produce a plurality of separated signals. and an adaptive compression circuit that detects the level occurrence distribution of at least one of the plurality of separated signals and performs level compression on the at least one separated signal in accordance with the level occurrence distribution. an addition circuit that adds a plurality of signals sent out by the adaptive compression circuit; a quantization circuit that quantizes and outputs the signal sent out by the addition circuit; and a prediction signal predicted from the output signal of the quantization circuit. and a prediction circuit that sends out the prediction.

Next, embodiments of the present invention will be described with reference to the drawings.

FIG. 3 is a block diagram showing a first embodiment of the present invention. In the figure, a color television signal is inputted from an input terminal 1, passes through a subtraction circuit 2, becomes an error signal which is the difference from a prediction signal from a prediction circuit 3, and is sent to a separation circuit 9. The separation circuit 9 is a duplexer composed of a bandpass filter that passes only the chromaticity signal component modulated at the subcarrier frequency, and a bandstop filter that passes the luminance signal component and blocks the chromaticity signal component. The error signal passes through the separation circuit 9 and is separated into a chromaticity signal component and a luminance signal component, resulting in two separated signals, each of which is sent to the adaptive compression circuit 1.
0 and 11 and subjected to level compression adapted to the level generation distribution of each signal component, the chromaticity signal component and the luminance signal component are again added and synthesized in the adder circuit 13. The signal from the adder circuit 13 is sampled and quantized at a period T in the quantization circuit 4 to become a DPCM code, which is output from the output terminal 5 and sent to the prediction circuit 3. In the prediction circuit 3, the DPCM code and the output of the predictor 7 are added in the adder circuit 6 and sent to the predictor 7.
The predictor 7 sends out a predicted signal at the next sampling time. The predicted signal passes through the digital-to-analog converter 8 and is sent to the subtraction circuit 2 as a predicted signal converted into an analog signal. A frame synchronization pulse is input from the synchronization signal separation device to the synchronization input terminal 12 in synchronization with the input signal from the input terminal 1 . This frame synchronization pulse is sent to the adaptive compression circuit 1 along with the sampling pulse sent from the quantization circuit 4.
0,11 to control the operation of the adaptive compression circuits 10,11 as described below.

4 is a block diagram showing the adaptive compression circuits 10 and 11 in the block diagram of the first embodiment shown in FIG. 3, and FIG. FIG. 2 is a level generation distribution diagram of an error signal and a characteristic diagram of a level compression circuit for explaining the operation. As shown in FIG. 4, each of the adaptive compression circuits 10 and 11 includes a compression circuit 20 and a control circuit 21.

The control circuit 21 used in this embodiment measures the level of the error signal, that is, the distribution of occurrence of prediction errors for each frame, and switches the level compression characteristics of the compression circuit 20 according to the result, thereby reducing the occurrence of prediction errors. The distribution is controlled so that it always concentrates around 0. Quantization circuit 4 in the block diagram of Figure 3
is the quantization characteristic shown in Fig. 2 so that the optimum signal-to-noise ratio can be obtained for a signal whose occurrence distribution of prediction errors is concentrated near 0 as shown in the distribution 40 of Fig. 6a. It has 30. When the occurrence distribution of prediction errors is less concentrated in the vicinity of 0, as shown in distribution 41 or 42 in FIG. 6a, the level compression characteristic 45 or 46 shown in FIG. 6b, respectively, is applied.
It is possible to convert the level so that the signal level generation distribution is concentrated around 0 using a compression circuit having the following. That is, the control circuit 21 measures for each frame to which of the distributions 40, 41, and 42 the prediction error of the separated signal belongs as shown in FIG. The separated signal passes through the level compression characteristic 44, that is, in FIG.
When it is 1 (or 42), the separated signal is as shown in Figure 6b.
In other words, in FIG.
8) is closed and sent to the adder circuit 13. Compression circuits 15 and 16 are both well-known circuits consisting of resistors and diodes, and switch circuits 14, 17,
Reference numeral 18 designates analog gates that are closed and conductive when the applied control signal is at a high level, and open and disconnected when the control signal is at a low level. Although quantization noise can be reduced by compressing the level of each separated signal as described above, waveform distortion occurs in the demodulated signal and distortion noise increases, so the level compression characteristics are based on quantization noise and distortion noise. The signal-to-noise ratio is determined by making a trade-off with the maximum signal-to-noise ratio. If the level compression characteristics are determined in this way, there is no need for the receiving side to have an expansion circuit that corresponds to the compression circuit on the transmitting side, and therefore there is no need to transmit information that indicates which compression circuit is being used on the transmitting side. do not have.

The control circuit 21 determines for each frame which of the distributions 40, 41, and 42 shown in FIG. 6a the generation distribution of the level of the separated signal from the separation circuit 9 belongs, and controls the compression circuit shown in FIG. control signal R for opening and closing switch circuits 14, 17, 18 in 20;
Generate S and U. First, the separated signal is sent to the comparator circuit 2.
Sent to 2,23. Comparison circuit 22 (or 2
3) consists of two voltage comparators and one OR gate, one of which is connected to the voltage E of the separated signal.
and the prediction error voltage B determined as shown in Figure 6a.
(or A) compared to E<B (or E<
A) A high level signal H (hereinafter abbreviated as H) is generated and input to one of the OR gates, and the other voltage comparator receives the voltage E and the prediction determined as shown in Figure 6a. H only when E>C (or E>D) compared to error voltage C (or D)
is generated and used as the other input of the OR gate. OR gate is H if either input is H.
Since the comparator circuit 22 (or 23)
sends H to the counting circuit 25 (or 26) only when E<B or E>C (or E<A or E>D). The counting circuit 25 (or 26) is reset to zero on every frame synchronization pulse,
Only while H is being sent from the comparator circuit 22 (or 23), the sampling pulses sent from the encoding circuit 4 are counted and a count signal P (or Q) is sent out. Further, the counting circuit 24 is reset to zero for each frame synchronization pulse, counts the sampling pulses from the encoding circuit 4, and sends out a counting signal N. Note that if the number of sampling pulses in one frame is determined in advance, the counting circuit 24 is not necessary. Read-only memory (hereinafter abbreviated as ROM)
27 uses the combination of the count signals N, P, and Q as address signals (N, P, Q) to read out signals W, X, and Y, and sends them to the control signal generation circuit 28. The correspondence between the address signals (N, P, Q) and the signals W, X, Y is determined as follows. The ratios α and β are determined in advance to determine which of the distributions 40, 41, and 42 the occurrence distribution of prediction errors belongs to, based on the address signals (N, P, Q) shown in FIG. , if the conditions Q/N<α and (P-Q)/N<β are satisfied, it belongs to distribution 40, and if the conditions Q/N<α and (P-Q)/N≧β are satisfied, it belongs to distribution 4.
1 or satisfies the condition Q/N≧α, it is determined that the distribution 42 belongs. The count signals N, P, and Q sent out each time a sampling pulse is input to the control circuit 21 satisfy only one of the three determination conditions described above. If the address signal (N, P, Q) satisfies the condition of belonging to the distribution 40, 1, 0, 0 is stored in the location corresponding to the address code (N, P, Q) of the ROM 27 as the signal W, X, Y.
If the condition belonging to the distribution 41 is satisfied, the signal W,
If X, Y are 0, 1, 0 or the conditions belonging to distribution 42 are satisfied, the signals W, X, Y are 0,
Store 0 and 1 respectively. The control signal generation circuit 28 includes a delay device that delays the signals W, X, and Y from the ROM 27 by the period T of the sampling pulse and sends them out, and a frame synchronization pulse as a set input, and outputs the signals from the delay device at the time of the frame synchronization pulse. It consists of a flip-flop that continues to send out a signal from the time of the frame synchronization pulse to the time of the next frame synchronization pulse. That is, the control signal generation circuit 28 measures the occurrence distribution of prediction errors in the previous frame for each frame and sends out the determined control signals as the control signals R, S, and U for each frame. The switch circuits 14 (or 17, 18) are closed when the control signal R (or S, U) is 1, that is, high level, and the switch circuit 14 (or 17, 18) is closed when the control signal R (or S, U)
is 0, that is, a low level, it is opened and the separated signal is sent to the adder circuit 13 through a compression circuit corresponding to the generation distribution of that level. Note that the adaptive compression circuits 10 and 11 in this embodiment each have three types of level compression characteristics, but one of the adaptive compression circuits 10 and 11 has one type of level compression characteristics, and the other has multiple types of level compression characteristics. It is obvious that a configuration in which the two types of level compression characteristics are provided, or a configuration in which both of them each have a plurality of types of level compression characteristics is also possible.

As explained above, by separating the luminance signal component and chromaticity signal component of a color television signal, passing them through a compression circuit that adapts to the respective level generation distributions, and then adding the two to perform quantization,
It is possible to directly encode each of the luminance signal and chrominance signal in a way that adapts to the prediction error occurrence distribution, and it is possible to improve the signal-to-noise ratio or It becomes possible to improve encoding efficiency.

FIG. 5 is a block diagram showing a second embodiment of the present invention. In the figure, an input signal from an input terminal 1 passes through an analog-to-digital conversion circuit 31, is converted into a digital signal, and is further sent to a subtraction circuit 32, where it becomes an error signal that is the difference from the prediction signal from a prediction circuit 30. Separation circuit 33 is subtraction circuit 32
It is composed of a band-pass type digital filter that passes only the chromaticity signal component of the error signal from the filter, and a band-elimination type digital filter that passes the luminance signal component and blocks the chromaticity signal component. The separated signals sent from the separating circuit 33 undergo level compression in adaptive compression circuits 34 and 35 that adapts to the occurrence distribution of prediction errors of each separated signal, and then are sent to the adding circuit 3.
6, the chromaticity signal and the luminance signal are added together to produce a signal that is a composite of both the chromaticity signal and the luminance signal. The adaptive compression circuits 34 and 35 are connected to the voltage comparator circuit 2 in FIG.
2 and 23 are replaced with digital comparison circuits, and compression circuits 15 and 16 are replaced with digital compression circuits. The digital compression circuit consists of a ROM, and compresses the signal level by using the separation signal from the separation circuit 33 as an address signal and reading out a signal corresponding to the compression characteristics shown in FIG. 6b. The signal sent from the adder circuit 36 is quantized in the quantizer circuit 37 and converted into DPCM.
It becomes a sign and is output from the output terminal 5, and
The signal is sent to the prediction circuit 30, where the adder 6 adds the prediction signal sent from the predictor 7, and the resultant signal is sent to the predictor 7. The prediction signal sent from the predictor 7 becomes an error signal, which is the difference between the prediction signal and the signal from the analog-to-digital conversion circuit 31, in the subtraction circuit 32.
Further, a frame synchronization pulse is input from the synchronization input terminal 12, and the operation of the adaptive compression circuits 34 and 35 is controlled in the same manner as in the first embodiment. This embodiment is a digitalized configuration of the first embodiment, and has an LSI
It has the advantage that it is suitable for industrial applications and the device can be made compact.

As explained above, the present invention enables direct coding by one predictive coding device while adapting to changes in the occurrence distribution of prediction errors of each signal component of a signal obtained by multiplexing a plurality of signal components over time. This has the effect of making it possible to construct a predictive coding device with a higher signal-to-noise ratio, lower transmission rate, and better coding efficiency than conventional direct coding devices without significantly increasing the scale. be.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing a conventional predictive coding device using a direct coding method, FIG. 2 is a characteristic diagram showing the quantization characteristics of the quantization circuit 4 in FIG. 1, and FIG.
FIG. 4 is a block diagram showing a first embodiment of the present invention, FIG. 4 is a block diagram showing the adaptive compression circuits 10 and 11 in FIG. 3, and FIG. 5 is a block diagram showing a second embodiment of the present invention. The block diagram and Figures 6a and b
4 are a prediction error occurrence distribution diagram and a level compression characteristic diagram, respectively, for explaining the operation of the circuit of FIG. 4. In the figure, 1... input terminal, 2, 32... subtraction circuit, 3, 30... prediction circuit, 4, 37... quantization circuit, 5... output terminal, 6... adder, 7...
...Predictor, 8...Digital-to-analog converter,
9, 33...separation circuit, 10, 11, 34, 35
...Adaptive compression circuit, 12...Synchronization input terminal, 1
3, 36...addition circuit, 31...analog-digital conversion circuit.

Claims (1)

[Claims] 1. A subtraction circuit that sends out an error signal representing the difference between an input signal containing a plurality of signal components and a predicted signal, and a separation circuit that separates the plurality of signal components included in the error signal and sends out a plurality of separated signals. Then, upon receiving one of the separated signals, it is determined which of a plurality of predetermined distributions the level generation distribution of the separated signal belongs to, and in response to the determination result, one of the plurality of predetermined level compression characteristics is determined. a level compression section that has at least one adaptive compression circuit that selects one of the plurality of separated signals and applies it to the separated signal; a quantization circuit that quantizes and outputs the signal sent out by the addition circuit; and a prediction circuit that predicts the prediction signal from the output signal of the quantization circuit and sends it out. A predictive encoding device comprising: