JPH01171376A - Predictive coding device for video signal - Google Patents

Abstract

PURPOSE:To widely decrease the generating degree of picture quality degradation by equipping a leak quantity controller, which sets the degree of a leak based on the output of a controller, and a variable leak device to multiply a first predictive value by a prescribed coefficient for the leak according to the instruction of this leak quantity controller and to introduce a multiplied result to a second adder. CONSTITUTION:Now, when correlatively is lowered, the absolute value of a predictive error 4 goes to be large and a switch 7 is closed to the side of a first quantizer 5. At such a time, a quantization representative value is also the large absolute value and for a value to be impressed to a second adder 17, the value from a first predictor 15 is occupational. At first, a coefficient in a variable leak device 19 is set to be small by a leak quantity controller 18 and the degree of the leak is strengthened. This is form a countermeasure against erroneous propagation at a predictive coding and decoding time. Here, the first coefficient in the variable leak device 19 is temporarily set to 0.5. After that, during a prescribed period, this condition is maintained. Here, the condition is maintained for five times operation. After that, the degree of the leak in the variable leak device is decreased by the leak quantity controller 18.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of the Invention The present invention relates to a predictive coding device, and more particularly to a predictive coding device for video signals that reduces error propagation and enables high-quality transmission.

Conventional technology In the transmission, recording and playback of video signals, various types of deterioration occur during the process, but in order to minimize these deteriorations and achieve high quality transmission, recording and playback, various techniques have been developed. A method is being considered. As one method, research is actively being conducted on digitizing the signal source and digitally performing processing such as transmission and recording/playback. By the way, this method makes it less susceptible to image quality deterioration, but on the other hand, it results in a large amount of data and a large increase in frequency bandwidth. Therefore, from the standpoint of effective use of transmission paths and recording media, a method of reducing the amount of data without substantially reducing image quality, that is, band compression (high efficiency coding), has become an important technical issue. (Hereafter referred to as “bandwidth compression”).

Predictive coding is one effective method for band compression.
It is also used in some cases to transmit video signals. This predictive coding (hereinafter referred to as "DPCM") is a relatively good method, but its biggest drawback is the error propagation phenomenon. Therefore, it is unsuitable for transmission lines and recording devices with high error rates. When magnetic tape is used as a recording medium, the S/N ratio of the transmission path is not very high, so there are many random errors, and burst errors due to dropouts also occur frequently, so there are problems in using DPCM.

Therefore, various methods have been proposed to reduce error propagation in DPCM, and examples of their application to recording and playback devices that use magnetic tape as a recording medium have been reported in recent years, such as in ``An Experimental Digital Video Recording System''. VI
DEORECORDING SYSTEM)j IEE Transactions on Consumer Electronics,
tions on Consumer Electro
nics), Vol CE-32, A3゜Augu
st 1986°, a conventional example of DPCM in which error propagation is reduced will be described below with reference to the drawings.

FIG. 7 is a block diagram showing a conventional example. In the figure, 1 is an input terminal, 2 is a subtracter, 8 is a controller, 5 is a first quantizer, 6 is a second quantizer, 7 is a switch, and 11 is a first adder. , 13 is the first nonlinear circuit, 14 is the second nonlinear circuit, 15 is the first predictor, 16 is the second predictor, 49 is the fixed leaker, 17 is the second adder, 9 is the word Converter, 1o is an output terminal, 3 is a predicted value, 4 is a prediction error, and 12 is a local signal. The difference between the value applied via input terminal 1 and predicted value 3, that is, prediction error 4, is calculated by subtracter 2.
and is guided to the controller 8, the first quantizer 15, and the second quantizer 16. The two quantizers 5 and 6 have different nonlinear quantization characteristics, and the controller 8 controls the switch 7 according to the degree of the prediction error 4, and selects one of the quantizers (first quantizer). quantizer 5 or the second quantizer 6).The output of the switch 7 is added to the predicted value 3 in the first adder 11,
This becomes the local signal 12. Local signal 12 is applied to first nonlinear circuit 13 and second nonlinear circuit 14 . Both valve linear circuits 13 and 14 have different nonlinear characteristics, for example coring and limiter characteristics, respectively. The outputs of the first nonlinear circuit 13 and the second nonlinear circuit 14 are led to a first predictor 16 and a second predictor 16, respectively.

The output of the first predictor 15 is applied to the second adder 17 via a fixed leaker 49. The fixed leak device 49 is the first
The output value of the predictor 15 is fixedly multiplied by a value less than one. The output of the second predictor 16 is also applied to the second adder 17, and the second adder 17 adds the two input values to create a predicted value 3, and the subtracter 2 and It leads to the first adder 11. Here, if the prediction error 4 is large, the controller 8 controls the switch 7 to switch the first quantizer 6
Close to the side. Moreover, the value of the quantization representative point output from the switch 7 is naturally a large value, and by passing it through the two nonlinear circuits 13 and 14, most of it is transmitted to the first predictor 15.
You will take the side route.

Generally, small errors do not significantly degrade DPCM decoding, but when large errors occur, the image quality after DPCM decoding is significantly degraded due to error propagation. When such a large error occurs, the DPCM decoding side usually applies a large prediction error representative point value, so if the prediction error is large, the fixed leakage The settings are such that a predicted value is created via the device 49. This works to suppress error propagation even if a large error is mixed in. The word converter 9 encodes the quantized representative value created in this way using a small number of bits, and sends it out via the output terminal 10.

By the way, in order to simplify the description of the two predictors 16 and 16, it is assumed here that two-point prediction is performed. In addition, the method of prediction will be further explained with reference to the drawings. Figure 8 is a pixel diagram, in which 50 is the nth line,
51 is the (n+1)th line, and 52 to 64 are pixels. In the first predictor 15 in FIG. 7, prediction of pixel 64 is performed using pixels 52 and 53, and each weight is tentatively set as %. That is, the average value of the pixel 52 directly above the pixel 64 and the pixel 53 directly to the left of the pixel 64 is used as the predicted value.

FIG. 9 is a block diagram showing the configuration of the first predictor 15 and fixed leak device 49. In the figure, 20 is a data input terminal, 21 is a first delay device, 22 is a second delay device, 23 and 24 are coefficient units, 25 is a third adder, 16
is a first predictor, 55 is a leak coefficient device, 49 is a fixed leak device, and 56 is a data output terminal. Data output from the first nonlinear circuit 13 is input to the first delay device 21 and the second delay device 22 via the data input terminal 2o.

The two delay devices 21 and 22 have one horizontal scanning period and one horizontal scanning period, respectively.
This is the delay of the pixel period, and the coefficients of the coefficient multipliers 23 and 24 are both %. The output values of the coefficient multipliers 23 and 24 are added by a third adder 25. 21-25 form the first predictor 15. The fixed leak device 49 includes a leak coefficient device 55, which multiplies the output of the third adder 25 by a coefficient less than 1. This result is transmitted via the data output terminal 56 to the second
The signal is supplied to the adder 17.

Problems to be Solved by the Invention Incidentally, in such a conventional method, deterioration in image quality is greater in areas where correlation is low than in ordinary DPCM. The reason is shown below.

When the correlation between images deteriorates significantly, the absolute value of the prediction error becomes extremely large. In this case, the output of the first quantizer of the quantization step is passed through the switch, and the representative value with large quantization distortion is sent to the prediction system. Since the large representative value is added to the first adder, most of it passes through the path of the first predictor and leaker. As a result, the prediction effect decreases due to the leaked predicted value, which leads to a vicious cycle in which image quality deterioration occurs and continues.

In view of the above-mentioned drawbacks, the present invention proposes a DPCM method that causes almost no deterioration in image quality even when correlation is significantly reduced, and effectively suppresses error propagation.

Means for Solving the Problems Therefore, in order to solve the above-mentioned problems, the present invention provides a subtracter that calculates a prediction error from a video signal and a predicted value, and a two-series quantizer that quantizes this prediction error. a switch that selects one of these two series of quantizers, a controller that controls this switch according to the prediction error described above, and creates a local signal by adding the output of this switch and the predicted value described above. a first adder for nonlinearly converting the local signal;
and a second nonlinear circuit; first and second predictors that create first and second predicted values based on the outputs of the first and second nonlinear circuits; An image comprising a second adder for creating the above-mentioned predicted value by adding the product multiplied by a coefficient for leakage and a second predicted value, and a word converter for converting the output of the above-mentioned switch into a word. The signal predictive coding device includes a leak amount controller that sets the degree of leak based on the output of the controller, and a leak amount controller that sets the degree of leak to the first predicted value according to instructions from the leak amount controller. This is a predictive coding device for a video signal, which includes a leak device that multiplies the video signal by a predetermined coefficient and then sends the video signal to the above-mentioned second adder.

Function: By configuring as described above, the period of leakage is limited in pixel areas where the correlation is low. As a result, error propagation during DPCM decoding is suppressed, and image quality in pixel areas with low correlation is greatly improved.

Example Embodiments of the present invention will now be described with reference to the drawings.

FIG. 1 is a block diagram showing an embodiment of the present invention, and in the same figure, 1 to 17 are the same as 1 to 17 in FIG. 7, respectively.
8 is a leak amount controller, and 19 is a variable leak device. 1~
Since 17 is the same as 1 to 17 in FIG. 7, detailed explanation thereof will be omitted. By the way, the controller 8 monitors the prediction error 4 and controls the switch 7. At the same time, data indicating the state of the prediction error 4 is applied to the leak amount controller 18.

Now, if we consider the case where the correlation is decreasing, the absolute value of the prediction error 4 will increase and the switch 7 will be switched to the first quantizer 6.
Close towards. At this time, the quantization representative point also has a large absolute value, and the value applied to the second adder 17 is dominated by the value from the first predictor 15. Therefore, initially, the coefficient of the variable leak device 19 is set small by the leak amount controller 18 to increase the degree of leak. This is to prevent error propagation during DPCM decoding.

Here, we assume that the initial coefficient of the variable leak device 19 is 0.5.
Let's set it to . After that, this state is continued for a predetermined period. Here, it is assumed that the process continues six times.

Thereafter, the leak amount controller 18 reduces the degree of leak in the variable leak device 19. Here, the coefficient is from 0.5 to 1
．． Let's change it to 0. If a large error is mixed in during DPCM decoding, leakage integration is performed during the decoding process, so if the leakage state can be maintained for a certain period of time, the effect of suppressing error propagation can be sufficiently exerted. Furthermore, even if the correlation is significantly reduced during the DPCM encoding process, the degree of leakage is reduced after a predetermined period of time, and the degree of deterioration is sufficiently improved even in images with low correlation.

Next, an example of the configuration of the first predictor 15 and variable leak device 19 in FIG. 1 is shown in a block diagram in FIG. In the figure, 16 and 20 to 26 are the same as 15 and 20 to 25 in FIG. 9, 19 is a variable leak device, 27 is a leak amount control terminal,
28 is a data output terminal, and 26 is a leak coefficient unit. 1
5 and 20 to 25 are similar to 15 and 20 to 26 in FIG. 9, so a detailed explanation will be omitted. On the other hand, the leak amount controller 18
This signal controls the coefficient of the leak coefficient unit 2e of the variable leak device 19 via the leak amount control terminal 27. That is, the coefficient here is 0.5 according to the instruction from the leakage controller 18.
Or controlled to 1.0.

Further, an example of the configuration of the leak amount controller 18 shown in FIG. 1 is shown in a block diagram in FIG. 3. In the figure, 29 is a control signal input terminal, 31 is a clock input terminal, 30 is a continuous length limiter, 3
2 is a coefficient generator, and 33 is a coefficient output terminal. The output of the controller 8 of FIG. 1 is applied to the control signal input terminal 29.

When the absolute value of the prediction error 4 exceeds a predetermined value, a signal is applied to the continuous length limiter 30 via the control signal input terminal 29.

On the other hand, a clock signal matching the pixel rate is applied to the clock input terminal 31, and a signal whose continuous length is limited to a period of five clocks is supplied to the coefficient generator 32 using the clock signal. The coefficient generator 32 outputs 0.5 or 1 depending on the output of the continuous length limiter 30. ○ is sent to the variable leak device 19 via the coefficient output terminal 33.

This situation will be further explained with reference to the waveform diagram shown in FIG. In the figure, 36 is a zero check signal, 37 is the output of the controller 8, 38 is the output of the continuous length limiter 30, 39 is the output of the coefficient generator 32, and 4° and 41 are times, respectively.

36 and 37 are signal waveforms applied to the clock input terminal 31 and the control signal input terminal 29, respectively;
At 0, the control signal 37 rises, indicating that the correlation has decreased significantly. After this, the continuous length limiter 30 outputs a high level output for 5 clocks, and then returns to the low level again at time 41, resulting in a waveform 38. Coefficient generator 32 generates waveform 38
"0" like 39 only between times 40 and 41.
．． 5" and then "1.0" again. Eventually, the coefficient shown in the waveform 39 is applied to the leak amount control terminal 27 in FIG. 2, and the leak amount is controlled.

Next, another embodiment of the leak controller 18 will be described with reference to the block diagram shown in FIG. 4 and the waveform diagram shown in FIG. 6.

In Figure 4, 29, 31.33 are 3 as in Figure 3.
4 is a prediction error input terminal, and 35 is a continuous length limiter. 2
9.31-33 are the same as 29°31-33 in FIG. 3, and detailed explanation will be omitted.

The prediction error 4 in FIG. 1 is input to the continuous length limiter 35 via the prediction error input terminal +3. In the embodiment shown in FIG. 3, the continuous length limiter 30 uniformly limits the continuous length by a predetermined number of clocks, but in the embodiment shown in FIG. 4, the continuous length is determined according to the situation of the prediction error 4. It is something to do.

This situation will be further explained with reference to the waveform diagram shown in FIG. In the same figure, 36.37 is the same as 36.37 in FIG. 5, 42 is a prediction error, 43 and 44 are thresholds, 46 is the output of the continuous length limiter 35, 46 is the output of the coefficient generator 32, 4
7 and 48 are times. At time 47, the control signal 37 becomes high level, and the output of the continuous length limiter 36 becomes high level as shown at 45. After that, it is determined by the relationship between the prediction error 42 inputted via the prediction error input terminal 34 and the two thresholds 43 and 44, and the level is high while the prediction error 42 is outside the two thresholds 43 and 44. is output. At time 48, the prediction error 42 falls between the two thresholds 43 and 44, and henceforth a low level is output, resulting in a waveform 45. Waveform 46 is continuous limiter 3
The coefficient generator 32 is controlled based on this output, and the output is "0.6" only between times 47 and 48 as shown in 46.
will be output.

By the way, in the detailed explanation of the present invention, in the embodiments shown in FIGS. 3 and 5, the period during which the amount of leakage is increased is set to 5.
The clock period may include, but is not limited to,
In the two embodiments shown in FIGS. 3 to 6, the leak coefficients are 0.5 and 1.0 when the amount of leak is large and when the amount of leak is small, but the invention is not limited to this.

Furthermore, in order to simplify the explanation of the prediction method, two-point prediction is taken as an example as shown in FIG. 2, but it goes without saying that the present invention is not limited to this method and can be applied to any prediction method. .

Furthermore, although the present invention can be applied to any transmission path, it is suitable for digital magnetic recording and reproducing of video signals using a magnetic recording and reproducing system as a transmission path where random errors and parsing errors coexist and occur frequently. It is extremely effective in fields such as equipment.

Effects of the Invention As is clear from the above explanation, the present invention not only effectively suppresses error propagation even if an error occurs in the transmission path, but also significantly reduces the correlation and increases the absolute value of the prediction error. By controlling the period of leakage, the degree of image quality deterioration can be significantly reduced, making it possible to perform DPCM encoding and decoding with high efficiency and high image quality.

[Brief explanation of the drawing]

Fig. 1 is a block diagram showing an embodiment of the present invention, Fig. 2 is a block diagram showing the configuration of the first predictor and variable IJ-reducer in Fig. 1, and Fig. 3 is the same as Fig. 1. 4 is a block diagram showing the configuration of the leak amount controller in FIG. 1, and FIG. 5 is a waveform showing the state of each part in FIG. 3. 6 is a waveform diagram showing the state of each part in FIG. 4, FIG. 7 is a block diagram showing a conventional example, FIG. 8 is a pixel diagram for explaining FIGS. 7 and 9, and FIG. FIG. 9 is a block diagram showing the configuration of the first predictor and leaker in FIG. 7. 6...First quantizer, 6...Second quantizer, 8...Controller, 9...Word converter, 13 ......First nonlinear circuit, 14.
...Second nonlinear circuit, 15...First predictor, 16...Second predictor, 19...
...Variable leak device. Name of agent: Patent attorney Toshio Nakao and 1 other person No. 1
Figure S Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 B; 8 Figure 3S4

Claims (3)

[Claims] (1) A subtracter that calculates a prediction error by subtracting a predicted value from the video signal to be transmitted, and a subtracter that quantizes this prediction error.
a quantizer and a second quantizer; a switch for selecting the output of either the first quantizer or the second quantizer; and a switch for selecting the output of either the first quantizer or the second quantizer; a controller for controlling, a first adder for creating a local signal by adding the output of the switch and the predicted value, and a first nonlinear circuit and a second nonlinear circuit for nonlinearly converting the local signal. , a first predictor that creates a first predicted value based on the output of this first nonlinear circuit, and a second predictor that creates a second predicted value based on the output of this second nonlinear circuit. the first predicted value multiplied by the coefficient for leakage, and the second predicted value
and a second adder for creating the predicted value by adding the predicted value to the predicted value, and a word converter for converting the output of the switch into a word. a leak amount controller that sets the degree of the leak;
A video signal characterized by comprising a variable leak device that multiplies the first predicted value by a predetermined coefficient for leakage according to instructions from the leak amount controller and leads the result to the second adder. predictive coding device. (2) The leak controller consists of a continuous length limiter that continuously outputs a signal for a predetermined number of pixels after receiving a signal from the controller, and only for the period during which this continuous length limiter continues to output a signal. 2. The predictive coding device for a video signal according to claim 1, further comprising a coefficient generator that outputs a value less than 1. (3) The leak controller consists of a continuous length limiter that outputs a signal continuously only during a period when the prediction error is outside a predetermined range after receiving a signal from the controller, and a continuous length limiter that outputs a signal continuously only during a period when the prediction error is outside a predetermined range. 2. The predictive coding device for a video signal according to claim 1, further comprising a coefficient generator that outputs a value less than 1 only during a continuous period.