JPH02159892A - Picture signal encoder - Google Patents

Abstract

PURPOSE:To improve the efficiency of encoding and transmission by providing a mode determining means to regulate a total code length for each picture and to allocate an encoding mode for each block so that minimum distortion can be obtained in a decoding side. CONSTITUTION:Mode encoding signals C1-Cn and mode local decoding signals F1-Fn are formed from a picture signal A, which is inputted to an input and P3 by first to n-th mode encoding circuits 11, 13 and 15 of a different quantization characteristic and forecasting processing. Then, the signals are added to a mode determining circuit 17 and an encoding signal selecting circuit 19. In the circuit 17, the total code length for each picture and the encoding mode, for which the distortion is made minimum in a restoring side, are determined and a mode signal J is outputted to show the encoding mode. In the circuit 19, the signals C1-Cn are selected based on the signal J and outputted as a transmitting encoding signal Cx to a signal rearranging circuit 23. ln the circuit 23, the signal Cx and a transmitting mode signal K from an error detecting code adding circuit 21 are outputted with being collected for all the blocks.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to an image signal encoding device using a predictive encoding method. Regarding equipment.

[Conventional technology]

Conventionally, a configuration as shown in FIG. 6 has been known as an image signal encoding device using a predictive encoding method.

That is, a prediction error signal E obtained by subtracting a prediction signal B, which will be described later, from a digital image signal in a subtraction circuit 1.
is output to the quantization circuit 3, and the quantization circuit 3 outputs an encoded signal C obtained by quantizing the prediction error signal E as a transmission signal.

On the other hand, the encoded signal C from the quantization circuit 3 is dequantized by the dequantization circuit 5 and converted into a representative value signal.

This representative value signal is added to the adder circuit 7 and the prediction signal B is added to output the local decoded signal F to the prediction circuit 9. , a predetermined prediction process based on the pixel combination of the previous sample and the previous line is performed to form a prediction signal B.
The subtraction circuit 1 and the addition circuit 7 are fed to the subtraction circuit 1 and the addition circuit 7, respectively. Note that the symbols P, , P2 in Figure 1 are input/output terminals.

The image signal encoding device having such a configuration predicts the value of the original pixel currently being encoded using the locally decoded signal F of the already encoded pixel to form a preliminary 11J signal B, and The prediction error signal E obtained by subtracting the prediction signal B from the prediction signal B is quantized and transmitted.

Generally, there is a strong tendency that image data values do not change significantly between adjacent pixels in one image signal, so the difference between the predicted value and the original pixel value becomes very small.

The use of the above-mentioned predictive coding method has the advantage that one image signal can be highly compressed and transmitted.

[Problem to be solved by the invention]

However, such image signal encoding devices quantize and transmit the prediction error signal obtained from the original pixel signal and the prediction signal of the already encoded pixel, so if an error occurs in the transmission signal on the transmission path, the decoding On the other hand, there is a drawback that errors in subsequent pixels spread, greatly deteriorating the quality of the decoded image.

However, as a method for suppressing the spread of signal errors that occur on the transmission path, the transmission signal is multiplied by a predetermined coefficient before being transmitted, so that the error component in the transmitted transmission signal is attenuated. It has been proposed to employ a so-called leaky integral predictive coding method in an image signal coding apparatus, but this configuration has the disadvantage that the coding efficiency decreases because the transmission signal is multiplied by a coefficient.

In addition, in order to improve the coding efficiency of a single image signal, a technology has been proposed in which the quantization characteristics are switched according to the image state. signal) must be added and transmitted, and the mode signal is mistakenly transmitted to the decoding side during the transmission path.

The problem is that the entire screen is destroyed.

Furthermore, a technique has been proposed in which a single image signal encoding device quantizes one image signal and performs encoding processing to make all quantized signals of equal length before transmission.

However, 2. Image characteristics 2. For example, in cases where the prediction error is large and a long code length is required, such as in the outline of an image, image quality deterioration is likely to occur due to waveform distortion of the decoded signal due to gradient overload. On the other hand, when the prediction error is small and can be processed with a short code length, such as in a flat area,
Since extra bits are added and transmitted, there is still room for improvement from the viewpoint of signal processing and transmission efficiency.

The present invention has been made to solve these conventional drawbacks, and even if a code error occurs in a transmission signal on a transmission path, it is possible to prevent the error from spreading to pixels that are restored on the decoding side for subsequent transmission signals. It is an object of the present invention to provide an image signal encoding device capable of making deterioration in one image quality less noticeable within an extremely narrow range.

Another object of the present invention is to provide an image signal encoding device capable of restoring an image that can be recognized without destroying the entire screen on the decoding side even if an error occurs in a mode signal in a transmission path.

Furthermore, the purpose of the present invention is to provide an image signal encoding device that can easily obtain a desired amount of transmission codes for each screen, and that can obtain good encoding processing and transmission efficiency tailored to the characteristics of one screen.

[Means to solve the problem]

In order to achieve such an object, the present invention is configured to include a plurality of encoding means, mode determination means, encoded signal selection means, and signal rearrangement means.

The plurality of encoding means divides one image signal into a plurality of blocks of a predetermined size each consisting of a plurality of pixels for each screen,
In each of these blocks, a quantization mode that quantizes the prediction error between the predicted value using only the pixels in the block and the original pixel value, and an encoding mode that differs in the code length of the quantized signal are used. In each block, one specific pixel is left as a PCM signal, and the PCM signal and the other pixels are predictively encoded using the encoding mode. The configuration is such that it outputs as an encoded signal and also outputs two local subsequent signals.

Further, the mode determining means determines a mode signal to which the plurality of encoding modes are assigned for each program using each local decoded signal and the original image signal, and the encoded signal selecting means determines the mode signal to which the plural encoding modes are assigned for each program, and the encoded signal selecting means One of the above encoded signals is selected based on the signal.

Further, the signal rearranging means rearranges the mode signals, PCM signals, and encoded signals in each block, and sequentially collects the mode signals, PCM signals, and encoded signals of each program for each screen. It is configured to transmit.

In one aspect of the present invention, the mode determining means is configured to determine a mode signal for assigning to each block an encoding mode that provides a predetermined transmission code amount for one screen as a whole and minimizes distortion with respect to the original image signal. is also possible.

[For production]

In the present invention equipped with such a means, one image signal is applied to a plurality of encoding means each having a different encoding mode, and each pixel constituting each block is processed by each encoding means according to its position. The pixels in each block are predictively encoded, and each encoding means generates a PCM signal, a predictively encoded encoded signal with different quantization characteristics and code length, and a local decoded signal used in the predictive encoding. The mode determining means allocates and determines a coding mode for each block and outputs a mode signal.

This mode signal causes the encoded signal selection means to select an encoded signal from one of the encoding means for each block.

In the signal rearranging means, the mode signal of each block.

The PCM signals and encoded signals are rearranged, and the mode signals, PCM signals, and encoded signals of all blocks in each screen are collectively transmitted sequentially.

Further, in a configuration in which the mode determining means assigns to each block an encoding mode that provides a predetermined transmission code amount and minimum distortion for one screen as a whole, encoding is performed by assigning a code length according to the characteristics of one screen.

〔Example〕

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

FIG. 1 is a block diagram showing an embodiment of an image signal encoding device according to the present invention.

First, the outline of the present invention will be briefly explained using FIG.

In the figure, P CM (Pulse Code
The input terminal P3 to which the modulated digital image signal is input is connected to a plurality of first to nth mode encoding circuits 11, 13.15 having different encoding modes and a mode determining circuit 17. The first to nth mode encoded signals 01 to C0 and the first to nth mode local decoded signals F1 to F are output from the first to nth mode encoding circuits 11.13.15. .

Each of the first to nth mode local decoded signals Fl-FTI is connected to the mode determining circuit 17, and each of the first to nth mode encoded signals 01 to CTl is output to the encoded signal selection circuit 19.

Mode determination circuit 17 outputs mode signal J to encoded signal selection circuit 19 and error detection code addition circuit 21.

The encoded signal selection circuit 19 selects the first to nth mode encoded signals C! for each block based on the mode signal J. ~CTl is selected and output to the signal rearrangement circuit 23.

The error detection code addition circuit 21 adds an error detection code to the mode signal J and outputs the transmission mode signal K to the signal rearrangement circuit 23.

The signal rearrangement circuit 23 rearranges the PCM signal and the encoded signal into the transmission mode signal of each block in the current screen, and outputs them all together as a transmission signal from the output terminal P4.

FIG. 2 shows the first to nth mode encoding circuits 1 shown in FIG.
It is a blog diagram showing an example of any one of 1, 13 and 15.

In FIG. 2, an input terminal P3 to which the image signal A is input is connected to a subtraction circuit 25 and a first selection circuit 27.

The subtraction circuit 25 extracts a prediction signal B1, which will be described later, from one image signal 8.
~Bn is subtracted to output a prediction error signal E, and is connected to a quantization circuit 29 that quantizes the prediction error signal E with a predetermined quantization characteristic and outputs a quantized signal G.

The quantization circuit 29 is connected to a first selection circuit 27, and this first selection circuit 27 quantizes the image signal A from the input terminal P3 and the quantization based on the pixel position signal I from the pixel position detection circuit 31. The quantized signal G from the quantization circuit 29 is selected and outputted as the i-th mode encoded signal C1, and the i-th mode encoded signal Ci is outputted from the output terminal P5 (not shown in FIG. 1).

It is connected to an inverse quantization circuit 33 and a second selection circuit 35.

For example, as shown in FIG. 3, the pixel position detection circuit 31 determines to which position of the pixel in each block the input image signal A is applied when the screen is divided into a plurality of blocks each consisting of a plurality of pixels. It outputs a pixel position signal H indicating information on whether the pixel is applicable or not.

The first selection circuit 27 receives an image signal applied to the input terminal P by inputting a pixel position signal H indicating a pixel indicated by - (specific pixel) shown in the upper left with diagonal lines in each block in FIG. i-th mode encoded signal C with A as a PCM signal
It has a function of outputting a quantized signal G for pixels other than the specific pixel.

The inverse quantization circuit 1i33 inversely transforms the input i-th mode encoded signal Ci into a representative value signal indicating a predetermined representative value and outputs it to the second selection circuit 35. The circuit 35 selectively switches the i-th mode encoded signal Ci or the representative value signal from the first selection circuit 27 according to the pixel position signal H, and outputs it to the addition circuit 37. That is, depending on the pixel position in each block,
It selectively outputs the mode encoded signal Ci and the representative value signal.

The adder circuit 37 adds a predicted signal B, which will be described later, to the selected signal from the second selection circuit 35 and outputs an i-th mode locally decoded signal 1"i, so that the first mode locally decoded signal Fi is It is connected to the first to third prediction circuits 39, 41, 43 and the output terminal Ps (not shown in FIG. 1).

Each of the first to third prediction circuits 39, 41, and 43 performs different prediction processing. for example.

The first prediction circuit 39 performs prediction processing using the previous sample, the second prediction circuit 41 performs prediction processing using the pixels of the previous line, and the other prediction circuits perform prediction processing using the previous field. Prediction processing is performed using the average of the pixels of the previous sample and the previous line, and each is connected to the third selection circuit 45.

The third selection circuit 45 always performs pixel prediction within each block, and is connected to each of the first to third prediction circuits 39 .
The prediction signals B1 to Bn selected from the prediction signals B1 to B7 from 41.43 based on the pixel position signal H from the pixel position detection circuit 3I are outputted as a feed bank to the subtraction circuit 25 and the addition circuit 37.

In such an i-th mode encoding circuit, when an image signal A is inputted to the input terminal P3, it is applied to the first selection circuit 27, and the subtraction circuit 25 subtracts it from the prediction signals B1 to Bi. The prediction error signal E is quantized by the quantization circuit 29 to become a quantized signal G, which is sent to the first selection circuit 27.
added to.

The first selection circuit 27 uses the pixel position signal I from the pixel position detection circuit 31 to output the image signal A as a PCM signal to the output terminal PS only in the case of a specific pixel in each block as the i-th mode encoded signal Ci. On the other hand, for other pixels, the quantized signal G is selected and output.

Then, the PCM signal and quantized signal G as the first mode encoded signal C1 are applied to the second selection circuit 35 directly or via the inverse quantization circuit 33.

In the second selection circuit 35, one of them is selected according to the pixel position signal H and added to the addition circuit 37.

In the adder circuit 37, the selected signal I and the prediction signal B are added to form an i-th mode locally decoded signal Fi, which is output to the first to first prediction circuits 39, 41, 43 and the output terminal Ps.

Each of the first to third prediction circuits 39, 41.43 performs different predictions to form prediction signals B, -BT1.
In addition to the selection circuit 45, the third selection circuit 45 selects each of the first to fourth prediction circuits 39, 4 based on the pixel position signal H.
The prediction signals B1 to BT1 from 1.43 are selected and added to the subtraction circuit 25 and the addition circuit 37, and this process is repeated thereafter.

In this way, in the image signal A input to the input terminal P3, each specific pixel in each block is output as a PCM signal, and the other pixels are predicted using only the pixels in each block as shown in Figure 3. Prediction signal B selected according to the position of the pixel! ~B0 is used to form a prediction error signal E, and 2M quantized i-th mode encoded signal Ci and i-th mode locally decoded signal Fi are output.

Each of the first to nth mode encoding circuits 11°13.1
5, the quantization characteristics when quantizing the prediction error signal E and the prediction processing of the first to first prediction circuits 39, 41, and 43 are different, and the first to first prediction circuits 39, 41, and 43 have different code lengths. The n-mode encoded signals cl to C7 and the first to n-th mode locally decoded signals F! to F are output.

Returning to FIG. 1, the first to nth mode encoding circuits 11.1
1st to nth mode local decoded signals F1 to F from 3.15
Mode determining circuit 17 to which TI and image signal A are input
selects the total code length for each screen to a desired value, determines the encoding mode to be assigned to each block so that the distortion of the restored image to the input image signal A is minimized, and selects the mode signal J.
This is what is output as.

The mode determining circuit 17 includes a distortion calculating circuit 47, for example, as shown in FIG. Sorting circuit 49. Threshold determination circuit 51. It is composed of a mode assignment circuit 53. here,
- As an example, mode determination of two encoding modes, a first mode with a low compression rate and a second mode with a high compression rate, will be explained with reference to FIG.

The distortion calculation circuit 49 compares the second mode locally decoded signal F2 with the image signal A, calculates distortion for each encoding mode for each block, and outputs the result to the rearrangement circuit 49 and the mode assignment circuit 51. Note that the distortion is, for example, the sum of absolute differences.

The rearrangement circuit 49 rearranges the distortions in each block in ascending order for each encoding mote, and is connected to a threshold determination circuit 51 that determines a threshold value.

For example, in order to obtain the desired code length for one screen as a whole, the number of blocks in the first mode is reduced to 1/3 of the total number of blocks, and the number of blocks in the second mode is reduced to 1/3 of the total number of blocks.
Considering a case where the number of mode blocks is set to 2/3 of the total number of blocks, the threshold value determination circuit 51 determines the threshold value at 1/3 of the one with the largest distortion.

The mode assignment circuit 53 assigns the first mode to blocks having distortions less than the threshold value and the second mode to the remaining blocks based on the determination signal from the threshold determination circuit 51, for example.

The encoding mode is determined so that a constant desired code length is obtained for each screen and the distortion of the entire screen is minimized, and a ζ mode signal is output.

The circuit configuration shown in FIG. 4 is a configuration for outputting two types of mode signals and is shown for convenience of explanation, and the first to nth mode locally decoded signals F! ~FT1 will be configured to be selectively determined.

The encoded signal selection circuit 19 selects the '1st to nth mode encoder circuits 1 according to the mode signal J from the mode determination circuit l7.
Each of the first to nth mode encoded signals C from 1.13.15
1 to C, are selected for each block to generate the transmission encoded signal Cx.
The error detection code adding circuit 21 adds an error detection code 2, for example, parity, to the mode signal J and outputs it to the signal rearrangement circuit 23 as a transmission mode signal.

The signal rearrangement circuit 23 has 1. Transmission coded signal Cx from coded fd code selection circuit 19 and error detection code addition circuit 2
A total of 1072 minutes of transmission mode signals K from 1 to 1 are output together.

That is, as shown in FIG. 5, for the current screen data for one field or one frame, all 1072 parts of the transmission mode signal K are summarized.9Then, all 1072 parts of the PCM signals of specific pixels of each block are summarized. , a total of 1072 encoded signals predicted at the end, and all 1072 of these
The transmission mode signal, the PCM signal, and the encoded signal are transmitted in sequence.

In the image signal encoding device of the present invention configured as described above, the image signal A input to the input terminal P3 is input to each of the first to nth mode encoding circuits 11, 13 . 15, the first to nth mode encoded signals cl -cm and the first to nth mode locally decoded signals F1
~FTl is formed and added to the mode determination circuit 17 and the encoded signal selection circuit 19, and the mode determination circuit 17 determines the encoding mode that minimizes the total code length for each screen and the distortion on the restoration side. A mode signal J indicating the encoding mode is output.

The encoded signal selection circuit 19 selects the first to nth mode encoded signals 0 from each of the first to nth mode encoding circuits 11 and 13.15 based on the mode signal J from the mode determination circuit 17.
1 to C7 are selected and output to the signal rearrangement circuit 23 as a transmission encoded signal Cx.

The signal rearrangement circuit 23 combines the transmission mode signal to which the error detection code has been added, the transmission encoded signal, that is, the PCM signal and the predictive encoded signal, into a transmission mode signal in the order of the PCM signal and the predictive encoded signal. Output to the transmission line as a transmission signal.

Note that in the present invention, it is not necessary to set the total code length for each screen to a constant value or to suppress distortion.

The configuration of the mode determining circuit 17 can be simplified.

〔Effect of the invention〕

As explained above, the present invention has a configuration in which closed predictive encoding is performed within each block for each pixel of each block using a plurality of encoding means with different encoding modes.
Even if a code error occurs in the PCM signal or encoded signal in the middle of each transmission path, the spread of the error remains within each narrow block, thereby suppressing deterioration in image quality.

Furthermore, select the encoding mode for each block.

Since the PCM signal, the predictively coded coded signal, and the mode signal indicating the coding mode for each block are transmitted together in all blocks for each screen, even if the mode signal is erroneously transmitted on the transmission path, it will not be detected on the decoding side. It becomes possible to decode the image using the PCM signal, and it is possible to roughly restore the image to the extent that the contents of the two images can be recognized, thereby preventing image failure.

Therefore, in the image signal encoding device of the present invention, it is easy to make deterioration in image quality on the decoding side less noticeable.

Furthermore, in a configuration in which the mode determining means specifies the total code length for each screen and assigns the encoding mode to each block so as to obtain the minimum distortion on the decoding side, the characteristics of nine screens, such as the It is easy to give a long code length to blocks where large prediction errors occur, such as contours, and short code lengths to blocks where only small prediction errors occur, such as flat areas, improving encoding and transmission efficiency. be able to.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing an example of the image signal encoding device according to the present invention, FIG. 2 is a block diagram showing an example of the mode encoding circuit in FIG. 1, and FIG. 3 is a block diagram showing an example of the mode encoding circuit in FIG. FIG. 3 is a diagram schematically showing pixel prediction. FIG. 4 is a block diagram showing an example of the mode determination circuit shown in FIG. 1, FIG. 5 is a diagram showing signal states transmitted from the signal rearranging circuit shown in FIG. 1, and FIG. 6 is a diagram showing conventional image signal encoding. FIG. 2 is a block diagram showing the device. 1.25... Subtraction circuit 3.29... Quantization circuit 5.33...
・・・・・・Inverse quantization circuit 7.37・・・・・・・・・
Addition circuit 9・・・・・・・・・・・・Prediction circuit I
1.13.15 Encoding means (first to nth mode encoding circuits) 17 Mode determining means (mode determining circuit) 19・・・・・・・・・・・・・・・Encoded signal selection means (encoded signal selection circuit) 21・・・・・・・・・・・・・・・Error Detection code addition circuit 23・・・・・・・・・・・・・・・・・・
Signal rearranging means (signal rearranging circuit)

Claims (2)

[Claims] (1) Divide the image signal into multiple blocks of a predetermined size each consisting of multiple pixels for each screen, and in each of these blocks, calculate the predicted value using only the pixels in the block and the original pixel value. a plurality of encoding means each having a quantization mode for quantizing a prediction error of the quantization signal and an encoding mode for each having a different code length of the quantized signal; a plurality of codes for outputting the PCM signal and a signal predictively encoded by the encoding mode for pixels other than the specific one pixel as encoded signals, and outputting a locally decoded signal; mode determining means for determining a mode signal for assigning the plurality of encoding modes to each block using each of the local decomposition signals and the original image signal; coded signal selection means for selecting each of the coded signals, and rearranging the mode signal, PCM signal, and coded signal in each block to select the mode signal, PCM signal, and coded signal of each block for each screen. An image signal encoding device comprising: signal rearranging means for sequentially transmitting signals individually; (2) The mode determining means determines the mode signal that allocates to each block an encoding mode that provides a predetermined transmission code amount for one screen as a whole and minimizes distortion with respect to the original image signal. The image signal encoding device described above.