JP2687973B2 - Predictive encoding method and decoding method - Google Patents

Description

TECHNICAL FIELD The present invention relates to a predictive coding and decoding method applied to transmission of image data such as a television signal. [Prior Art] DPCM is known as one of predictive coding methods.
The DPCM predicts the value of the current pixel from the value of the already coded pixel, and codes the prediction error between the value of the current pixel and the value of the predicted pixel. As another predictive coding method, the inventor of the present application, as described in Japanese Patent Application No. 59-174412, uses a linear primary combination of a plurality of neighboring pixels and a plurality of parameters to determine the pixel of interest. We propose a method that can make the compression rate extremely high by predicting and transmitting parameters that minimize the sum of squared errors. In these predictive coding methods, if the correction signal is not transmitted, errors will accumulate with time on the decoding side. In particular, it is important to start up from an initial state in which no reference data exists and to process when a scene change occurs. [Problems to be Solved by the Invention] When transmitting a predictively coded output at an extremely low bit rate, the receiving side may obtain a restored image little by little from the edge of the screen when rising from the initial state. It had the disadvantage of being ordinary and requiring a long time to complete the whole picture. Even when changing scenes, there was a drawback that the time required to switch from a previous scene to a new scene was long. Therefore, an object of the present invention is to provide a predictive coding and decoding method capable of transmitting an outline of the entire screen in a short time in the initial state and scene change, and then gradually improving the image quality of the entire restored image. To do. [Means for Solving Problems] According to the present invention, a difference between input image data and restored image data of a previous frame is formed for each block formed of a two-dimensional array of a plurality of pixel data, and the difference block The step of detecting the polarity of the integrated value in the above, and the correction amount having the polarity corresponding to the detected polarity is generated, and the correction amount is added to the restoration image data of the previous frame to obtain new restoration image data. The block size is set to the first predetermined value by the forming step and the reset signal generated when the temporal continuity of the image is broken, and the block size is set according to the set order with the passage of time. The step of gradually decreasing the power from the first predetermined value, and the reset signal sets the correction amount to the second predetermined value, and the set order is set with the lapse of time. Therefore, the predictive coding method is characterized in that it comprises a step of successively decreasing the correction amount from the second predetermined value. Further, according to the present invention, the difference between the input image data and the restored image data of the previous frame is formed for each block composed of a two-dimensional array of a plurality of pixel data, and the polarity of the integrated value in the difference block is detected. Generating a correction amount having a polarity corresponding to the detected polarity and adding the correction amount to the restored image data of the previous frame to form new restored image data; The block size is set to the first predetermined value by the reset signal generated when the continuous continuity is broken, and the block size is changed from the first predetermined value according to the set order with the passage of time. The correction amount is set to the second predetermined value by the step of sequentially decreasing and the reset signal, and the correction amount is set to the second position according to the set order with the passage of time. Decrease the values sequentially, send a signal indicating the detected polarity and a reset signal, receive a signal indicating the detected polarity and a reset signal, and send the block size by the reset signal The step of changing in the same sequence as the side and the reset signal to change the value in the same sequence as the transmitting side, and generate the correction amount having the polarity corresponding to the detected polarity to the restored image data. And a correction amount is added to form new restored image data, and a predictive coding and decoding method is provided. [Operation] A reset signal is generated in the initial state or when a scene change occurs. Starting with this reset signal, the blocks are successively reduced in size. At first, the block is made into a large block, and the rough step correction amount provides an overview of the entire image restored in a short time. After that, the blocks are gradually made smaller and the steps of the correction amount are made finer, so that the image quality of the restored image is gradually improved. When the steady state is reached, the block size is small and fine correction is performed, so that a good restored image can be obtained. Furthermore, the compression rate can be increased by reducing the number of bits of the correction amount. [Embodiment] An embodiment in which the present invention is applied to DPCM will be described below with reference to the drawings. FIG. 1 shows the configuration of a transmission system including both a transmitting side and a receiving side. In FIG. 1, a 1-sample 8-bit digital television signal from an input terminal 1 is supplied to a blocking circuit 2. Blocking circuit 2
Is supplied with the restoration signal of the previous frame from the frame memory 3. The blocking circuit 2 is a circuit for converting each of the input digital television signal and the restored signal in the order that coincides with the scanning of the television signal into the order of the block. The subtraction circuit 4 detects the difference between the restored signal and the input digital television signal in the block order by the blocking circuit 2. That is, assuming that the current signal of one pixel is Zk and the restoration signal of the corresponding pixel one frame before is k−1, the subtraction circuit 4 causes the frame difference ΔZk.
(= Zk-k-1) is calculated. This frame difference ΔZk
Is supplied to the integrating circuit 5. As the initial value of the restoration signal, an intermediate value, for example, in the case of 8 bits, a value of 128 is used. The frame difference ΔZk is supplied to the integrating circuit 5. The accumulator circuit 5 is 1 of the result of accumulating the frame difference ΔZk of 1 block.
The sign bit (representing + or-) of the bit is output.
This sign bit is taken out to the output terminal 6 as a correction signal. For example, "0" means + and "1" means-.
Further, the correction signal is supplied to the correction value generation circuit 7. The correction value formed by the correction value generation circuit 7 is supplied to the addition circuit 8. The addition circuit 8 adds the restoration signal from the blocking circuit 2 and the correction value to generate a restoration signal. The restored signal from the adder circuit 8 is supplied to the block decomposition circuit 9. The block decomposition circuit 9 is a circuit for returning the order of data from the order of blocks to the order of television signals. The restoration signal from the block decomposition circuit 9 is supplied to the frame memory 3. In this embodiment, a reset signal which is generated when the initial state is cut off or when a scene is changed is supplied from a terminal 10. This reset signal is taken out to the output terminal 11 and supplied to the control unit 12. A pulse signal having a field frequency synchronized with the input digital television signal is supplied to the control unit 12 from the input terminal 13. Control unit 12
Is a block size control signal for the block formation circuit 2, the integration circuit 5, and the block decomposition circuit 9 and the correction value generation circuit 7.
And a correction value control signal for. The control unit 12 is composed of, for example, a ROM. Blocking circuit 2 is controlled by the block size control signal.
The block size in the block decomposition circuit 9 is changed, and the range of data to be integrated in the integration circuit 5 is defined. Further, the correction value generated from the correction value generation circuit 7 is controlled by the correction value control signal. Control unit
When the reset signal is supplied to 12, the control of the block size and the correction value is started, and the block size is gradually reduced and the width of the correction value is gradually reduced according to a predetermined order. An example of the control of the block size and the correction value performed by the control unit 12 is shown below. The maximum block size is (8 lines × 16 pixels). One image (effective portion) is (184 lines × 640 pixels), as shown in FIG. Therefore,
In the case of a block size of (8 lines × 16 pixels), the number of blocks is (vertical 23 × horizontal 40 = 920). The minimum block size is (1 line × 1 pixel). "(1 line x 1 pixel)" means that no blocking process is performed, and means a steady state in which encoding is performed for each pixel. As the sign of the correction value, one of + and − is designated by the correction signal from the integrating circuit 5. That is, when the correction signal is +, the polarity of the correction value from the correction value generation circuit 7 is +,
When the correction signal is −, the polarity of the correction value is −. A 1-bit output code and a reset signal for each block taken out to the output terminals 6 and 11 are sent to the receiving side via the transmission path. In this case, the output code and the reset signal are in the form of serial data so as to be suitable for transmission, and encoding processing for error detection or error correction is performed as necessary. The reset signal from the input terminal 21 on the receiving side is supplied to the control unit 22. The 1-bit correction signal per block received from the input terminal 20 is supplied to the correction value generation circuit 24. A pulse signal having a field frequency is supplied from the terminal 23 to the control unit 22. The block size control signal and the correction value control signal are output from the control unit 22. These control signals are generated in the same sequence as that generated by the control unit 12 provided on the transmission side. Therefore, starting with the reset signal, the block size is gradually reduced from the larger one, and the correction value is gradually finer from the coarser one. A correction value control signal from the control unit 22 is supplied to the correction value generation circuit 24. The polarity of the correction value is controlled by the received correction signal. The correction value is supplied to the adding circuit 25. The adder circuit 25 is supplied with the data from the blocking circuit 26. The block circuit 26 is supplied with the restoration signal of the previous frame from the frame memory 27. A restoration signal is formed by the adder circuit 25, and the restoration signal is generated by the block decomposition circuit.
Supplied to 28. The restoration signals in the same order as the television signals are supplied to the frame memory 27 by the block disassembling circuit 28, and taken out to the output terminal 29. Frame memory
In 27, as an initial value, an intermediate value (128) is set for all pixels in the case of 8-bit quantization. The block size control signal from the control unit 22 is supplied to the block formation circuit 26 and the block decomposition circuit 28, and the block size is sequentially reduced. In addition, as described above, the quantization width of the correction value for each block is gradually changed from coarse to fine. Therefore, in the initial state, the restored image shows a rough overall image, and thereafter changes to a fine pattern. The present invention can also be applied to a predictive coding system other than the DPCM system. For example, the present invention can be applied to a method in which a pixel in the current field is assumed to be a linear combination of a large number of pixels in the vicinity of the past three fields, a parameter is identified by the least square method, and the parameter is transmitted. [Advantages of the Invention] According to the present invention, when there is no reference data at the receiving side as in the initial state of transmission or a scene change, it is possible to transmit almost the contents of an image in a short time, and In the state, a good restored image can be obtained by a small block size and fine correction. Also, by setting the correction signal to 1 bit per block,
There is an advantage that the amount of data to be transmitted hardly increases.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram of a transmission system according to an embodiment of the present invention, and FIG. 2 is a schematic diagram used for explaining an embodiment of the present invention. Description of main symbols in the drawings 1: Digital television signal input terminal, 2: Blocking circuit, 3: Frame memory, 5: Integration circuit, 7: Correction value generation circuit, 12: Control unit.

Continuation of front page    (56) References IEEE TRANSACTIONS                 ON COMMUNICATIONS                 VOL. COM-31 [4] (1983)                 P. 532-540 (The Lapla               cyan Pyramid as a               compact Image Cod               e. )

Claims (1)

(57) [Claims] Forming a difference between the input image data and the restored image data of the previous frame for each block composed of a two-dimensional array of a plurality of pixel data, and detecting the polarity of the integrated value of the difference in the block, Generating a correction amount having a polarity corresponding to the detected polarity and adding the correction amount to the restored image data of the previous frame to form new restored image data; The reset signal generated when the continuity is interrupted is set to the first predetermined value for the size of the block, and with the passage of time, the size of the block is set to the first predetermined value according to the set order. The step of successively decreasing from the value and the reset signal set the correction amount to the second predetermined value, and with the passage of time, follow the set order. Te, predictive coding method in which the correction amount, characterized in that it comprises a step of sequentially smaller from the second predetermined value. 2. Forming a difference between the input image data and the restored image data of the previous frame for each block composed of a two-dimensional array of a plurality of pixel data, and detecting the polarity of the integrated value of the difference in the block, Generating a correction amount having a polarity corresponding to the detected polarity and adding the correction amount to the restored image data of the previous frame to form new restored image data; The reset signal generated when the continuity is interrupted is set to the first predetermined value for the size of the block, and with the passage of time, the size of the block is set to the first predetermined value according to the set order. The step of successively decreasing from the value and the reset signal set the correction amount to the second predetermined value, and with the passage of time, follow the set order. The step of sequentially reducing the correction amount from the second predetermined value, the step of transmitting the signal indicating the detected polarity and the reset signal, and the step of transmitting the signal indicating the detected polarity and the reset signal. Receiving, changing the size of the block in the same sequence as the transmitting side by the reset signal, and changing the value in the same sequence as the transmitting side by the reset signal, and Generating a correction amount having a polarity corresponding to the detected polarity and adding the correction amount to the restored image data to form new restored image data. Encoding and decoding method.