JPH0453379A - Moving picture decoder - Google Patents

Abstract

PURPOSE:To prevent a reproduced picture from being disturbed even when an incorrectable error takes place by obtaining movement information between a correct decoding picture frame just before a frame in which erroneous picture information exists and a correct decoding picture frame reached afterward and using the movement information so as to interpolate it to the erroneous picture information. CONSTITUTION:The decoder consists of a channel decoding means 1, a separation means 3, a decoding means 4, a frame interpolation means 9, a picture storage means 10, a movement detection means 11 and an error compensation inter-frame decoding means 12. When an incorrectable error is detected during the transmission of picture information subject to band compression, movement information is obtained between a correct decoding picture frame just before a frame in which erroneous picture information exists and a correct decoding picture frame reached afterward and the moving information is used to interpolate it to the erroneous picture information. Thus, even when an incorrectable error takes place, the moving picture decoder is realized, in which the reproduced picture is not disturbed.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to a decoding device for moving image information transmitted by band compression.

(Prior Art) In recent years, research on the transmission of digital moving images and their encoding and decoding has been progressing mainly in the communication field.

FIG. 4 is a block diagram showing a conventional video encoding device.
41 is a preprocessing means, 42 is a determination means, 43 is an orthogonal transform encoding means, 44 is a quantization means, 45 is an inverse quantization means, 4
6 is orthogonal transform decoding means, 47 is motion compensated interframe coding means, 48 is motion detection means, 49 is variable length coding means, and 50 is channel coding means.

Each frame of the original image input from the transmission path T1 is divided into an encoded frame and a thinned-out frame. That is, first, the original image is input to the preprocessing means 41,
A block signal is obtained by subsampling and blocking each frame of the original image and is input to a determining means 42, which determines whether the frame is a coded frame or a thinned-out frame.

In the case of a coded frame, it is then determined whether the block signal is to be coded by interframe coding or intraframe coding. In the case of interframe coding, a block difference signal between the block signal and the prediction signal predicted by the motion compensated interframe coding means 47 is sent to the orthogonal transform coding means 43, and in the case of intraframe coding, the block signal is sent as it is. It is sent to the orthogonal transform encoding means 43.

If an uncorrectable error occurs in the transmission path during consecutive frames using interframe coding, the error will continue to propagate to subsequent frames, so to avoid this, intraframe coding is periodically performed. A refresh frame consisting only of blocks that have been previously inserted is inserted, but normally when there is a scene change, the first frame acts as a refresh frame.

On the other hand, the motion detection means 48 refers to the block signal and the prediction signal in the frame buffer of the previous frame in the motion compensated interframe coding means 47.

Detects motion information between the previous frame and the previous frame.

In the case of a thinned frame, only motion information between the previous frame and the previous frame is detected, and subsequent encoding is not performed.

The orthogonal transform encoding means 43 performs two-dimensional orthogonal transform on the block signal or the block difference signal and converts it into transform coefficients. The spatial correlation of the image signal is removed from the transform coefficients, and energy is concentrated in some coefficients. The quantization means 44 performs quantization by weighting the coefficients in which energy is concentrated among the transform coefficients. The variable length encoding means 49 assigns codes to each of the quantized outputs of the previous stage so that the average code length becomes short. Thereafter, the main information to which the code has been assigned is transmitted to the channel encoding means 50 such as encoding determination information from the determining means 42 and motion information from the motion detecting means 48.
It is combined with sub-information such as quantization step information from the quantization means 44 and motion information of each block of the thinned-out frame, is further added with an error correction code, etc., and is sent to the transmission path T2.

On the other hand, the dequantization means 45 dequantizes the quantized output using the quantization step information from the quantization means 44 to create local decoding transform coefficients, and the orthogonal transform decoding means 46 performs orthogonal transform decoding to perform local decoding. Output.

If the local decoding is determined to be an inter-frame code by the determination means 42, the predicted signal is added and sent to the motion-compensated inter-frame encoding means 47, and if it is determined to be intra-frame encoding, the local decoding is performed directly as a motion-compensated frame. Intercoding means 47
sent to. The motion compensation interframe encoding means 47 performs motion compensation using the motion information detected by the motion detection means 48, and outputs a predicted signal.

FIG. 5 is a block diagram of a conventional moving image decoding device.

In FIG. 5, 51 is a channel decoding means, 52 is an error detection and correction means, 53 is a separation means, 54 is a decoding means, 55 is a variable length decoding means, 56 is an inverse quantization means, and 57 is an orthogonal transform decoding means. 58 is a motion compensated inter/intraframe decoding means, and 59 is a frame interpolation means.

The image information from the transmission path is applied to the communication path decoding means 51, where error detection and correction, etc. are performed, and after the image information is separated into main information and sub information by the separation means 53, information corresponding to one block is further processed. is detected. The separated main information and sub information are each sent to decoding means 54. In this example, the decoding means 54 includes a variable length decoding chemical stage 55. It is composed of an inverse quantization means 5G, an orthogonal transform decoding means 57, and a motion compensation inter/intraframe decoding means 58. The main information sent to the decoding means 54 is first converted into code-assigned information by the variable length decoding means 55 into a decoded and quantized output. The inverse quantization means 56 inversely quantizes the quantized output using the quantization step information of the side information to create decoded transform coefficients. The orthogonal transform decoding means 57 performs orthogonal transform decoding on the decoded transform coefficients to obtain a decoded block signal.

The motion compensated inter/intraframe decoding means 58 includes at least two
It has a frame buffer for one frame, and detects whether the encoding means of the decoded block output is interframe encoding or intraframe encoding using encoding method determination information of the sub information. In the case of intra-frame encoding, the decoded block output is directly written as a decoded block signal to a designated position in the frame buffer of the current frame. In the case of interframe coding, a decoded block signal is created by adding the block signal at the position specified by the motion information in the frame buffer of the previous frame and the decoded block output using the motion information of the sub information. Writes to the specified position in the frame buffer.

As described above, the decoded image frame for one frame is
A motion compensated inter/intraframe decoding means 58 is formed in a frame buffer.

Next, the encoder interpolates the thinned frames using the motion information of the thinned frames in the sub information. The frame interpolation means 59 calculates block signals at positions in the frame buffer of the immediately preceding and following decoded image frames indicated by the motion information of the thinned frame in the sub-information, and writes them into the specified positions of the frame buffer of the interpolated frame. In this way, an interpolated frame is generated and interpolated to the thinned-out frame.

(Problem to be Solved by the Invention) However, with the above configuration, if an error that occurs during transmission cannot be corrected by a decoder on the communication path, the image of the frame containing erroneous image information cannot be correctly decoded. The image is distorted. Furthermore, if interframe coding is used, or if frames are thinned out using preprocessing means during encoding and the thinned out frames are interpolated during decoding, incorrect image information may be generated in subsequent frames. There is a problem in that the error propagates to the referenced location and the image becomes distorted, which continues until a regularly inserted refresh frame is input.

In view of the above, the present invention interpolates an erroneous block using an error-free block in the previous and subsequent frames even when an uncorrectable error occurs during transmission, so that a good image can be reproduced. The purpose is to provide a decoding device.

(Means for Solving the Problems) The present invention achieves the above object by providing an error detection and correction means for detecting and correcting errors in band-compressed image information, and separating error-corrected image information into main information and sub information. a decoding means for decoding main information using sub information to generate a decoded image frame; an image storage means for storing a plurality of decoded image frames; frame interpolation means for interpolating frames using the image frame and sub information; motion detection means for detecting motion information between correct decoded image frames immediately before and after the error frame containing uncorrectable error pixels; This is achieved by comprising error compensation interframe decoding means for interpolating uncorrectable error pixels existing in a frame using motion information.

(Function) With the above-described configuration, the present invention corrects uncorrectable errors during transmission between the correct decoded image frame immediately before the frame containing the incorrect image information and the correct decoded image frame that arrives thereafter. Since motion information is obtained and erroneous image information is interpolated using the motion information, a moving image that does not disturb the reproduced image even if an uncorrectable error occurs can be obtained.

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

FIG. 1 is a block diagram of a moving picture decoding device according to an embodiment of the present invention, in which 1 is a channel decoding means, 2 is an error detection and correction means, 3 is a separation means, 4 is a decoding means, and 5 is a variable length a decoding means, 6 an inverse quantization means, 7 an orthogonal transform decoding means, 8 a motion compensation inter/intra frame decoding means, 9 a frame interpolation means, 10 an image storage means, 11 a motion detection means, Reference numeral 12 denotes error compensation interframe decoding means.

FIG. 2 is a block diagram of an image storage means, which is a main part of an embodiment of the present invention. In FIG. 2, 21 is a frame buffer group, 2
2 is the previous refresh frame, 23 is the error frame,
24 is a frame immediately before the error frame, 25 is a subsequent refresh frame, 26 is an error block, and 27 is a block that refers to the error block 26.

FIG. 3 is a block diagram of an image storage means of a moving picture decoding device which is an embodiment of the present invention. In FIG. 3, 21 is a frame buffer group, 32 is the immediately previous decoded image frame, and 33 is a frame buffer group.
is an error frame, 34 is the immediately following decoded image frame, 35
is an error block.

The operation of the moving picture decoding apparatus configured as shown in FIGS. 1, 2, and 3 will be described below. It should be noted that the moving picture encoding device of this embodiment is the same as that shown in the conventional example, so the explanation will be omitted.

In FIG. 1, image information encoded by a moving image encoding device and arriving via a transmission path is transmitted to a communication path decoding means 1.
If an uncorrectable error occurs after error detection and correction is performed, if the error is information on an encoded frame, processing of the corresponding block is stopped; if the error is information on a thinned frame, processing is stopped. Continuing as it is, the error detection and correction means 2 and the error compensation interframe decoding means 12, respectively,
The type of error frame and the position of the error block are notified to the motion compensation inter/intra frame decoding means 8 and the frame interpolation means 9, and the process moves on to the next block.

If the error correction is successfully performed, the image information is transferred to the separation means 3.
The information is separated into main information and sub information by the decoding means 4.
sent to. The decoding means 4 is composed of a variable length decoding means 5, an inverse quantization means 6, an orthogonal transform decoding means 7, and a motion compensated inter/intra frame decoding means 8, and the main information is first processed by the variable length decoding means. 5 converts the code-assigned information into a decoded and quantized output, which is dequantized by the dequantizing means 6 using the quantization step information of the sub information to create decoded transform coefficients. The orthogonal transform decoding means 7 performs orthogonal transform decoding on the decoded transform coefficients to obtain a decoded block output.

The motion inter/intra frame decoding means 8 has a frame buffer for at least two frames, and uses encoding method determination information among the sub information to determine whether the encoding means of the decoded block output is interframe encoding or not. Detect whether it is inner encoding. If it is intraframe coding, the decoded block output is written as is as a decoded block signal to the specified position of the frame buffer of the current frame, and if it is interframe coding, it is written to the previous frame using the motion information of the sub information. The block signal of the previous frame, which is specified by the motion information in the frame buffer of , and the decoded block output are added together and written as a decoded block signal to the specified position of the frame buffer of the current frame.

At this time, the motion compensated inter/intraframe decoding means 8 uses the error block position information from the error detection and correction means 2 to determine whether the block signal of the previous frame specified by the motion information is an error block or a block that refers to the error block. If this is the case, the current frame is not written to the frame buffer.

In this way, one frame worth of decoded image frames is formed in the frame buffer in the motion compensated inter/intra frame decoding means 8.

Next, the encoder interpolates the thinned frames using the motion information of the thinned frames in the sub information. The frame interpolation means 9 calculates a block signal at the position in the frame buffer of the decoded image frame immediately before and after that indicated by the motion information of the thinned frame in the sub information, and writes it into the specified position of the frame buffer of the interpolation frame. Interpolate thinned frames. In addition, in the above, the frame interpolation means 9
uses the error block position information from the error detection and correction means 2, and if the motion information is incorrect, the corresponding block is not processed.

These decoded image frames and interpolated frames are sequentially updated and stored in the image storage means 10. This image storage means 10 is composed of at least two refresh frames inserted periodically and frame buffers corresponding to the number of frames existing between them. In the case of this example,
Assuming that one refresh frame is inserted into every six frames, the image storage means 10 has a frame buffer group 21 (FIG. 3) for at least seven frames. The decoded image frames are sequentially stored in the image storage means 10, and once all the reproduced video information of the frames existing between the refresh frames are decoded and stored, they are sequentially read out and displayed according to the display timing.

When the error detection and correction means 2 detects an uncorrectable error, the error compensation interframe decoding means 2 is informed of the type of error frame made up of erroneous data and the position of the error block.

If the error frame is a coded frame, the error compensation interframe decoding means 12 detects the error frame and position where the error block exists, and notifies the motion detection means 11 of the detected error frame and position. If the error frame is a thinned-out frame, only the position of the error block is notified to the motion detection means 11.

Further, when the error frame is a coded frame, when the later refresh frame 25 (FIG. 3) arrives at the image storage means 10, the motion detection means 11 detects the decoded image frame 24 immediately before the error frame and the later refresh frame 25 (FIG. 3). Motion information near the position where an error block 26 exists between frames 25 is detected and notified to the error compensation interframe decoding means 12. Thereby, the error compensation interframe decoding means 12
interpolates the error block 26 of the error frame 23 from the decoded image frame 24 immediately before the error frame using the error block position and motion information, and records the interpolated block at the corresponding position of the error frame 23 in the image storage means 10. .

Further, when the error frame is a thinned-out frame, when the immediately following decoded image frame 34 (FIG. 3) arrives in the image storage means 10, the motion detection means 11 detects the decoded image frames 32 and 34 before and after the error frame. Motion information near the position where an error block exists between the two frames is detected and notified to the error compensation interframe decoding means 12. Thereby, the error compensation interframe decoding means 12 uses the error block position and motion information to convert the error block 35 of the error frame 33 into the decoded image frames 32.3 immediately before and after the error frame 33.
4, and the interpolated block is stored at the corresponding position of the error frame 33 in the image storage means 10.

By compensating for an error block using another block as described above, even if an uncorrectable error occurs, the reproduced image will not be disturbed.

After that, the error compensation interframe decoding means 12 decodes the interframe coded blocks that referenced the error block after the error frame using the interpolation block,
It is recorded at the corresponding position in the frame buffer group 21 of the image storage means 10. After all the reproduced video signals of frames existing between refresh frames are decoded, according to the display timing.

These are sequentially read out and displayed.

Above, the present invention has been described with reference to an example in which an encoded image signal is transmitted to a decoding device via a transmission path. It goes without saying that similar effects can be obtained with other devices such as decoding.

(Effects of the Invention) As is clear from the detailed explanation above, the video decoding device of the present invention detects the existence of erroneous image information when an uncorrectable error is detected during transmission of band-compressed image information. This method calculates motion information between the correct decoded image frame immediately before the frame to be processed and the correct decoded image frame that arrives after that, and uses that motion information to interpolate incorrect image information using other information. Since this is a moving image decoding device that does not disturb the reproduced image even if an unavoidable error occurs, it can be used for moving image transmission in the recent information society and contribute greatly.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the configuration of an embodiment of the present invention, FIGS. 2 and 3 are block diagrams of an image storage means of the main part of the embodiment of the present invention, and FIG. 4 is a conventional moving image code. FIG. 5 is a block diagram of a conventional moving image decoding apparatus. 1...Communication path decoding means, 2...Error detection and correction means, 3...Separation means, 4...Decoding means,
5... Variable length decoding means, 6... Inverse quantization means,
7... Orthogonal transform decoding means, 8... Motion compensated inter/intra frame decoding means, 9... Frame interpolation means, 10... Image storage means, 11... Motion detection means, 12 ...Error compensation interframe decoding means. Patent applicant Matsushita Electric Industrial Co., Ltd.

Claims (1)

[Claims] an error detection and correction means for detecting and correcting errors in band-compressed image information; a separation means for separating the output error-corrected image information into main information and sub information; a decoding means for decoding a decoded image frame to generate a decoded image frame; an image storage means for storing a plurality of output decoded image frames; and between the adjacent decoded image frames, the adjacent decoded image frames and the sub information are stored. frame interpolation means for interpolating frames using a frame interpolation means; motion detection means for detecting motion information between the correct decoded image frames immediately before and after an error frame containing uncorrectable error pixels;
A moving picture decoding device comprising: error compensation interframe decoding means for interpolating uncorrectable error pixels existing in the error frame using the motion information.