KR0123083B1 - Packet data structure for packet full image encoder and decoder - Google Patents

Abstract

The present invention relates to a method of efficiently decoding with respect to packet loss in packet moving picture coding & decoding system. According to this method, when transmitting to a B-ISDN cell with N byte in packet length, header data, start pointer data, absolute address data, reserved data, and invalid bit data are inserted into the cell, and data to be practically transmitted to the K byte.

Description

Packet Data Structure in Packet Video Encoder and Decoder

The present invention relates to packet transmission in a packet video encoder and a decoder, and more particularly, to a packet data structure suitable for efficiently suppressing packet loss generated when transmitting packet video encoded data in a packet form having a predetermined length. will be.

Recently, thanks to the rapid development of digital video signal processing technology, digital video compression technology has been made available for applications in various telecommunication fields such as video conferencing, digital broadcasting codec, and video reduction. In particular, the signal band compression method of the digital VCR, which has been developed over the last several years, has been mainly mainstream of the coding method in an intra frame without considering the redundancy between images. However, the data compression technique using the intra frame coding method does not remove redundancy outside the frame, and thus has a disadvantage in that the compression efficiency of data is not high.

Therefore, in recent years, an attempt has been made to apply an inter frame coding scheme, which is a video compression scheme that removes redundancy in the time axis direction, to data compression of a digital VCR. Since the coding method between frames can remove redundancy between images, data compression efficiency is higher than that of intra-frame coding method.

However, despite the advantages of the inter-frame coding method, the inter-frame coding method has not been applied to digital VCRs until now, due to the characteristics of the magnetic recording medium called tape. In other words, when data is recorded on a magnetic recording medium such as a tape, data must be fixed, because data is difficult to be fixed by an inter-frame coding method.

Meanwhile, a typical encoder for interframe coding, which includes an intra frame coding scheme and solves a problem of data fixation, is an encoder of MPEG (Motion Picture Experts Group) scheme.

1 shows an example of a configuration of a general packet video encoder. The operation is briefly described as follows.

Referring to FIG. 1, the first subtractor 10 subtracts an input source image and image signal data output from the motion compensator 28 (a prediction frame obtained by motion estimation and compensation using previous frames between current frames). The difference image (error signal or error signal) obtained through the signal is applied to a discrete cosine transform unit 12.

In addition, the discrete cosine transform unit 12 converts the difference image frame output from the first subtractor 10 into a DCT transform coefficient expressed in a frequency domain in order to compress the video signal using the correlation present in the frame. .

On the other hand, the quantization unit 14 quantizes the video signal, that is, the DCT transform coefficient, which is energy-compressed by the discrete cosine transform unit 12 to a predetermined quantization level, and performs quantization by performing run-length encoding to quantize the DCT transform coefficient. The quantized DCT transform coefficients are provided to a variable length coding (VLC) unit 16 and an inverse quantizer 20, respectively.

The variable length coding (VLC) unit 16 quantizes the quantization unit 14, and compresses the variable length coded video signal (quantized DCT transform coefficient) by variable length coding.

That is, in the variable length encoder 16, data having a high frequency is represented by a few bits among the signals (0 to 256 levels) represented by 8 bits, and data having a low frequency is represented by many bits. The total number of bits can be reduced.

On the other hand, the packetizer 18 packetizes the compressed data provided by the variable length encoder 16 into a 53-byte packet having a predetermined length, for example, 5 bytes of header information and 48 bytes of data information. The packetized data here is forwarded to a transmitter not shown for transmission.

On the other hand, if the quantized data is input from the above-described quantization unit 14, the inverse quantization unit 20 restores the data before quantization (that is, the DCT transform coefficient), and the inverse discrete cosine transform unit 22 When the restored data is input from the quantization unit 20, the restored data is restored to the data before the discrete cosine transform and provided to the adder 22.

Next, the adder 24 adds the reconstructed data (ie, the difference image) provided by the inverse discrete cosine transform unit 22 and the predicted frame data provided by the motion compensator 28 to be described later to add the reconstructed previous frame. And the restored previous frame signal generated here is stored in the recovery memory 26.

On the other hand, in the motion estimation unit 30, for example, using a technique such as a block matching algorithm, the current video signal (frame) provided from the input side and the restored previous video signal (frame) provided from the recovery memory 26. The motion vectors of the liver are detected as N × N unit blocks (eg, 8 × 8 blocks, 16 × 16 blocks, etc.), and the detected motion vectors are provided to the motion compensation unit 28.

Although not shown in FIG. 1, the motion vectors detected by the motion motion estimation unit 30 are provided to the variable length encoder 16 for transmission to the receiving side.

In addition, the motion compensator 28 generates a predictive frame signal through motion compensation based on the motion vectors provided by the motion estimator 30 for the restored previous frame provided by the recovery memory 26. The predicted frame signal generated here is provided to the above-described subtractor 10 and adder 24, respectively.

2 shows an example of the configuration of a general packet video decoder. The operation is briefly described as follows.

Referring to FIG. 2, the depacketizer 40 restores a signal before packetization from the packet video encoder of FIG. 1 to the demultiplexer 42, and the demultiplexer 42 depacketizes the packet. By demultiplexing the video signal data, the mode data, and the vector data from the signal output from the video unit 42, the video signal data is a variable length decoding (VLD) unit 44, and the mode data and the vector data are stored in a frame. / Approve to 48 respectively.

At this time, the variable length decoding unit 44 and the inverse quantization unit 46 perform a variable length decoding, inverse quantization, and inverse discrete cosine transformation on the variable length coded image signal data output from the demultiplexer 42. The signal is reconstructed, and the reconstructed image signal (difference image) is provided to the adder 50 of the next stage.

On the other hand, the frame store and predictor (48) is provided with an image signal (i.e., image frame data) output from the adder 50 and stored as a previous frame signal, which is provided by the demultiplexer 42. A predictive frame signal is generated through motion compensation using data, motion vector data and previously stored previous frame signals, wherein the generated predicted frame signals (ie, reconstructed previous frame signals) are provided to adder 50.

In addition, the adder 50 adds the difference image signal output from the variable length decoder 44 and the reconstructed previous frame signal output from the frame storage / estimation 48, thereby reconstructing the image frame restored to the original signal before encoding. Generate a signal.

That is, conventionally, a video encoder compresses and encodes video data at a predetermined compression rate, packetizes the packet data into packet data having a predetermined length, and transmits the packet data. Restoring

In this case, data loss (or error) in the packet data may occur due to interference with a transmission path or an external signal during packet transmission of data through a transmission channel. When decoding the original video signal by collecting the generated continuous packet data, there is a problem that the original video signal cannot be faithfully restored.

Accordingly, an object of the present invention is to provide a packet data structure that can efficiently cope with packet data loss that has been packet-encoded using additional information in a packet video encoder and a decoder. .

In order to achieve the above object, the present invention compresses and encodes a predetermined bitrate through a packet video encoder and packetizes the packet into packets having a predetermined length, and transmits the packet encoded data to the packet video decoder through a transport channel. In a packet data structure capable of coping with packet loss occurring, B-coded data is packet-encoded through the packet video encoder, wherein one packet length is N bytes, the header information is Nn bytes, and the data area length is M bytes. -When transmitting to the cell for the ISDN, P bytes in the M-byte data area following the header information are allocated to the additional information area corresponding to the transmission error of the packet encoded data, and 9-bit start pointer information is provided in the additional information area. 11 bits of absolute address information, 7 bits of reservation information, 5 bits of incremental bit information Provides packet data structure in the packet video encoder and decoder, characterized in that the transmission by inserting a compression-encoded data to be inserted, and transmits the remaining 44 bytes of the data area, respectively.

1 is a block diagram showing the configuration of a general packet video encoder.

2 is a block diagram showing the configuration of a general packet video decoder.

3 is a block diagram of a transmission cell showing a packet data structure in a packet video encoder and a decoder according to a preferred embodiment of the present invention.

* Explanation of symbols for main parts of the drawings

10: adder 12: discrete cosine conversion unit

14 quantizer 16 variable length encoder

18: packetization unit 20,46: inverse quantization unit

22: inverse discrete cosine conversion unit 26: recovery memory

28: motion compensation unit 30: motion estimation unit

40: reverse packetization unit 42: demultiplexer

44: variable length decoder 48: frame storage / estimation

24,50: Adder

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

The video signal compression-coded through the packet video encoder of FIG. 1 is packetized into packet data having a predetermined length. In this case, the video signal data encoded by the packet compression encoding is, for example, B-ISDN for a packet length of 53 bytes. Since it has 5 bytes of header information when loaded in a cell, the data actually transmitted is 48 bytes, that is, 384 bits.

In this case, when the variable length coded data, that is, 384 bits are inserted into 48 bytes of the B-ISDN cell, the variable length coded value of the last part is assumed to be inserted to span the current cell and the next cell. Assuming that macro block addressing is 22, the variable length code corresponding thereto is '0000 0100011' according to a code book. If this is inserted at the end of the cell and only up to '00001' is inserted and the remaining '00011' is inserted into the next cell, if the transmission loss occurs in the next cell, the data after '00001' is lost when the decoder decodes it, so it is lost in the decoder. An error occurs when the variable length information is decoded by the variable length, thereby causing a problem that image quality is degraded in the reconstructed reproduced video.

Therefore, in the present invention, to prevent such a phenomenon (degradation of the playback image caused by a data error due to transmission loss), the decoder can faithfully restore the original signal even if a loss occurs in the transmission channel. In the cell, as shown in FIG. 3 as an example, new information (additional information) is inserted, i.e., 4 bytes of 48 bytes allocated to the data area are allocated to the additional information area, and predetermined in each of the allocated additional information areas. Insert a lot of additional information with bits.

That is, according to the present invention, in a B-ISDN cell having a packet length of 53 bytes, 4 bytes following 5 bytes of header information include 9 bits of start pointer information, 11 bits of absolute address information, and 7 bits of reservation information. 5 bits of the insignificant bit information are respectively inserted, and data to be substantially transmitted is inserted into the remaining 44 bytes.

Here, the start point is a macroblock address in a frame, the invalid bit information is an address based on the back of the packet of a non-belidated bit in one packet, and the absolute address indicates a macroblock start in the packet. Note is an address.

Accordingly, even if a transmission loss occurs during packet data transmission, the variable length decoder of the receiving side decoder can effectively restore the error without using an additional information inserted in 4 bytes of the valid data interval in the cell.

As described above, according to the present invention, in the packet video encoder and the decoder, start pointer information, absolute address information, reservation information, and incremental bit information in addition to header information in a transmission cell for efficient data recovery in response to packet loss of data. By inserting the video data into a packet, the packetized video can be efficiently decoded without errors in a system having packet loss generated during transmission of packet video encoded data.

Claims (1)

The packet video coder compresses and encodes a packet having a predetermined bit rate by using a packet video encoder, and packetizes the packet coded data into packets having a predetermined length. In the structure, when the packet-coded data is transmitted to the B-ISDN cell having one packet length of N bytes, its header information of Nn bytes, and its data area length of M bytes, the header information is transmitted. P bytes in the M-byte data area subsequent to the sub-information area corresponding to the transmission error of the packet-encoded data are allocated. Insert reservation information and 5-bit invalid bit information, respectively, Packet data structure in the packet video encoder and decoder, characterized in that the transmission by inserting a compression-encoded data to be transmitted in 44 byte.