KR950010764B1 - Vld subtract method of position information - Google Patents

Abstract

The method for contracting variable length code according to position information that if a zero length and a discrete cosine transform value are not recorded at a variable length code table, an escape coder, the zero length and the discrete cosine transform coefficient are all added to assign a new code is characterized in that the zero length representing the number of "0" is coded by selectively changing the number of the bits according to the zig-zag scanning sequence position of the right-before-coded variable length code.

Description

Shortening Method of Variable Length Codes According to Position Information

1 is a diagram illustrating a scanning sequence in a zigzag scanning method in an image encoding apparatus.

2 is a block diagram illustrating a survey rectangle of a variable length code according to the present invention.

* Explanation of symbols for main parts of the drawings

101: discrete cosine transform unit 102: quantization unit

103: zig-zag scanning unit 104: variable length code table

105: code allocation unit 106: bit control unit

107 variable length encoder

The present invention relates to a method of shortening a variable length code (VLC) according to position information in a moving picture expert group (MPEG) II. More specifically, a pair input to a VLC table is not described in a VLC table. When an exceptional case occurs, the amount of bits is reduced by selectively changing the number of bits for encoding a zero length according to the Zig-Zag scanning order of the VLC coded immediately before. It relates to a method of shortening the VLC according to the location information.

In general, a Discrete Cosine Transform Quantization Parameter (QP; hereinafter referred to as DCT) is performed in (8 × 8) units in a change encoding scheme for reducing code redundancy using spatial correlation. After quantization is performed in units of a predetermined quantization parameter (QP), zig-zag scanning is performed on each coefficient to improve coding efficiency.

FIG. 1 is a diagram illustrating an order of scanning by a zigzag scanning method in an image encoding apparatus, wherein the VLC code has a valid value other than '0' when scanning each position in the order as shown. A short length code is assigned to the pair with the highest frequency corresponding to successive pairs of '0' lengths and valid values. If the pair input to the VLC table is exceptional in which the VLC table is not described, the following code is used correspondingly. (Escape code; 6 bits) + (zero length, 6 bits) + (valid values; 8 to 16 bits), i.e., an extension code indicating that it is exceptional, followed by the number of '0's 6 bits are allocated to represent all cases in the range of 0 to 63. Finally, for the valid value, 8 bits are allocated when the absolute value is 129 or less, and 16 bits are allocated and more than that. That is, since the length information of '0' is added to the information configured through the zig-zag scanning method, the position information corresponding to each code in the 8 × 8 DCT block has the VLC immediately formed before the corresponding code. Can be obtained by using the position information of the sign.

However, the described method always encodes a certain bit even when an exceptional case occurs. Such a prior art coding method ignores the fact that the 'zero length' part has location information, thus reducing the possibility of reducing the amount of data. There was a problem excluding.

Accordingly, an object of the present invention is to solve the above problems, and an object of the present invention is to select a bit for selectively encoding a zero length according to a zigzag-jag scanning order of a VLC coded immediately before an exceptional case occurs. The present invention provides a VCL shortening method based on location information which increases the efficiency of image compression according to VLC shortening by changing the number.

In the VLC reduction method according to the position information of the present invention for achieving the above object, if the zero length and DCT value obtained by performing quantization and zigzag scanning with DCT are exceptional pairs not listed in the VLC table, ZLC-Zag scanning of the VLC coded just before, using VLC abbreviation method according to the location information that newly adds both the extension code indicating the exception and the zero length indicating the number of zeros and the magnitude of the DCT value. And encoding the number of bits selectively with respect to the zero length indicating the number of zeros according to the order position.

Hereinafter, a preferred embodiment of the VLC reduction method according to the positional information of the present invention will be described in detail with reference to the accompanying drawings.

1 is a block diagram for performing a VLC reduction method according to the positional information of the present invention. The DCT unit 101 compresses and encodes using spatial correlation and the DCT transform coefficients from the DCT unit 101. A quantization unit 102 for approximating a finite number of values, a zigzag-scanning unit 103 for scanning the coefficients of the quantization using a zig-zag scanning method in order to improve coding efficiency, and the jig -When the result (zero length, DCT value) of the zigzag scanning unit 103 is input, the VLC table 104 which assigns a short length code to the pair having the highest frequency corresponding to the zero length and the DCT value, and the jig If the zero length and the DCT transform coefficient value output from the zag scanning unit 103 are exceptional pairs not listed in the VLC table 104, a code assignment unit 105 for newly allocating a VLC code and the code assignment Zero length of the part 105 The bit control unit 106 for changing the number of bits to be coded according to the zig-zag scanning order of the VLC just encoded, and the code or VLC table 104 assigned by the code assignment unit 106. The VLC unit 107 reduces the amount of data to be transmitted by performing VLC on the code assigned by the.

According to the present invention configured as described above, the DCT using the spatial correlation in the DCT unit 101 is performed in units of 8 × 8 blocks, and is quantized in the quantization unit 102, and then zigzag-scanned in the zigzag scanning unit 103. Scanning is done in a ning manner.

At this time, if the zero length and the DCT conversion coefficient values input from the zig-zag scanning unit 103 to the VLC table 104 are pairs described in the VLC table 104, the zero length and DCT conversion in the VLC table 104 are performed. The VLC unit 107 performs a VLC after assigning a short-length code to the pair with the highest frequency corresponding to the pair of coefficient values.

On the other hand, if the zero length and the DCT transform coefficient value input to the VLC table 104 are exceptional pairs not described in the VLC table, then the code assignment unit 105 is as follows with respect to the zero length and the DCT value. The sign is assigned.

That is, they are allocated as (extension code; 6 bits) + (zero length; 6 bits) + (DCT value; 8 to 16 bits). At this time, the maximum 'zero length' that can be obtained at the position where the exceptional case occurs can be easily estimated using the zigzag-jagging scanning sequence position of the previous VLC as follows. Possible zero length = 63- (Zig-zag scanning sequence position of previous VLC)

Accordingly, the sizes of bits required for encoding the 'zero length' size and the 'zero length' according to the range of the zig-zag scanning order positions of the VLC just encoded are as follows.

Thus, even if an exception occurs, the bit control unit 106 changes the number of bits for selectively coding a 'zero length' according to the zig-zag scanning order position of the immediately-encoded VLC. Compared to this, the amount of bits can be reduced more efficiently.

As described above, according to the VLC shortening method according to the positional information of the present invention, as the number of exceptional cases occurs, the number of bits generated as a whole can be greatly reduced compared to the conventional method, thereby improving the coding efficiency.

Claims (1)

If the zero length and discrete cosine transforms obtained by performing discrete cosine transform and quantization and zig-zag scanning are exceptional pairs not listed in the variable-length code table, then the escape code and zero indicate that they are exceptional. In the variable length code shortening method according to the position information of adding a zero length representing a number and a magnitude of a discrete cosine transform coefficient value and newly assigning a code, the zero length representing the number of zeros is a variable length previously encoded. Zig-zag scanning order of code A variable length code shortening method according to position information, characterized by selectively changing the number of bits according to the position.