KR100753282B1 - VLC table selection method for CAVLC decoding and CAVLC decoding method thereof - Google Patents

Abstract

The present invention relates to a method of decoding Context Adaptive Variable Length Coding (CAVLC) for compressing surplus information in H.264 / AVC, which is a video coding standard, and determines a Lev-VLC table using only level_prefix based on the result of analyzing CAVLC. To decode the level. Accordingly, even a very simple process can select the Lev-VLC table to decode the level, improve the complexity of the processing method and improve the processing speed.

CAVLC, H.264 / AVC, Conversion Factors, Levels

Description

VLC table selection method and CAVLC decoding method for CALC decoding {VLC table selection method for CAVLC decoding and CAVLC decoding method}

1 is a diagram illustrating a method of selecting a second VLC table in conventional level decoding;

2 is a flowchart illustrating a decoding method of CAVLC;

3 is a view showing a VLC table used in a CAVLC encoding / decoding method;

4 is a diagram illustrating a structure of a codeword included in a VLC table of FIG. 3;

5 is a flowchart illustrating a level decoding method of CAVLC according to the present invention;

6 is a view showing a method of selecting a second VLC table according to the size of a level;

7 is a view provided for explaining the second and subsequent table selection method of the VLC table selection method of the present invention; and

8 is a flowchart illustrating a VLC table selection method of the present invention.

The present invention relates to a method of encoding / decoding surplus information by Context Adaptive Variable Length Coding (CAVLC) in H.264 / AVC, which is a video coding standard. More specifically, by analyzing CAVLC, a Lev-VLC table is determined by using only level_prefix. A method of selecting a Lev-VLC table for level decoding CAVLC decoding and a method of decoding the CAVLC thereof.

The H.264 / AVC video coding standard uses CAVLC and CABAC (Context Adaptive Binary Arithmetic Coding) to compress the redundant information. Among them, CAVLC is supported in all profiles of H.264 / AVC such as Baseline, Main, Advanced, and FRExt profile, and it has less computation and complexity than CABAC. The CAVLC technique is a technique developed from the variable length coding (VLC) used in the conventional MPEG-1 and MPEG-4.

Hereinafter, a conventional CAVLC level decoding method will be described based on US Pat. No. 6,646,578.

The conventional CAVLC level decoding step includes the number N of nonzero quantization transform coefficients in a 4x4 block compressed with CAVLC, the number of trailing ones 'T1s', the current level calculated with the level_prefix and level_suffix codewords, and the currently used The VLC table is used to select the next level decoding using the VLC table, and the next level is decoded using the selected VLC table.

For example, after the T1s-th level decoding ends, a new VLC table is selected for (T1s + 1) th level decoding.

The VLC table is determined using the decoded level size and the Lev-VLC (n) table used for current level decoding. The selection method is selected using a pseudo code of Equation 1 below, and selects the current table (n = n) or an increased table (n = n + 1) than the current table.

if (| level |> (vlc_inc [n]) & (n ≠ 6)) n = n + 1

 else n = n

And exceptionally, after applying Equation 1 | If level |> 3, the Lev-VLC (2) is used as the VLC table for decoding the T1s-th level. The vlc_inc table used to select the VLC table in Equation 1 is shown in Table 1 below and has a different value depending on n.

n vlc_inc 0 0 One 3 2 6 3 12 4 24 5 48 6 infinite

FIG. 1 is a diagram illustrating a method of selecting a VLC table for (Tis + 1) th level decoding during conventional level decoding. FIG. 1 illustrates a VLC table and a (Tis + 1) layer for Tisth level decoding based on Equation 1. The VLC table for the 9th level decoding is selected.

The VLC table for Tis-th level decoding is determined among Lev-VLC (0) (a1) or Lev-VLC (1) (a2). The VLC table for decoding the (Tis + 1) th level is determined by Equation 1, but when the size of the first level is greater than 3, Lev-VLC (2) (a4) is selected as the second table.

An object of the present invention is to provide a VLC table selection method for CAVLC residual decoding and a CAVLC decoding method in an H.264 / AVC decoder / encoder.

Another object of the present invention is to propose a Lev-VLC table selection method using a level_prefix codeword of a level.

In order to achieve the above object, a method of selecting a VLC table for CAVLC decoding according to the present invention includes selecting a Lev-VLC for level decoding of an i-1 th transform coefficient of a bit stream coded by CAVLC (Context Adaptive Variable Length Coding). (n) determining whether the level_prefix extracted from the table is greater than 2 (wherein the integer i> (T1s + 1) and the integer n is 6≥n≥0) and the i-th conversion factor according to the determination result. Select the Lev-VLC table for level decoding, but if the level_prefix is greater than 2, select the Lev-VLC (n + 1) table; if the level_prefix is not greater than 2 or the n is 6, the Lev-VLC (n) table Maintaining.

Further, if the T1s parsed from the bit stream is equal to 3, setting a predetermined virtual code to 5; if the T1s is less than 3 and the number N of non-zero quantization transform coefficients is 10 or less, the virtual code is 3 Determining the virtual code as 1 when the T1s is less than 3 and N is greater than 10, and the level_prefix extracted from the Lev-VLC (n) table selected for the level decoding of the T1s-th transform coefficient. Determining whether the value is greater than the code and selecting a Lev-VLC table for level decoding of the (T1s + 1) th conversion coefficient. If level_prefix is larger than the virtual code, the Lev-VLC (n + 1) is selected. The method may further include maintaining a Lev-VLC (n) table if it is not larger than the virtual code, but applying 1 instead of 0 if n is 0.

In addition, decoding a token which is a symbol in which the number N of non-zero quantization transform coefficients and the number of trailing ones are combined from the CAVLC coded bit stream according to the present invention, after decoding the token. And decoding whether the signs of the T1s transform coefficients are + or-, respectively, while decoding one of the plurality of Lev-VLC tables to obtain a level of each of the transform coefficients other than the T1s, i-1 Selecting a Lev-VLC table for level decoding of the i-th transform coefficient using level_prefix extracted from the Lev-VLC (n) table selected for level decoding of the first transform coefficient (wherein the integer i> (T1s + 1) And n decodes 6 ≧ n ≧ 0) and TotalZeros, and decodes a run, which is the number of consecutive zero coefficients existing between the levels.

Hereinafter, the present invention will be described in detail with reference to the drawings.

2 is a flowchart illustrating a decoding method of CAVLC.

The Context Adaptive Variable Length Coding (CAVLC) decoding method is used to compress the redundant information in the H.264 / AVC video coding standard. When a video bit stream coded with CAVLC is received, the quantized transform coefficient is decoded according to the CAVLC decoding method.

First, a token (Token), which is a symbol obtained by combining the number N of non-zero quantization transform coefficients in a 4 × 4 block compressed with CAVLC and 'T1s', the number of trailing ones, is decoded (S201).

After decoding the token, the T1s-1th level is decoded from the first, the +/- sign of the trailing ones is decoded (S203), and (N + 1-T1s) level values other than the trailing ones are decoded ( S205). The data compressed by CAVLC is decoded, including the step of decoding TotalZeros (step S207) and the step of decoding a run that is the number of consecutive zero coefficients existing between levels (S209).

In step S205, the method of decoding a level includes both a method of adaptively selecting a VLC table and a method of decoding a level using the selected table.

Hereinafter, a method of decoding the level of the transform coefficient will be described in more detail.

3 is a diagram illustrating a VLC table used in a CAVLC encoding / decoding method, and FIG. 4 is a diagram illustrating a structure of a codeword included in the VLC table of FIG. 3.

Referring to Figure 3, the VLC table is Lev-VLC (0), Lev-VLC (1), Lev-VLC (2), Lev-VLC (3), Lev-VLC (4), Lev-VLC (5) And Lev-VLC (6). That is, the VLC table has Lev-VLC (n) (where 0 ≦ n ≦ 6).

Referring to FIG. 4, the bit stream codewords included in the VLC table indicate the first part b1 indicates level_prefix and the second part b2 indicates level_suffix.

The first half b1 has a structure in which consecutive bits '0' and final bits are '1', and level_prefix has a value equal to the number of consecutive bits '0'.

On the other hand, the second half b2 is the code following the first half b1 and the 'xx... s' is the part. In general, a bit length codeword equal to n of the table Lev-VLC (n), but there is an escape code that is longer. The escape code is used to represent a larger level. In the case of Lev-VLC (0), when level_prefix has a value of 14 or 15, the length of level_suffix has 4 or 12 bits, respectively. However, the tables of Lev-VLC (1) to Lev-VLC (6) have a 12-bit level_suffix codeword when level_prefix is 15.

5 is a flowchart illustrating a level decoding method of CAVLC according to the present invention.

First, a VLC table (hereinafter referred to as a 'first VLC table') for decoding the T1s level is selected by using the number N of non-zero transform coefficients in a block and T1s having a coefficient of a high frequency region of 1. Selection of the first VLC table will be described again below (S501).

Using the first VLC table obtained in step S501, level_prefix is parsed (S505), level_suffix is parsed (S509), and the level of the corresponding transform coefficient is decoded (S511).

If the level decoding has not been completed, the levels of the next transform coefficient are decoded by repeating S503 to S511. However, in this process, the level decoding after the T1s-th conversion coefficient uses a new VLC table, and the selection of the new VLC table is performed immediately at step S507 using the level_prefix parsed at step S505.

For convenience of explanation, the VLC table for decoding of the T1s level is referred to as the 'first VLC table', and the VLC table for decoding of the (T1s + 1) level is referred to as the 'second VLC table'. This is called. Therefore, hereinafter, 'first' and 'second' in the first VLC table and the second VLC table do not mean the order n on the Lev-VLC (n) table including seven in total.

In step S511, the intermediate variable levelCode for obtaining the level of the transform coefficient is calculated using level_prefix, level_suffix and n by the following equation (1).

levelCode = ( level _prefix << n) + level _ suffix

The level of the coefficient is decoded using the following equation (3) from the levelCode. The least significant bit (LSB) of levelCode is a sign, and the remaining bits are equal to the level absolute value minus one.

coefficient level = (1-2 * ( levelCode & 0x 01)) × (( levelCode +2)) >> 1)

The higher the frequency region, the higher the probability that the quantized transform coefficient is. The CAVLC decodes the coefficients in the high frequency region in the order of the low frequency coefficients, and the level table Lev-VLC (n) to be used newly selects each time the level is decoded. Since the Lev-VLC (n) table is suitable for a level at which n is likely to have a large value, the selection of the VLC table proceeds in the direction of increasing n.

Hereinafter, a method of obtaining the first VLC table for obtaining the first level, which is performed in step S501, will be described.

The first VLC table uses the number N of coefficients in the block and T1s having a coefficient of 1 in the high frequency region to estimate the residual amount of surplus information in the block. The first VLC table is represented by Lev-VLC (0) and Choose from Lev-VLC (1).

Here, n means order n of the Lev-VLC (n) tables, which are seven VLC tables.

Hereinafter, a method of selecting a new VLC table in step S507 will be described.

1 is rearranged to the approach direction of the present invention as shown in FIG. 6 is a diagram illustrating a method of selecting a second VLC table according to the size of a level.

Lev-VLC (0) (a) is divided into two cases where T1 = 3 (c1) and when T1 <3 and N≤10 (c2), and the level size is compared to 3 in the second table selection. Only case-by-case table selection is organized into valid rules. In other words, if the level size is 3 or less, Lev-VLC (1) is selected, and if the level size is larger than 3, Lev-VLC (2) is selected.

Table 2 shows the first table selection according to T1 and N together with the codeword and level size.

level_prefix codeword level_suffix codeword Level T1 = 3 (d1) T1 <3, N≤10 (d2) T1 <3, N> 10 (d3)   One - One 2 -   01 - -One -2 -   001 - 2 3 -   0001 - -2 -3 -   0000 1 - 3 4 -   0000 01 - -3 -4 -   0000 001 - 4 5 -   One 0 - - 2   One One - - -2   01 0 - - 3   01 One - - -3   001 0 - - 4

According to Table 2, in Lev-VLC (0), when T1 = 3 (d1), the level size is 3 or less when level_prefix is 5 or less. In Lev-VLC (0), when T1 <3 and N≤10 (d2), the level size is 3 or less when level_prefix is 3 or less.

When T1 <3, N> 10 (d3), in the Lev-VLC (1), the level size is 3 or less when level_prefix is 1 or less.

That is, when selecting the second VLC table, considering the Figure 6 and Table 2, it is possible to select the second VLC table by level_prefix rather than the size of the level.

Table 2 may be summarized as in Equation 5 below using pseudo code T_ prefix .

if (T1 = 3) T_ prefix = 5

elseif (N≤10) T_ prefix = 3

elseif T_ prefix = 1

Here, since T1 always has a value of 3 or less, "elseif (N≤10)" is equal to "elseif (T1 <3 and N≤10).

The second VLC table may be selected as shown in Equation 6 below.

if (n = 0) n = 1

if ( level _ prefix > T_ prefix ) & (n ≠ 6)) n = n + 1

else n = n

Here, n means order n of the Lev-VLC (n) tables, which are seven VLC tables.

In Equation 6, considering that the first VLC table is Lev-VLC (0), let if (n = 0) n = 1.

Hereinafter, the second and subsequent VLC table selection methods will be described with reference to FIG. 7.

FIG. 7 is a diagram provided to explain a second and subsequent table selection method of the VLC table selection method of the present invention.

Referring to FIG. 7, a current VLC table e1, levle_prefix (e4) and level_suffix (e5), and a VLC table e6 to be selected are shown in the corresponding VLC table. For the sake of proper explanation, the level (e3) obtained by using the vlc_inc (e2) of Table 1 and the current VLC table is shown.

Referring to Lev-VLC (n) of FIG. 3 and FIG. 7, in the Lev-VLC (1) to Lev-VLC (5), if the level_prefix is equal to or smaller than '001' (code number 2), the same as the VLC table of the next step is obtained. If the table is selected and level_prefix is greater than '001', the table in which n is increased by 1 is selected for the next VLC table.

Since Lev-VLC 6 does not have Lev-VLC 7, Lev-VLC 6 is selected as the VLC table of the next step even if level_prefix is greater than '001'. Since Lev-VLC (0) is used only for first-level decoding, it is irrelevant since it is determined by Equations 5 and 6 described above.

Therefore, a method of selecting the VLC table in the decoding step of the (T1s + 1) th subsequent level may be defined as in Equation 8 below.

if (( level _ prefix > T_ prefix ) & (n ≠ 6)) n = n + 1

else n = n

Here, T_ prefix = 2 and n denotes the order n of the Lev-VLC (n) tables, which are seven VLC tables.

In conclusion, the table selection method of the present invention comprises two steps of determining the virtual code T_prefix and selecting the next table through T_prefix and case, as shown in FIG. 8.

8 is a flowchart illustrating a VLC table selection method of the present invention.

Only when the second VLC table is selected, there are some selectable T_prefixes (S801 and S803), and when the second and subsequent VLC tables are selected, the T_prefix is 2 (S801 and S805). Equation 7 is then applied using the corresponding T_prefix (S807).

By the above method, the VLC table selection method of the present invention is realized.

While the above has been shown and described with respect to preferred embodiments of the invention, the invention is not limited to the specific embodiments described above, it is usually in the technical field to which the invention belongs without departing from the spirit of the invention claimed in the claims. Various modifications can be made by those skilled in the art, and these modifications should not be individually understood from the technical spirit or prospect of the present invention.

As described above, according to the present invention, in decoding Context Adaptive Variable Length Coding (CAVLC) for compressing redundant information in the video coding standard H.264 / AVC, the VLC table is determined only by level_prefix to decode the level. Can be.

In addition, the process of obtaining a level with level_prefix and level_suffix can be separated from the process of selecting the next VLC table.

Accordingly, the VLC table can be selected and the level can be decoded through a very simple process. The complexity of the processing method is improved and the processing speed is also improved.

This method can be used in all areas where the CAVLC method is used by using the existing CAVLC standard as it is.

Claims (5)

Determining whether or not the level_prefix extracted from the selected Lev-VLC (n) table is greater than 2 for level decoding of the i-1 th transform coefficient of the bit stream coded with Context Adaptive Variable Length Coding (CAVLC). The integer i> (T1s + 1) and the integer n is 6 ≧ n ≧ 0; And According to the determination result, a Lev-VLC table for level decoding of an i-th conversion coefficient is selected, and if the level_prefix is greater than 2, a Lev-VLC (n + 1) table is selected, and the level_prefix is not greater than 2 or the n If 6, maintain the Lev-VLC (n) table; VLC table selection method for CAVLC decoding comprising a. The method of claim 1, If the T1s parsed from the bit stream equal to 3, providing a predetermined virtual code to be compared with the level_prefix and setting the virtual code to 5; Setting the virtual code to 3 when the T1s is less than 3 and the number N of non-zero quantization transform coefficients is 10 or less; Setting the virtual code to 1 when the T1s is less than 3 and the N is greater than 10; Determining whether level_prefix extracted from a Lev-VLC (n) table selected for level decoding of a T1s-th transform coefficient is greater than the virtual code; And According to the determination result, a Lev-VLC table is selected for level decoding of the (T1s + 1) th conversion coefficient. If level_prefix is larger than the virtual code, a Lev-VLC (n + 1) table is selected. If not large, the Lev-VLC (n) table is maintained as it is, if n is 0, 1 instead of 0; Lev-VLC table selection method for CAVLC decoding further comprises. Decoding a token which is a symbol in which the number N of non-quantization quantization transform coefficients and the number of trailing ones are combined from the CAVLC coded bit stream; After decoding the token, decoding whether the symbols of the T1s transform coefficients are each + or-; The level is decoded while selecting one of a plurality of Lev-VLC tables to obtain the level of each of the transform coefficients other than the T1s, and extracted from the selected Lev-VLC (n) table for level decoding of the i-1 th transform coefficient. selecting a Lev-VLC table for level decoding of the i-th transform coefficient using level_prefix, where integer i> (T1s + 1) and integer n is 6 ≧ n ≧ 0; And Decoding TotalZeros and decoding a run that is a number of consecutive zero coefficients existing between the levels. The method of claim 3, wherein Selecting a Lev-VLC table for level decoding of the i-th transform coefficient using the level_prefix, Determining whether level_prefix extracted from the Lev-VLC (n) table selected for level decoding of the i-1 th transform coefficient is greater than 2; And According to the determination result, a Lev-VLC table for level decoding of an i-th conversion coefficient is selected, and if the level_prefix is greater than 2, a Lev-VLC (n + 1) table is selected, and the level_prefix is not greater than 2 or the n Maintaining a Lev-VLC (n) table if the 6; CAVLC decoding method comprising a. The method of claim 3, wherein Selecting a Lev-VLC table for level decoding of the (T1s + 1) th transform coefficient using the level_prefix, When the T1s is equal to 3, providing a predetermined virtual code to be compared with the level_prefix and setting the virtual code to 5; Setting the virtual code to 3 when the T1s is less than 3 and the number N of non-zero quantization transform coefficients is 10 or less; Setting the virtual code to 1 when the T1s is less than 3 and the N is greater than 10; Determining whether level_prefix extracted from a Lev-VLC (n) table selected for level decoding of a T1s-th transform coefficient is greater than the virtual code; And According to the determination result, a Lev-VLC table for level decoding of a T1s + 1 th transform coefficient is selected. If level_prefix is larger than the virtual code, a Lev-VLC (n + 1) Lev-VLC table is selected. If not greater than Lev-VLC (n) maintains the table, if n is 0, the step of applying a 1 instead of 0; CAVLC decoding method further comprising.

Images