CN103384329A - CAVLC decoding method based on no code table lookup - Google Patents

Abstract

The invention relates to a CAVLC decoding method based on no code table search, which includes selecting a variable length code table corresponding to the Coeff-token element according to the value of NC; reading the input code stream, and calculating the number of continuous prefix 0 in the input code stream Number, the number is used as the first level of the program jump statement entry; the corresponding codeword length is obtained by the number of codeword prefix 0, and determined according to the relationship between the codeword length and the number of codeword prefix 0 The length of the codeword suffix; read the codeword suffix from the input code stream according to the length of the codeword suffix; interpret the value of the codeword suffix, and use the value of the codeword suffix as a program jump statement The second level of entry; according to the execution of the program at the first level and the second level of the jump statement, the corresponding decoding codeword can be obtained. Its beneficial effects are: the present invention can greatly save time for searching code tables, locating code words, judging and processing, and improves standard CAVLC decoding performance.

Description

A CAVLC Decoding Method Based on No Code Table Lookup

technical field

The present invention relates to the technical field of video image encoding and decoding, in particular to a CAVLC decoding method based on no code table lookup. the

Background technique

H.264/AVC is an international standard for video compression coding jointly formulated by ITU-VCEG (Video Coding Experts Group) and ISO MPEG (Moving Picture Experts Group). Because of its high coding and compression efficiency, friendly network-oriented interface and high-quality image transmission at a lower bandwidth, it has been widely used in the field of digital video communication and storage, and is generally considered to be the most influential powerful industry standards. CAVLC (Context-Adaptive Variable Length Coding) is the most common encoding method of entropy encoding in the H.264/AVC standard, because it can perform dynamic encoding in multiple sets of code tables with different structures according to different video streams. Switching can greatly improve video encoding efficiency and video quality. However, while CAVLC improves video encoding efficiency and video quality, it greatly increases the power consumption required for encoding and decoding video due to its complex calculations. The high performance of H.264/AVC is at the cost of increasing its encoding and decoding power consumption. of. the

Code table lookup is a very power consuming operation. Since the standard CAVLC decoding table look-up method uses a continuous code table lookup (TLSS) method until the expected codeword is found, it consumes a lot of video decoder power consumption. According to statistics, the code table lookup consumes more than 60% of the memory power consumption and about 70% of the decoding time of the video decoder, which is a major factor that seriously affects the decoding performance of CAVLC, and is also a hindrance to the development of H.264/AVC standard technology. important bottleneck. In the CAVLC decoding process of the H.264/AVC standard, since the CAVLC decoding process needs to involve the decoding of five elements (Coeff_token, T1s, Level, Run_before and Total_zeros), decoding the three elements of Coeff_token, Run_before and Total_zeros requires a large number of The variable length code table decoding, therefore consumes a lot of power consumption and time of the H.264/AVC video decoder. Therefore, how to reduce the memory consumption and reduce the time of searching the variable-length code table in CAVLC decoding has become a major issue that our current research focuses on. the

Contents of the invention

The main purpose of the present invention is to, aiming at the deficiencies in the above-mentioned prior art, provide a kind of CAVLC decoding method based on no code table search, the method adopts the mode of program code to replace the code table continuous search method in the H.264/AVC standard , which saves the memory consumption caused by code table lookup, and improves the performance of CAVLC decoding in H.264/AVC. the

The technical solution adopted by the present invention to solve the problems of the prior art is: a kind of CAVLC decoding method based on no code table search, it comprises the following steps:

S1. According to the value of NC, select the variable length code table corresponding to the Coeff-token element;

S2, read the input code stream, calculate the number of continuous prefix 0 in the input code stream, and use this number as the first level of the program jump statement entry;

S3, obtain the corresponding codeword length by the number of the codeword prefix 0, determine the length of the codeword suffix according to the relationship between the codeword length and the number of the codeword prefix 0;

S4. Read the codeword suffix from the input code stream according to the length of the codeword suffix;

S5. Interpret the value of the codeword suffix, and use the value of the codeword suffix as the second level of the program jump statement entry;

S6. According to the execution of the program at the first level and the second level of the jump statement, the corresponding decoded codeword can be obtained.

The above technical solution is further elaborated below: Further, in the step S3, when there are two possible values for the codeword length corresponding to the number of the codeword prefix 0, it is necessary to judge whether the first digit of the codeword suffix is is "0";

If the first bit of the codeword suffix is "0", then determine that the length of the codeword is one of two possible values, and then continue to step S4 and subsequent steps;

If the first bit of the codeword suffix is "1", then search for the corresponding decoded codeword without jumping.

Further, in the step S5, when the value of the codeword suffix is 0, then search for the codeword corresponding to the value 0 of the codeword suffix, and when the value of the codeword suffix is 1, then search for the codeword corresponding to the value of the codeword suffix The codeword corresponding to the value 1 of the codeword suffix. the

Further, the number of continuous prefix 0 in the input code stream is n, the mark of the selection jump statement of the first level is Case n, then in step 1, the program jumps to the selection jump statement Case of the first level Execute at n. the

The beneficial effect of the present invention is: a kind of CAVLC decoding method based on non-code table search provided by the present invention adopts the method based on program code instead of code table search, realizes standard CAVLC decoding table look-up operation, can greatly save the code table Lookup, code word positioning, judgment and processing time, overcome the disadvantages of the TLSS method in the current standard CAVLC decoding that the table lookup takes up a lot of memory, takes a long time to look up the table, and occupies a large storage space of the code table, greatly improving the standard CAVLC decoding performance. the

Description of drawings

Fig. 1 is a flow chart of the method of the present invention;

Fig. 2 is that the method of the present invention and other several commonly used methods compare Fig. 1 in memory consumption of code table lookup when different image sequences;

Fig. 3 is that the method of the present invention and other several commonly used methods compare Fig. 2 of memory consumption of code table lookup when testing different image sequences;

Fig. 4 is that the inventive method and several other commonly used common methods consume code table lookup time comparison figure when decoding different image sequences;

Fig. 5 is a comparison chart of code table storage space consumption between the method of the present invention and the standard TLSS method. the

The realization of the purpose of the present invention, functional characteristics and advantages will be further described in conjunction with the embodiments and with reference to the accompanying drawings. the

Detailed ways

The technical solution of the present invention will be described in detail below in conjunction with the accompanying drawings and specific embodiments, so as to understand the essence of the present invention more clearly and intuitively. the

The present invention mainly aims at the problem that the continuous code table lookup method (TLSS) in the current standard CAVLC decoding takes up a large amount of memory, takes a long time to look up the table, and consumes a large storage space of the code table when decoding the table lookup, and provides a search method based on no code table The CAVLC decoding method uses program code execution to realize the code table lookup, improves the standard CAVLC decoding code table lookup process, and expands the existing CAVLC decoding table lookup technology to improve the decoding of CAVLC decoding on the H.264/AVC data stream ability. the

Referring to Figure 1, the method specifically includes the following steps:

S1. According to the value of NC, select the variable-length code table corresponding to the Coeff-token element; in this step, the Coeff_token element corresponds to three variable-length code tables, and different variable-length codes need to be selected according to different NC values surface.

S2, read the input code stream, calculate the number of continuous prefix 0 in the input code stream, and use this number as the first level of the entry of the program jump statement, for example, the number of continuous prefix 0 in the input code stream is n, the label of the first-level selective jump statement is Case n, then in step 1, the program jumps to the first-level selective jump statement Case n for execution. the

S3, obtain the corresponding codeword length by the number of the codeword prefix 0, determine the length of the codeword suffix according to the relationship between the codeword length and the number of the codeword prefix 0; Specifically,

When the codeword length corresponding to the number of the codeword prefix 0 has two possible values, it is necessary to judge whether the first bit of the codeword suffix is "0";

If the first bit of the codeword suffix is "0", then determine that the length of the codeword is one of two possible values, and then continue to step S4 and subsequent steps;

If the first bit of the codeword suffix is "1", then search for the corresponding decoded codeword without jumping.

S4. Read the codeword suffix from the input code stream according to the length of the codeword suffix;

S5. Interpret the value of the codeword suffix, and use the value of the codeword suffix as the second level of the program jump statement entry; in this step, the value of the codeword suffix may be 0 or 1, when the value of the codeword suffix When the value of the suffix is 0, the code word corresponding to the value 0 of the code word suffix is searched, and when the value of the code word suffix is 1, the code word corresponding to the value 1 of the code word suffix is searched.

S6. According to the execution of the program at the first level and the second level of the jump statement, the corresponding decoded codeword can be obtained. the

Assuming that the NC value is 0≤NC<2, the decoded input code stream is 0000100011..., taking the decoded Coeff_token as an example, the method of the present invention can be completed by the following steps:

Step 1: According to the value of NC 0≤NC<2, select the variable length code table 0 of Coeff_ token as the program entry;

The second step: read the input code stream, calculate the number of continuous prefix 0 of the code word in the input code stream 0000100011 to be 4, and the program jumps to the first-level selection jump statement Case 4 to execute;

The third step: according to the corresponding relationship between the number 4 of the code word prefix 0 and the code length, the code word length can be obtained as 6 or 7; because the code word length corresponding to the number 4 of the code word prefix 0 has two possible values 6 or 7 , by further judging that the first bit of the codeword suffix is 0, the codeword length corresponding to the number 4 of the codeword prefix 0 can be obtained as 7, and then the codeword suffix length can be obtained according to the relationship between the number of codeword prefix 0 and the code length is 2;

Step 4: Read the 2-bit codeword suffix value from the code stream according to the length 2 of the codeword suffix, and judge whether its value is 0 as the second level of program judgment jump statement entry;

The corresponding decoding code word 0x65 can be found by executing the above program at the first and second levels of the jump statement.

Part of its program code is as follows:

Case 4: //The number of codeword prefix 0. ---The first level judgment jump statement program

Codesuff=1;//The first digit of codeword suffix

codesuffword=ShowBits(buf, frame-bitoffset, BitstreamLengthInBits, 1); // the first digit of the codeword suffix

if(codesuffword)//The value of the first digit of codeword suffix is 1

cod=0x64;//code word

else

{

codesuff=2;//code word suffix length is 2;

codesuffword = ShowBits(buf, frame_bitoffset, BitstreamLengthInBits, 2);

if(codesuffword=00)//judging whether the code word suffix value is 0, -- the second level judges the jump statement program

cod=0x64;//code word

else if(codesuffword=01)

cod=0x43;//code word

else

cod="decoding error"

}

The decoding method of other Run_before and Total_zeros elements is the same as that of Coeff_token, and will not be repeated here.

Fig. 2 and Fig. 3 are the memory comparison diagrams of the code table lookup consumption when testing the image sequences of different parameters between the method of the present invention and several other commonly used common methods respectively; wherein Fig. 2 is under the image sequence of setting Frame=60, QP=28 The comparison chart of the memory consumption of the measured code table, Figure 3 is the comparison chart of the memory consumption of the code table measured under the image sequence with Frame=120 and QP=32. It can be seen from Figure 2 and Figure 3 that in terms of saving the memory consumption of code table lookup, the method we proposed is higher than all other methods, and can achieve 100% saving of code table search memory occupation and consumption. The main reason for saving is that our proposed method uses program code to replace the code table continuous search method in the H.264/AVC standard, which can save the memory consumption caused by the code table search and improve the H.264/AVC standard. CAVLC decoding performance in AVC. It can also be seen from Figure 2 and Figure 3 that the Kim method can achieve better memory saving characteristics than the TLBS and Moon methods. The TLBS method occupies a large memory saving advantage at a large QP value, and Moon can achieve a relatively good memory saving effect at a low QP value. the

Fig. 4 is a comparison diagram of code table lookup time consumed by the method of the present invention and several other commonly used common methods when decoding different image sequences. As can be seen from Fig. 5, compared with the standard TLSS method, the method of the present invention can improve the code table lookup speed by about 45%, which is higher than other commonly used table lookup algorithms. The main reason for the analysis is that our proposed method uses program codes to realize code table lookup, which can greatly save time for code table lookup, code word positioning, judgment and processing. the

Fig. 5 is a comparison chart of code table storage space consumption between the method of the present invention and the standard TLSS method. As can be seen from Fig. 5, compared with the standard TLSS method, the inventive method can save 2320 bytes of code table storage space, saving the main reason is that our proposed method adopts the program code mode to replace the code table, which can save the code table Storage space, which improves the storage space utilization of embedded devices with limited storage resources in mobile environments with small areas and low power consumption. the

In summary, a kind of CAVLC decoding method based on no code table search provided by the present invention adopts a program code-based method instead of the code table search method to realize the standard CAVLC decoding table lookup operation, which can greatly save the search of the code table. The positioning, judgment and processing time of the code word overcomes the disadvantages of the TLSS method in the current standard CAVLC decoding that the table lookup takes up a lot of memory, takes a long time to look up the table, and occupies a large storage space of the code table, greatly improving the standard CAVLC decoding. performance. the

The above is only a preferred embodiment of the present invention, and does not limit the scope of its patents. Any equivalent structure or equivalent process transformation made by using the description of the present invention and the contents of the accompanying drawings is directly or indirectly used in other related technical fields. All are included in the scope of patent protection of the present invention in the same way. the

Claims (4)

1. a kind of CAVLC decoding method based on no code table, is characterized in that: it comprises the following steps S1. According to the value of NC, select the variable length code table corresponding to the Coeff-token element; S2, read the input code stream, calculate the number of continuous prefix 0 in the input code stream, and use this number as the first level of the program jump statement entry; S3, obtain the corresponding codeword length by the number of the codeword prefix 0, determine the length of the codeword suffix according to the relationship between the codeword length and the number of the codeword prefix 0; S4. Read the codeword suffix from the input code stream according to the length of the codeword suffix; S5. Interpret the value of the codeword suffix, and use the value of the codeword suffix as the second level of the program jump statement entry; S6. According to the execution of the program at the first level and the second level of the jump statement, the corresponding decoded codeword can be obtained. 2. search CAVLC decoding method based on no code table according to claim 1, it is characterized in that: in described step S3, when the code word length corresponding to the number of described code word prefix 0 has two possible values, need Determine whether the first digit of the codeword suffix is "0"; If the first bit of the codeword suffix is "0", then determine that the length of the codeword is one of two possible values, and then continue to step S4 and subsequent steps; If the first bit of the codeword suffix is "1", then search for the corresponding decoded codeword without jumping. 3. search CAVLC decoding method based on no code table according to claim 1, it is characterized in that: in described step S5, when the value of described code word suffix is 0, then search and the value 0 correspondence of described code word suffix codeword, when the value of the codeword suffix is 1, then search for the codeword corresponding to the value 1 of the codeword suffix. 4. search CAVLC decoding method based on no code table according to claim 1, it is characterized in that: the number of continuous prefix 0 is n in the described input code stream, the mark of the selection jump sentence of the first level is Case n, Then in step 1, the program jumps to the first-level selection jump statement Case n for execution.

Images