Method for reducing computational complexity of video compression standard

Abstract

A method for reducing computational complexity of video compression standard is provided, and it includes an intra 4×4 macroblock (I4MB) search algorithm, an intra 16×16 macroblock (I16MB) search algorithm and a chroma search algorithm. The I4MB search algorithm and I16MB search algorithm accelerate the prediction process of the luma macroblock, and the chroma search algorithm accelerates the prediction process of chroma macroblock. The above algorithms can greatly reduce the computation of prediction mode of video compression standard.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

The accompanying drawings are included to provide a further understanding of the invention, and are incorporated in and constitute a part of this specification. The drawings illustrate embodiments of the invention and, together with the description, serve to explain the principles of the invention. In the drawings,

FIG. 1 a illustrates a condition-correlation search mode of the intra 4×4 macroblock search algorithm of the preferred embodiment of the present invention.

FIG. 1 b illustrates a half-full search mode of the condition-correlation search mode in FIG. 1a;

FIG. 1 c illustrates a context-correlation search mode of the condition-correlation search mode in FIG. 1a;

FIG. 2 illustrates an intra 4×4 macroblock mode determination table of the preferred embodiment of the present invention;

FIG. 3 a illustrates a luma 16×16 macroblock of the intra 16×16 macroblock search algorithm of the preferred embodiment of the present invention;

FIG. 3 b illustrates a 4×4 block of the intra 16×16 macroblock search algorithm of the preferred embodiment of the present invention;

FIG. 4 a illustrates a U macroblock of the chroma macroblock of the preferred embodiment of the present invention; and

FIG. 4 b illustrates a V macroblock of the chroma macroblock of the preferred embodiment of the present invention.

DESCRIPTION OF THE PREFERRED EMBODIMENTS

Reference will now be made in detail to the present preferred embodiments of the invention, examples of which are illustrated in the accompanying drawings. Wherever possible, the same reference numbers are used in the drawings and the description to refer to the same or like parts.

While the specification concludes with claims defining the features of the invention that are regarded as novel, it is believed that the invention will be better understood from a consideration of the following description in conjunction with the figures, in which like reference numerals are carried forward.

In H.264/AVC video compression standard, the luminance prediction mode can be divided into two different intra macroblock conditions, intra 4×4 macroblock (I4MB) and intra 16×16 macroblock (I16MB).

There are nine prediction modes in the I4MB macroblock: mode 0 (vertical), mode 1 (horizontal), mode 2 (DC), mode 3 (diagonal down-left), mode 4 (diagonal down-right), mode 5 (vertical-right), mode 6 (horizontal-down), mode 7 (vertical-left) and mode 8 (horizontal-up). Condition-correlation search table 11 is built by observing availableness of the upper and left blocks. According to the condition-correlation search table 11, the prediction mode can selectively predict some modes to reduce the computation.

Reference is made to FIG. 1a, which illustrates a condition-correlation search mode of the intra 4×4 macroblock search algorithm of the preferred embodiment of the present invention.

There are four different conditions in the condition-correlation search table in FIG. 1a: in the first condition, only mode 2 is performed when the left and upper blocks are both non available.

In the second condition, mode 1, mode 2 and mode 8 are performed when the left block is available but the upper block is non available.

In the third condition, mode 0, mode 2, mode 3 and mode 7 are performed when the left block is non available but the upper block is available.

In the fourth condition, modes of half-full search and context-correlation search are performed when the left and upper blocks are both available.

Reference is made to FIG. 1b, which illustrates a half-full search mode of the condition-correlation search mode.

The half-full search table is divided into two different conditions. In the first condition, only mode 0, mode 1, mode 2, mode 3 and mode 4 are performed when the left or upper block is mode 2 (DC) because the other modes possess their own spatial direction in the nine block prediction modes of I4MB except mode 2 (DC). Therefore, mode 2 can be separated from the nine block prediction modes of I4MB. In natural images, neighboring pixels change slightly such that each neighboring blocks are almost the same. Therefore this approach uses the spatial correlation between the blocks to select some possible block prediction modes to perform and simplify the computation. No matter what kind of prediction modes the neighboring block is, mode 2 is always possible to be the best prediction mode. If mode 2 happens in the neighboring blocks, all kinds of prediction modes are possible to be the best prediction modes because mode 2 does not possess any obvious spatial direction. Thus, all kinds of modes will be performed when mode 2 occurs in neighboring blocks. In order to reduce the computation complexity, the mode having higher occurrence probability is selected to substitute all previous modes. Therefore, the approach, wherein only mode 0, mode 1, mode 2, mode 3 and mode 4 are performed, is called as half-full search.

Reference is made to FIG. 1c, which illustrates a context-correlation search mode of the condition-correlation search mode.

In the second condition of the half-full search mode, when neither of the left and upper neighboring blocks is mode 2, the context-correlation search mode is performed. The prediction approach of context-correlation search mode is described as following: under the assumed condition when the prediction modes of two neighboring blocks are almost the same. In addition to original prediction modes of the upper and left blocks, other prediction modes similar to the upper and left blocks in spatial direction are also selected as the basis for prediction modes. In the motion vector and the context-correlation search table of FIG. 1c, the upper and left prediction modes are mode 6 and mode 7. Based on aforementioned approach, the prediction modes of mode 3, mode 7, mode 0 (from mode 7), mode 4, mode 6, mode 1 (from mode 6) and mode 2 (DC) are performed.

According to the above three search prediction modes, the condition correlation search mode, the half-full search mode and the context-correlation search mode, the I4MB mode determination table is provided in FIG. 2.

Reference is made to FIG. 3a and FIG. 3b, wherein FIG. 3a illustrates a luma 16×16 macroblock and FIG. 3b illustrates a 4×4 block of the intra 16×16 macroblock search algorithm of the intra 16×16 macroblock search algorithm of the preferred embodiment of the present invention.

The luma 16×16 macroblock is divided into sixteen 4×4 blocks. The 4×4 block denotes transformed residues that are computing the sum of absolute transformed difference (SATD) in FIG. 3b. The following Formula denotes the amount after accumulating from tr0 to tr15 in the FIG. 3b.

${\mathrm{SATD}}_{4\times 4\mathrm{blk}}=\sum ^{15}{\mathrm{tr}}_i$

The following formula is to accumulate the SATD_4×4blk of sixteen blocks, so the best I16MB prediction mode can be selected through the calculation result.

${\mathrm{COST}}_{l16\mathrm{MB}}=\sum _{4\times 4\mathrm{blk}=0}^{15}{\mathrm{SATD}}_{4\times 4\mathrm{blk}}$

Reference is made to the FIG. 4a and FIG. 4b, wherein FIG. 4a illustrates a U macroblock of the chroma macroblock, and FIG. 4b illustrates a V macroblock of the chroma macroblock.

There are U macroblock and V macroblock with four 4×4 blocks and four different kinds of chroma prediction modes in each color component. The spatial relation between the blocks in chroma prediction mode is not very obvious, and each predictor between the blocks is almost the same. Therefore, the following formula only calculates the SATD of top left block of the 4×4 block in each color component to perform the best prediction mode of chroma macroblock. This approach could greatly save 75% computation.

$\hspace{1em}\begin{array}{c}{\mathrm{COST}}_{\mathrm{Chroma}}={\mathrm{COST}}_u+{\mathrm{COST}}_v\\ ={\mathrm{SATD}}_{4\times 4\mathrm{blk}0}+{\mathrm{SATD}}_{4\times 4\mathrm{blk}0}\end{array}$

According to the composition and the embodiments above, there are many advantages of the present invention, such as:

1. This invention proposes an effective fast method, which can simplify the processes determining the best luma and chroma prediction mode of H.264/AVC video compression standard, and further reduces the computation of system.

2. Compared to prior art, the simplified method of the present invention for luma and chroma prediction can not only speed up the whole system but also reduce system power consumption. Thus, the present invention can be implemented into chip and then applied to every kind of portable device such as multi-media mobile phone.

Although the present invention has been described in considerable detail with reference to certain preferred embodiments thereof, other embodiments are possible. Therefore, their spirit and scope of the appended claims should no be limited to the description of the preferred embodiments contained herein.

It will be apparent to those skilled in the art that various modifications and variations can be made to the structure of the present invention without departing from the scope or spirit of the invention. In view of the foregoing, it is intended that the present invention cover modifications and variations of this invention provided they fall within the scope of the following claims and their equivalents.

Claims

1. A method for reducing computational complexity of video compression standard, at least comprising: providing an intra 4×4 macroblock search algorithm with a luma prediction mode;providing an intra 16×16 macroblock search algorithm with a luma prediction mode, and using the intra 16×16 macroblock search algorithm and the intra 4×4 macroblock search algorithm to accelerate the luma prediction modes of the intra 4×4 macroblock and the intra 16×16 macroblock;providing a chroma macroblock, including a U macroblock and a V macroblock; andusing a chroma search algorithm to accelerate a plurality of chroma prediction modes of the U macroblock and the V macroblock;whereby, the three search algorithms reduce the, computational complexity of luma and chroma prediction modes.

2. The method for reducing the computational complexity of video compression standard of claim 1, wherein the intra 4×4 macroblock search algorithm includes a condition-correlation search mode.

3. The method for reducing the computational complexity of video compression standard of claim 2, wherein the condition-correlation search mode includes a first upper block and a first left block.

4. The method for reducing the computational complexity of video compression standard of claim 3, wherein the intra 4×4 macroblock search algorithm determines a pre-determined prediction mode for computation according to the first upper block and the first left block.

5. The method for reducing the computational complexity of video compression standard of claim 4, wherein the condition-correlation search mode includes a half-full search mode.

6. The method for reducing the computational complexity of video compression standard of claim 5, wherein the condition-correlation search mode includes a context-correlation search mode.

7. The method for reducing the computational complexity of video compression standard of claim 6, wherein the intra 4×4 macroblock search algorithm has multiple prediction modes including a mode 0, a mode 1, a mode 2, a mode 3, a mode 4, a mode 5, a mode 6, a mode 7 and a mode 8.

8. The method for reducing the computational complexity of video compression standard of claim 7, wherein the condition-correlation search mode includes a plurality of search conditions comprising: performing only the mode 2 (DC) when the first upper block and the first left block are both non-available;performing the mode 1, the mode 2 and the mode 8 when the first upper block is non-available and the first left block is available;performing the mode 0, the mode 2, the mode 3 and the mode 7 when only the first upper block is available and the first left block is non-available; andusing the half-full search mode and the context-correlation search mode to perform when the first upper block and the first left block are both available.

9. The method for reducing the computational complexity of video compression standard of claim 8, wherein the half-full search mode includes a second upper block and a second left block.

10. The method for reducing the computational complexity of video compression standard of claim 9, wherein the half-full search mode includes a plurality of search conditions: performing the mode 0, the mode 1, the mode 2, the mode 3 and the mode 4 when the second upper block and the second left block are both mode 2(DC);performing the mode 0, the mode 1, the mode 2, the mode 3 and the mode 4 when the second upper block is mode 2(DC) and the second left block is other prediction mode;performing the mode 0, the mode 1, the mode 2, the mode 3 and the mode 4 when the second upper block is other prediction mode and the second left block is mode 2(DC); andusing the context-correlation search mode to perform when the second upper block and the second left block are both other prediction mode.

11. The method for reducing the computational complexity of video compression standard of claim 10, wherein the context-correlation search mode includes a third upper block and a third left block.

12. The method for reducing the computational complexity of video compression standard of claim 11, comprising: performing at least one of the prediction modes by the context-correlation search mode wherein each of the performed at least one of the prediction modes corresponding to the third upper block and the third left block in spatial directions.

13. The method for reducing the computational complexity of video compression standard of claim 1, wherein the intra 16×16 macroblock search algorithm includes sixteen macroblocks, and each macroblock has sixteen transformed residues.

14. The method for reducing the computational complexity of video compression standard of claim 13, wherein the intra 16×16 macroblock search algorithm comprises: computing a distortion of each macroblock by a SATD(Sum of Absolute Transform Difference);transforming the transformed residues by Hadamard; andsumming up the absolute values of the transformed residues and computing the SATD of the sixteen macroblocks.

15. The method for reducing the computational complexity of video compression standard of claim 1, wherein the U macroblock includes four chroma prediction modes.

16. The method for reducing the computational complexity of video compression standard of claim 15, wherein the four chroma prediction modes are four 4×4 blocks.

17. The method for reducing the computational complexity of video compression standard of claim 16, wherein the U macroblock only computes a distortion of a left upper block as the distortion of the U macroblock.

18. The method for reducing the computational complexity of video compression standard of claim 1, wherein the V macroblock includes four IS chroma prediction modes.

19. The method for reducing the computational complexity of video compression standard of claim 18, wherein the four chroma prediction modes are four 4×4 blocks.

20. The method for reducing the computational complexity of video compression standard of claim 19, wherein the V macroblock only computes a distortion of a left upper block as the distortion of the V macroblock.

21. The method for reducing the computational complexity of video compression standard of claim 1, further comprising: computing a distortion of the U macroblock;computing a distortion of the U macroblock; anddetermining a distortion of the chroma macroblock by summing up the distortion of the U macroblock and the distortion of the V macroblock.