CN100558169C - The fast selecting method of inter-frame mode in the video code conversion - Google Patents

Abstract

The present invention proposes the fast selecting method of inter-frame mode in a kind of video code conversion, make full use of the characteristic information of the input video stream of having encoded in the video code conversion, as coefficient mean value, parameter variance, weighting coefficient values, significantly reduced the computation complexity of video code conversion, transcoding efficiency then remains unchanged substantially.

Description

The fast selecting method of inter-frame mode in the video code conversion

Technical field

The invention belongs to method for video coding, be specifically related to video transcoding method, the fast selecting method of inter-frame mode in particularly a kind of video code conversion.

Background technology

Is that the operation of another kind of video data form is called video code conversion with the video data that compressed from a kind of formal transformation.Since video encoding standard of new generation as H.264, AVS all supports multiple inter-frame mode (as H.264 supporting 7 kinds of inter-frame modes such as 16x16,16x8,8x16,8x8,8x4,4x8,4x4, AVS supports 16x16,16x8,4 kinds of inter-frame modes such as 8x16,8x8), and the selection of optimal frames inter mode need be calculated comparison by the rate-distortion optimization of high complexity, therefore in the video code conversion of video encoding standard of new generation, the quick selection of inter-frame mode realizes that for reducing computation complexity real-time video code conversion is significant.

In present existent method, " Zhi Zhou, et al. " Motion information and codingmode reuse for MPEG-2 to is transcoding H.264 ", IEEE InternationalSymposium on Circuits and Systems, 23-26 May 2005 Page (s): 1230-1233Vol.2 " inter-frame mode selecting method of a kind of (from 16x16 to 4x4) from top to bottom proposed.In the method, if a certain inter-frame mode is not optimum in calculating all patterns of comparison, other inter-frame modes under this inter-frame mode also are considered to not be optimum so, therefore also just no longer calculate comparison, thereby have reduced computation complexity.And at " Xiaoan Lu, et al, Fast mode decisionand motion estimation for is with a focus on MPEG-2/H.264 transcoding.IEEE International Symposium on Circuits and Systems H.264,23-26 May 2005Page (s): 1246-1249 Vol.2 " in the method that proposes, the present frame information such as distortion, pattern, difference value of coded macroblocks is used to the selection of ancillary frame inter mode.But these two kinds of methods all do not make full use of the characteristic information of the input video stream of having encoded in the video code conversion, though so computation complexity decrease but still be difficult to satisfy the requirement of real-time transcoding, and the efficient of transcoding also has obvious reduction.

Summary of the invention

The fast selecting method that the purpose of this invention is to provide inter-frame mode in a kind of video code conversion.

Technical scheme of the present invention is: the fast selecting method of inter-frame mode in a kind of video code conversion may further comprise the steps:

(1) calculate the coefficient mean value Avg and the parameter variance Var of current macro respectively according to following two formulas:

$\mathrm{Avg}=\frac{1}{4}\underset{i=0}{\overset{3}{\mathrm{\Σ}}}({\mathrm{DCT}}_i\·A_i/\underset{i=0}{\overset{3}{\mathrm{\Σ}}}{A}_i)$

$\mathrm{Var}=\frac{1}{4}\underset{i=0}{\overset{3}{\mathrm{\Σ}}}{({\mathrm{DCT}}_i\·A_i/\underset{i=0}{\overset{3}{\mathrm{\Σ}}}{A}_i-\mathrm{Avg})}^2$

DCT wherein _iAnd A _iExpression is the number and the average absolute value of the summation about non-zero DCT coefficients of i piece in the current macro respectively;

(2) if Var＞T ₁, then change step (4) over to, otherwise change step (3), wherein T over to ₁Be threshold value, recommendation is 2.5;

(3) if Avg＞T ₂, then the utilance distortion optimization technology is carried out the selection of inter-frame mode, otherwise inter-frame mode is 16x16, wherein T ₂Be threshold value, recommendation is 3.25;

(4) calculate the weighting coefficient values F of four pieces in the current macro respectively according to following formula _i:

$F_i={\mathrm{DCT}}_i\·A_i/\underset{i=0}{\overset{3}{\mathrm{\Σ}}}{A}_i$

Wherein i is 0,1,2,3, represents the piece of upper left in the macro block, upper right, lower-left, position, bottom right respectively;

(5) with the weighting coefficient values F of four pieces in the current macro _iArrange according to ascending order, be designated as E ₁＜E ₂＜E ₃＜E ₄

(6) if E ₃＜Avg, inter-frame mode will the utilance distortion optimization technology be selected in 16x8,8x16 so, otherwise change step (7) over to;

(7) if E ₂＞Avg, inter-frame mode is 8x8 so, otherwise changes step (8) over to;

(8) if E ₁And E ₂Pairing two pieces are in diagonal position in current macro, inter-frame mode is 8x8 so, otherwise change step (9) over to;

(9) if E ₁And E ₂Pairing two pieces level in current macro is adjacent, and inter-frame mode will the utilance distortion optimization technology be selected in 8x8,8x16 so, otherwise inter-frame mode will the utilance distortion optimization technology be selected in 8x8,16x8.

The present invention and prior art contrast, effect is positive and tangible: the present invention makes full use of the characteristic information of the input video stream of having encoded in the video code conversion, as coefficient mean value, parameter variance, weighting coefficient values, significantly reduced the computation complexity of video code conversion, transcoding efficiency then remains unchanged substantially.

Embodiment

Below be the concrete steps of the embodiment of the invention, what this embodiment carried out is the video code conversion of MPEG-2 to AVS, the video sequence of test is silence, resolution is 176x144, be input as 100 frame MPEG-2 video flowings of 25 frame/seconds of frame per second, quantization parameter 16, be output as 100 frame AVS video flowings of 25 frame/seconds of frame per second, quantization parameter 28, the 1st frame of video flowing is the I frame, and all the other are the P frame:

(1) suppose that current macro is the 79th macro block of the 8th frame, calculate the coefficient mean value Avg and the parameter variance Var of this macro block, be respectively 2.75453 and 5.34568 according to following two formulas:

$\mathrm{Avg}=\frac{1}{4}\underset{i=0}{\overset{3}{\mathrm{\Σ}}}({\mathrm{DCT}}_i\·A_i/\underset{i=0}{\overset{3}{\mathrm{\Σ}}}{A}_i)$

$\mathrm{Var}=\frac{1}{4}\underset{i=0}{\overset{3}{\mathrm{\Σ}}}{({\mathrm{DCT}}_i\·A_i/\underset{i=0}{\overset{3}{\mathrm{\Σ}}}{A}_i-\mathrm{Avg})}^2$

(2) because Var＞T ₁So, change step (4) over to;

(3) if Avg＞T ₂, then the utilance distortion optimization technology is carried out the selection of inter-frame mode, otherwise inter-frame mode is 16x16, wherein T ₂Be threshold value, recommendation is 3.25;

(4) calculate the weighting coefficient values F of four pieces in the current macro respectively according to following formula _i, obtain F ₀=0, F ₁=3.38371, F ₂=1.45704, F ₃=6.17738;

$F_i={\mathrm{DCT}}_i\·A_i/\underset{i=0}{\overset{3}{\mathrm{\Σ}}}{A}_i$

(5) with F _iArrange according to ascending order, be designated as E ₁＜E ₂＜E ₃＜E ₄, E then ₁=F ₀=0, E ₂=F ₂=1.45704, E ₃=F ₁=3.38371, E ₄=F ₃=6.17738;

(6) because E ₃So＞Avg is the step of changing over to (7);

(7) because E ₂So＜Avg is the step of changing over to (8);

(8) because E ₁And E ₂Pairing two pieces are in upper left and the lower-left, are not diagonal positions, so the step of changing over to (9);

(9) because E ₁And E ₂Pairing two pieces are not that level is adjacent, so inter-frame mode will the utilance distortion optimization technology be selected in 8x8,16x8.

Claims (3)

1. the fast selecting method of inter-frame mode in the video code conversion is characterized in that may further comprise the steps:

(1) calculate the coefficient mean value Avg and the parameter variance Var of current macro respectively according to following two formulas:

$\mathrm{Avg}=\frac{1}{4}\underset{i=0}{\overset{3}{\mathrm{\Σ}}}({\mathrm{DCT}}_i\·A_i/\underset{i=0}{\overset{3}{\mathrm{\Σ}}}{A}_i)$

$\mathrm{Var}=\frac{1}{4}\underset{i=0}{\overset{3}{\mathrm{\Σ}}}{({\mathrm{DCT}}_i\·A_i/\underset{i=0}{\overset{3}{\mathrm{\Σ}}}{A}_i-\mathrm{Avg})}^2$

DCT wherein _iAnd A _iExpression is the number and the average absolute value of the summation about non-zero DCT coefficients of i piece in the current macro respectively;

(2) if Var＞T ₁, then change step (4) over to, otherwise change step (3), wherein T over to ₁Be threshold value;

(3) if Avg＞T ₂, then the utilance distortion optimization technology is carried out the selection of inter-frame mode, otherwise inter-frame mode is 16x16, wherein T ₂Be threshold value;

(4) calculate the weighting coefficient values F of four pieces in the current macro respectively according to following formula _i:

$F_i={\mathrm{DCT}}_i\·A_i/\underset{i=0}{\overset{3}{\mathrm{\Σ}}}{A}_i$

Wherein i is 0,1,2,3, represents the piece of upper left in the macro block, upper right, lower-left, position, bottom right respectively;

(5) with the weighting coefficient values F of four pieces in the current macro _iArrange according to ascending order, be designated as E ₁＜E ₂＜E ₃＜E ₄

(6) if E ₃＜Avg, inter-frame mode will the utilance distortion optimization technology be selected in 16x8,8x16 so, otherwise change step (7) over to;

(7) if E ₂＞Avg, inter-frame mode is 8x8 so, otherwise changes step (8) over to;

(8) if E ₁And E ₂Pairing two pieces are in diagonal position in current macro, inter-frame mode is 8x8 so, otherwise change step (9) over to;

(9) if E ₁And E ₂Pairing two pieces level in current macro is adjacent, and inter-frame mode will the utilance distortion optimization technology be selected in 8x8,8x16 so, otherwise inter-frame mode will the utilance distortion optimization technology be selected in 8x8,16x8.

2. the fast selecting method of inter-frame mode in the video code conversion according to claim 1 is characterized in that: threshold value T in the step (2) ₁Value be 2.5.

3. the fast selecting method of inter-frame mode in the video code conversion according to claim 1 and 2 is characterized in that: threshold value T in the step (3) ₂Value be 3.25.