CN102377993B - Intra-frame prediction mode selection method and system - Google Patents

Abstract

The invention discloses an intra-frame prediction mode selection method and an intra-frame prediction mode selection system. The method comprises the following steps of: selecting one of Intra_4*4, Intra_8*8 and Intra_16*16 as an intra-frame prediction type of a current macroblock according to a frequency characteristic of the current macroblock; when the Intra_16*16 is selected, selecting one of intra-frame prediction modes which is selectable for the Intra_16*16 and has the minimum prediction error as the intra-frame prediction mode of the current macroblock; and when the Intra_8*8 or the Intra_4*4 is selected, dividing the current macroblock into four 8*8 pixel matrixes, namely B0, B1, B2 and B3 or 16 4*4 pixel matrixes from D0 to D15; and as to any 8*8 pixel matrix bi or 4*4 pixel matrix Dj, selecting a prediction direction for the pixel matrix Bi or Dj according to the relevance among the pixels in the pixel matrix Bi or Dj, and selecting one of the intra-frame prediction modes which is selectable for the Intra_8*8 or the Intra_4*4 and has the minimum prediction error in the prediction direction for the pixel matrix Bi or Dj as the intra-frame prediction mode of the pixel matrix Bi or Dj.

Description

Method for choosing frame inner forecast mode and system

Technical field

The present invention relates to image processing field, relate more specifically to a kind of method for choosing frame inner forecast mode and system.

Background technology

H.264 the up-to-date international video encoding standard that Shi You International Telecommunications Union (ITU) and International Standards Organization (ISO) develop jointly.In standard H.264, infra-frame prediction is a kind of important technology that improves compression efficiency, in cataloged procedure, for the current some regions that will encode (being generally a rectangular block) in video image, search and its most akin and encoded piece in video image, with this, the most akin and encoded piece is predicted the current piece that will encode.

For color video frequency image, every frame video image comprises a luminance component frame and a chromatic component frame.For luminance component frame, H.264 standard has two kinds of type of prediction: Intra_16 * 16 (that is, 16 * 16 infra-frame predictions) and Intra_4 * 4 (that is, 4 * 4 infra-frame predictions).Particularly, Intra_4 * 4 are applicable to the infra-frame prediction that image detail changes obvious region, Intra_16 * 16 are applicable to the infra-frame prediction in image smoothing region, and it is available that wherein the former has 9 kinds of predictive modes, and it is available that the latter has 4 kinds of predictive modes.In supporting the H.264 standard of high accuracy expansion (FRExt), for luminance component frame except adopting Intra_16 * 16 and Intra_4 * 4 predict, can also adopt 8 * 8 infra-frame predictions (Intra_8 * 8) to predict, wherein have 9 kinds of predictive modes available.For chromatic component frame, can adopt with for the similar infra-prediction techniques of 16 * 16 infra-frame predictions of luminance component frame.

In order to improve coding quality and to reduce bit rate, H.264 standard has adopted rate-distortion optimization (RDO) technology to select infra-frame prediction type and the intra prediction mode for the macro block of needs coding is predicted.And RDO Technology Need is searched for totally by the infra-frame prediction type to all and intra prediction mode, just can select best infra-frame prediction type and intra prediction mode, so its computation complexity is very high, in hardware designs, be difficult to realize.

Summary of the invention

One or more problems in view of the above, the invention discloses a kind of method for choosing frame inner forecast mode and system of novelty.

Method for choosing frame inner forecast mode according to the embodiment of the present invention, comprising: according to the frequency characteristic of current macro, select one of 4 * 4 infra-frame predictions, 8 * 8 infra-frame predictions and 16 * 16 infra-frame predictions as the infra-frame prediction type for current macro; when selecting 16 * 16 infra-frame predictions as infra-frame prediction type for current macro, utilization is carried out infra-frame prediction to current macro respectively for the optional all intra prediction modes of 16 * 16 infra-frame prediction, and selection is used for the intra prediction mode of current macro for a kind of conduct of predicated error minimum in the optional intra prediction mode of 16 * 16 infra-frame prediction, when selecting 8 * 8 infra-frame predictions or 4 * 4 infra-frame predictions as infra-frame prediction type for current macro, current macro is divided into 48 * 8 picture element matrix B0～B3 or 16 4 * 4 picture element matrix D0～D15, and by following processing selecting the intra prediction mode for any one 8 * 8 picture element matrix Bi or any one 4 * 4 picture element matrix Dj, wherein, i and j are integer, 0≤i≤3,0≤j≤15: according to the correlation between the pixel in picture element matrix Bi or Dj, select the prediction direction for picture element matrix Bi or Dj, utilization is carried out infra-frame prediction to picture element matrix Bi or Dj in the prediction direction for picture element matrix Bi or Dj respectively for 8 * 8 infra-frame predictions or the optional all intra prediction modes of 4 * 4 infra-frame predictions, and the prediction direction being chosen in for picture element matrix Bi or Dj is used for the intra prediction mode of picture element matrix Bi or Dj for a kind of conduct of predicated error minimum in 8 * 8 infra-frame predictions or the optional intra prediction mode of 4 * 4 infra-frame predictions.

According to the Intra prediction mode selection system of the embodiment of the present invention, comprise: type of prediction selected cell, be configured to the frequency characteristic according to current macro, select one of 4 * 4 infra-frame predictions, 8 * 8 infra-frame predictions and 16 * 16 infra-frame predictions as the infra-frame prediction type for current macro; predictive mode selected cell, be configured to when type of prediction selected cell is selected 16 * 16 infra-frame predictions as infra-frame prediction type for current macro, utilization is carried out infra-frame prediction to current macro respectively for the optional all intra prediction modes of 16 * 16 infra-frame prediction, and selection is used for the intra prediction mode of current macro for a kind of conduct of predicated error minimum in the optional intra prediction mode of 16 * 16 infra-frame prediction, when type of prediction selected cell is selected 8 * 8 infra-frame predictions or 4 * 4 infra-frame predictions as infra-frame prediction type for current macro, current macro is divided into 48 * 8 picture element matrix B0～B3 or 16 4 * 4 picture element matrix D0～D15, and by following processing selecting the intra prediction mode for any one 8 * 8 picture element matrix Bi or any one 4 * 4 picture element matrix Dj, wherein, i and j are integer, 0≤i≤3,0≤j≤15: according to the correlation between the pixel in picture element matrix Bi or Dj, select the prediction direction for picture element matrix Bi or Dj, utilization is carried out infra-frame prediction to picture element matrix Bi or Dj in the prediction direction for picture element matrix Bi or Dj respectively for 8 * 8 infra-frame predictions or the optional all intra prediction modes of 4 * 4 infra-frame predictions, and the prediction direction being chosen in for picture element matrix Bi or Dj is used for the intra prediction mode of picture element matrix Bi or Dj for a kind of conduct of predicated error minimum in 8 * 8 infra-frame predictions or the optional intra prediction mode of 4 * 4 infra-frame predictions.

According to the method for choosing frame inner forecast mode of the embodiment of the present invention and system, by the method for choosing frame inner forecast mode with respect to traditional and the comparatively simple method of system, be identified for the intra prediction mode of the regional in current macro, in the coding quality that does not reduce video image, improved the code rate of video image.

Accompanying drawing explanation

Below in conjunction with accompanying drawing in the description of the specific embodiment of the present invention, the present invention may be better understood, wherein:

Fig. 1 shows according to the logic diagram of the Intra prediction mode selection system of the embodiment of the present invention;

Fig. 2 shows according to the flow chart of the method for choosing frame inner forecast mode of the embodiment of the present invention;

Fig. 3 shows the logic diagram of type of prediction selected cell 102;

Fig. 4 shows 102 selections of type of prediction selected cell for the flow chart of the processing of the infra-frame prediction type of current macro;

Fig. 5 shows in level, vertical, lower-left, lower right the intra prediction mode that can use for Intra_4 * 4 and Intra_8 * 8;

Fig. 6 shows the flow chart of the processing of the infra-frame prediction type that is identified for picture element matrix Bi;

Fig. 7 shows an example that gathers 4 * 4 picture element matrix A0 of 16 pixels compositions from current macro;

Fig. 8 shows an example of the picture element matrix A0 ' after conversion;

Fig. 9 shows an example that gathers 4 * 4 picture element matrix Ci of 16 pixels compositions from 8 * 8 picture element matrix Bi;

Figure 10 shows the logic diagram of predictive mode selected cell 104;

Figure 11 shows the flow chart of process that predictive mode selected cell is selected the prediction direction of any one 8 * 8 picture element matrix Bi comprise for current macro;

Figure 12 shows the flow chart of process that predictive mode selected cell is selected the prediction direction of any one 4 * 4 picture element matrix Bi comprise for current macro; And

Figure 13 shows and utilizes H01, the H10 of picture element matrix Ci or Dj and the process example of the prediction direction that H11 is identified for picture element matrix Bi and Dj.

Embodiment

Feature and the exemplary embodiment of various aspects of the present invention will be described in detail below.Many details have been contained in description below, to complete understanding of the present invention is provided.But, it will be apparent to one skilled in the art that the present invention can be in the situation that do not need some details in these details to implement.Below the description of embodiment is only used to provide the clearer understanding to the present invention by example of the present invention is shown.The present invention is never limited to any concrete configuration and the algorithm proposing below, but has covered under the premise of without departing from the spirit of the present invention any modification, replacement and the improvement of coherent element, parts and algorithm.

In standard H.264, the size of the prediction piece of predicting for any one luminance macroblock that luminance component frame is comprised can be 4 * 4,8 * 8 or 16 * 16.Wherein, in the situation that adopts the prediction piece that size is 4 * 4 (that is, adopting Intra_4 * 4) to predict luminance macroblock, exist 9 kinds of predictive modes available; In the situation that adopts the prediction piece that size is 8 * 8 (that is, adopting Intra_8 * 8) to predict luminance macroblock, exist 9 kinds of predictive modes available; And in the situation that adopts the prediction piece that size is 16 * 16 (that is, adopting Intra_16 * 16) to predict luminance macroblock, exist 4 kinds of predictive modes available.

Fig. 1 shows according to the block diagram of the Intra prediction mode selection system of the embodiment of the present invention.Fig. 2 shows according to the flow chart of the method for choosing frame inner forecast mode of the embodiment of the present invention.As shown in Figure 1, this Intra prediction mode selection system comprises type of prediction selected cell 102 and predictive mode selected cell 104.

The current macro of needs coding is being carried out in the process of infra-frame prediction, type of prediction selected cell 102 is according to the frequency characteristic of current macro, select one of 4 * 4 infra-frame predictions, 8 * 8 infra-frame predictions and 16 * 16 infra-frame predictions as the infra-frame prediction type for current macro (that is, execution step S202).When type of prediction selected cell 102 is selected 16 * 16 infra-frame predictions as infra-frame prediction type for current macro, predictive mode selected cell 104 utilizes and respectively current macro is carried out to infra-frame prediction for the optional all intra prediction modes of 16 * 16 infra-frame prediction, and selection is used for the intra prediction mode (that is, execution step S2042) of current macro for a kind of conduct of predicated error minimum in the optional intra prediction mode of 16 * 16 infra-frame prediction.When type of prediction selected cell 102 is selected 8 * 8 infra-frame predictions or 4 * 4 infra-frame predictions as infra-frame prediction type for current macro, predictive mode selected cell 104 is divided into 48 * 8 picture element matrix B0～B3 or 16 4 * 4 picture element matrix D0～D15 by current macro, and by following processing selecting for the intra prediction mode of any one 8 * 8 picture element matrix Bi or any one 4 * 4 picture element matrix Dj (wherein, i and j are integer, 0≤i≤3, 0≤j≤15): according to the correlation between the pixel in picture element matrix Bi or Dj, select the prediction direction for picture element matrix Bi or Dj, utilization is carried out infra-frame prediction to picture element matrix Bi or Dj in the prediction direction for picture element matrix Bi or Dj respectively for 8 * 8 infra-frame predictions or the optional all intra prediction modes of 4 * 4 infra-frame predictions, and the prediction direction being chosen in for picture element matrix Bi or Dj is used for the intra prediction mode (that is, execution step S2044 to S2046) of picture element matrix Bi or Dj for a kind of conduct of predicated error minimum in 8 * 8 infra-frame predictions or the optional intra prediction mode of 4 * 4 infra-frame predictions.

Here it should be noted that, according in the intra prediction mode method and system of the embodiment of the present invention, thereby in order to go out best intra prediction mode by comparatively simple process choosing, improve the coding rate for video image, the following four kinds of prediction direction of definition: horizontal direction (Horizontaldirection), vertical direction (Vertical direction), lower right (Diagonal Rightdirection) and lower left (Diagonal Left direction).When selecting Intra_4 * 4 or Intra_8 * 8 conduct to be used for the infra-frame prediction type of current macro, each marking off from current macro 4 * 4 or 8 * 8 picture element matrixs can be suitable for identical prediction direction, also can be suitable for different prediction direction.

Fig. 5 shows in level, vertical, lower-left, lower right the intra prediction mode that can use for Intra_4 * 4 and Intra_8 * 8.As shown in Figure 5, in a horizontal direction, for Intra_4 * 4 and Intra_8 * 8 can with intra prediction mode be Horizontal_Up (level _ on), Horizontal (level), Horizontal_Down (level _ under) and DC prediction (direct current prediction).In the vertical direction, for Intra_4 * 4 and Intra_8 * 8 can with intra prediction mode be Vertical_Left (vertical _ left side), Vertical (vertically), Vertical_Right (vertical _ right side) and DC prediction.In lower right, for Intra_4 * 4 and Intra_8 * 8 can with intra prediction mode be Vertical_Right (vertical _ right side), Diagonal_Down_Right (diagonal _ under _ right side), Horizontal_Down (level _ under) and DC prediction.In lower left, for Intra_4 * 4 and Intra_8 * 8 can with intra prediction mode be Diagonal_Down_Left (diagonal _ under _ left side), Vertical_Left (vertical _ left side), Horizontal_Up (diagonal _ on) and DC prediction.

Below, type of prediction selected cell 102 is described in detail.Fig. 3 shows the logic diagram of type of prediction selected cell 102.Fig. 4 shows 102 selections of type of prediction selected cell for the flow chart of the processing of the infra-frame prediction type of current macro.As shown in Figure 3, type of prediction selected cell 102 comprises that the first pixel acquisition module 1022, the first conversion Executive Module 1024, first select Executive Module 1026, the first matrix to divide module 1028, the second pixel acquisition module 1030, second converts Executive Module 1032, second and selects Executive Module 1034 and Selective type judging module 1036.

In the process of infra-frame prediction type that is identified for current macro, the first pixel acquisition module 1022 forms 4 * 4 picture element matrix A0 (that is, execution step S2022) for gather 16 pixels from current macro.1024 couples of picture element matrix A0 of the first conversion Executive Module carry out Hadamard transform, and the DC coefficient of the picture element matrix A0 ' after computational transformation and the ratio of ac coefficient (that is, execution step S2024).When the DC coefficient of picture element matrix A0 ' after conversion and the ratio of ac coefficient are not less than first threshold, first selects Executive Module 1026 to select 16 * 16 infra-frame predictions as the infra-frame prediction type for current macro (that is, execution step S2026).

When the DC coefficient of picture element matrix A0 ' after conversion and the ratio of ac coefficient are less than first threshold, the first matrix is divided module 1028 for current macro being divided into 48 * 8 picture element matrix B0～B3 (that is, execution step S2028).The second pixel sampling module 1030, the second conversion Executive Module 1032 and second select Executive Module 1034 to carry out the processing of the intra prediction mode of each the picture element matrix Bi the picture element matrix B0～B3 that is identified for marking off from current macro (that is, execution step 2030).Having determined the infra-frame prediction type of the picture element matrix B0～B3 for marking off from current macro, 1036 pairs of Selective type judging module adopt 8 * 8 picture element matrixs of 8 * 8 infra-frame predictions to count, and to adopting 8 * 8 picture element matrixs of 4 * 4 infra-frame predictions to count, and in the situation that adopt the number of 8 * 8 picture element matrixs of 8 * 8 infra-frame predictions to be not less than 2 in current macro, select 8 * 8 infra-frame predictions as the infra-frame prediction type for current macro, in the situation that adopt the number of 8 * 8 picture element matrixs of 4 * 4 infra-frame predictions to be greater than 2 in current macro, select 4 * 4 infra-frame predictions as the infra-frame prediction type for current macro (, execution step S2032).

Fig. 6 shows the flow chart of the processing of the infra-frame prediction type that is identified for picture element matrix Bi.As shown in Figure 6, in the process of infra-frame prediction type that is identified for picture element matrix Bi, the second pixel acquisition module 1030 gathers 16 pixels composition 4 * 4 picture element matrix Ci from picture element matrix Bi, and (wherein, i is integer, 0≤i≤3) (that is, execution step S2030-2); 1032 couples of picture element matrix Ci of the second conversion Executive Module carry out Hadamard transform, and the DC coefficient of the picture element matrix Ci ' after computational transformation and the ratio of ac coefficient (that is, execution step S2030-4); If the DC coefficient of picture element matrix Ci ' and the ratio of ac coefficient after conversion are not less than Second Threshold, second select Executive Module 1034 to select 8 * 8 infra-frame predictions as the infra-frame prediction type for picture element matrix Bi, otherwise second selects Executive Module 1034 to select 4 * 4 infra-frame predictions as the infra-frame prediction type for picture element matrix Bi (that is, execution step S2030-6).

Fig. 7 shows an example that gathers 4 * 4 picture element matrix A0 of 16 pixels compositions from current macro.As shown in Figure 7, current macro can be divided into 16 4 * 4 picture element matrix D0～D15, and a pixel groups that gathers the upper left corner of each picture element matrix in picture element matrix D0～D15 becomes picture element matrix A0.

According in the method for choosing frame inner forecast mode of the embodiment of the present invention and system, can to picture element matrix A0, carry out Hadamard transform according to following equation:

D′＝HT4·D·HT4

$\mathrm{HT}4=\left[\begin{array}{cccc}1& 1& 1& 1\\ 1& 1& -1& -1\\ 1& -1& -1& 1\\ 1& -1& 1& -1\end{array}\right]$

D1＝D·HT4

(d1 _i0d1 _i1d1 _i2d1 _i3)＝(d _i0d _i1d _i2d _i3)·HT4

D’＝HT4·D1

$\left(\begin{array}{c}{d}_{0j}^{\′}\\ d_{1j}^{\′}\\ d_{2j}^{\′}\\ d_{3j}^{\′}\end{array}\right)=\mathrm{HT}4\·\left(\begin{array}{c}{d1}_{0j}\\ {d1}_{1j}\\ {d1}_{2j}\\ {d1}_{3j}\end{array}\right)$

Wherein, D represents picture element matrix A0, and HT4 represents Hadamard transform matrix, and D1 represents intermediary matrix, and D ' represents the picture element matrix A0 ' after conversion.Fig. 8 shows an example of the picture element matrix A0 ' after conversion.As shown in Figure 8, a pixel coefficient in the upper left corner of the picture element matrix A0 ' after conversion is direct current (DC) coefficient, and other pixel coefficient are to exchange (AC) coefficient.

The DC coefficient of picture element matrix A0 ' after conversion and the ratio DC_ratio of ac coefficient can represent the frequency characteristic of current macro.Wherein, larger DC_ratio represents that current macro is smoother, and less DC_ratio represents current macro more complicated.Wherein, the DC coefficient DC_value of the picture element matrix A0 ' after conversion, ac coefficient AC_value and DC coefficient are defined as foloows with the ratio DC_ratio of ac coefficient:

DC_value＝abs(a ₀₀)

AC_value＝abs(a ₀₁)+abs(a ₀₂)+abs(a ₀₃)+

abs(a ₁₀)+abs(a ₁₁)+abs(a ₁₂)+abs(a ₁₃)+

abs(a ₂₀)+abs(a ₂₁)+abs(a ₂₂)+abs(a ₂₃)+

abs(a ₃₀)+abs(a ₃₁)+abs(a ₃₂)+abs(a ₃₃)

$\mathrm{DC}\_\mathrm{ratio}=\frac{\mathrm{DC}\_\mathrm{value}}{\mathrm{AC}\_\mathrm{value}}$

If DC_ratio is not less than first threshold TH1, selects Intra_16 * 16 as the infra-frame prediction type for current macro, otherwise select Intra_8 * 8 or Intra_4 * 4 as the infra-frame prediction type for current macro.Type of prediction selected cell 102 can determine it is finally to select Intra_8 * 8 or select Intra_4 * 4 as the infra-frame prediction type for current macro by step S2028 to S2032.

Fig. 9 shows the example that gathers 4 * 4 picture element matrix Ci of 16 pixels compositions from 8 * 8 picture element matrix Bi.As shown in Figure 9, can gather picture element matrix Bi first and third, five, seven pixel columns first become picture element matrix Ci with the 3rd of the pixel of the 4th pixel column and second, four, six, eight pixel columns of picture element matrix Bi with the pixel groups of the 7th pixel column.

Can adopt the Hadamard transform method identical with picture element matrix A0 to carry out Hadamard transform to picture element matrix Ci, thus the picture element matrix Ci ' after being converted.Can adopt with convert after the identical method of picture element matrix A0 ' come the DC coefficient of the picture element matrix Ci ' after computational transformation and the ratio of ac coefficient.If the DC coefficient of picture element matrix Ci ' and the ratio of ac coefficient after conversion are less than Second Threshold TH2, select Intra_4 * 4 as the infra-frame prediction type for picture element matrix Bi, otherwise select Intra_8 * 8 as the infra-frame prediction type for picture element matrix Bi.After the infra-frame prediction type of each comprising in determining current macro 8 * 8 picture element matrixs, respectively to adopting 8 * 8 picture element matrixs of Intra_8 * 8 and adopting 8 * 8 picture element matrixs of Intra_4 * 4 to count, and adopting the number of 8 * 8 picture element matrixs of Intra_8 * 8 to be not less than at 2 o'clock, selecting Intra_8 * 8 as the infra-frame prediction type for current macro.

In an embodiment of the present invention, can be set to 50 by TH1, and TH2 is set to 25, to be applicable to most of video images.

Figure 10 shows the logic diagram of predictive mode selected cell 104.As shown in figure 10, predictive mode selected cell 104 comprises the 3rd pixel acquisition module 1042, calculated for pixel values module 1044, set direction Executive Module 1046.

When type of prediction selected cell 102 selects Intra_16 * 16 conduct to be used for the infra-frame prediction type of current macro, predictive mode selected cell 104 utilizes and respectively current macro is carried out to infra-frame prediction for the optional all intra prediction modes in Intra_16 * 16, and selects to be used for for a kind of conduct of predicated error minimum in the optional intra prediction mode in Intra_16 * 16 intra prediction mode of current macro.When type of prediction selected cell 102 selects Intra_8 * 8 or Intra_4 * 4 conduct to be used for the infra-frame prediction type of current macro, the prediction direction of each 8 * 8 or 4 * 4 picture element matrix that first predictive mode selected cell is identified for comprising in current macro, then for each 8 * 8 or 4 * 4 picture element matrix, utilization is carried out infra-frame prediction to this 8 * 8 or 4 * 4 picture element matrix respectively at all intra prediction modes for using for Intra_8 * 8 or Intra_4 * 4 in its definite prediction direction, and select a kind of intra prediction mode of predicated error minimum as the intra prediction mode for these 8 * 8 or 4 * 4 picture element matrixs.

Figure 11 shows the flow chart of process that predictive mode selected cell is selected the prediction direction of any one 8 * 8 picture element matrix Bi comprise for current macro.As shown in figure 11, when type of prediction selected cell 102 selects Intra_8 * 8 conduct to be used for the infra-frame prediction type of current macro, for divide any one 8 * 8 picture element matrix Bi drawing from current macro, the 3rd pixel acquisition module 1042 gathers 16 pixels from picture element matrix Bi and forms 4 * 4 picture element matrix Ci (that is, execution step S2044-2).The poor H11 (that is, execution step S2044-4) of four pixel values in four pixel value sums in four pixel values in the upper left corner of the pixel value sum of the first and second pixel columns of calculated for pixel values module 1044 calculating pixel Matrix C i and the pixel value sum of the first and second pixel columns of poor H01, picture element matrix Ci between the pixel value sum of the third and fourth pixel column and the poor H10 between the pixel value sum of the third and fourth pixel column and picture element matrix Ci and the lower right corner and the upper right corner and four pixel value sums in the lower left corner.If | H01| >=2 * | H10| and | H01| >=| H11|, set direction Executive Module 1046 selects vertical direction as the prediction direction for picture element matrix Bi; If | H10| >=2 * | H01| and | H10| >=| H11|, set direction Executive Module 1046 selection level directions are as the prediction direction for picture element matrix Bi.If H01, H10, and the relation between H11 is discontented with two conditions that are enough to, the poor AdditionalValue2 of six pixel value sums in six pixel value sums in the upper left corner of set direction Executive Module 1046 calculating pixel matrix B i and the poor AddionalValue1 of six pixel value sums in the lower right corner and the upper right corner of picture element matrix Bi and six pixel value sums in the lower left corner, and in the situation that AdditionalValue1 is greater than AdditionalValue2, select lower left as the prediction direction for picture element matrix Bi, in the situation that AdditionalValue1 is not more than AdditionalValue2, select lower right as the prediction direction for picture element matrix Bi (, execution step S2044-6).

Figure 12 shows the flow chart of process that predictive mode selected cell is selected the prediction direction of any one 4 * 4 picture element matrix Bi comprise for current macro.As shown in figure 12, when type of prediction selected cell 102 selects Intra_4 * 4 conduct to be used for the infra-frame prediction type of current macro, for dividing from current macro any one 4 * 4 picture element matrix Dj drawing, (j is integer, and 0≤j≤15), the pixel value sum of the first and second pixel columns of calculated for pixel values module 1046 calculating pixel matrix D j and the poor H01 between the pixel value sum of the third and fourth pixel column, poor H10 between the pixel value sum of the first and second pixel columns of picture element matrix Dj and the pixel value sum of the third and fourth pixel column, and the poor H11 of four pixel values in four pixel value sums in four pixel values in the upper left corner of picture element matrix Dj and the lower right corner and the upper right corner and four pixel value sums in the lower left corner (, execution step S2044-8).If | H01| >=2 * | H10| and | H01| >=| H11|, set direction Executive Module 1048 selects vertical direction as the prediction direction for picture element matrix Dj, if | H10| >=2 * | H01| and | H10| >=| H11|, set direction Executive Module 1048 selection level directions are as the prediction direction for picture element matrix Dj, otherwise the poor AdditionalValue2 of six pixel value sums in six pixel value sums in the upper left corner of calculating pixel matrix D j and the poor AddionalValue1 of six pixel value sums in the lower right corner and the upper right corner of picture element matrix Dj and six pixel value sums in the lower left corner, if AdditionalValue1 is greater than AdditionalValue2, select lower left as the prediction direction for picture element matrix Dj, otherwise select lower right as the prediction direction for picture element matrix Dj (, execution step S2044-10).Figure 13 shows and utilizes H01, the H10 of picture element matrix Ci or Dj and the process example of the prediction direction that H11 is identified for picture element matrix Bi and Dj.

Particularly, example that can be using the picture element matrix shown in Figure 13 as picture element matrix Ci or Dj, calculates H01, H10 and the H11 of the picture element matrix shown in Figure 13 according to following equation.

H01＝(A00+A10+A20+A30+A01+A11+A21+A31)

-(A02+A12+A22+A32+A03+A13+A23+A33)

H10＝(A00+A01+A02+A03+A10+A11+A12+A13)

-(A20+A21+A22+A23+A30+A31+A32+A33)

H11＝(A00+A10+A01+A11)+(A22+A32+A23+A33)

-(A20+A30+A21+A31)-(A02+A12+A03+A13)

In addition, example that can be using the picture element matrix shown in Figure 13 as picture element matrix Ci or Dj, the AdditionalValue1 that calculates the picture element matrix shown in Figure 13 according to following equation is greater than AdditionalValue2.

AdditionalValue1＝(A00+A10+A01+A20+A11+A02)-(A31+A22+A13+A32+A23+A33)

AdditionalValue2＝(A03+A02+A13+A01+A12+A23)-(A10+A21+A32+A20+A31+A30)

More than described according to the method for choosing frame inner forecast mode of the embodiment of the present invention and system.But those skilled in the art should be understood that, the division of above functional unit is that a kind of logic function of the present invention is divided example, identical or akin function also can be realized by different logical blocks as required, also can be realized by same logical block.

According to the method for choosing frame inner forecast mode of the embodiment of the present invention and system, by the method for choosing frame inner forecast mode with respect to traditional and the comparatively simple method of system, be identified for the intra prediction mode of the regional in current macro, in the coding quality that does not reduce video image, improved the code rate of video image.

Below with reference to specific embodiments of the invention, the present invention has been described, but those skilled in the art all understand, can carry out various modifications, combination and change to these specific embodiments, and can not depart from the spirit and scope of the present invention that limited by claims or its equivalent.

Can perform step with hardware or software as required.Note, without departing from the scope of the invention, can add step, therefrom remove step or revise step wherein to the flow chart providing in this specification.In general, flow chart is just used to refer to a kind of possible sequence for the basic operation of practical function.

Embodiments of the invention can utilize programming general purpose digital computer, utilize application-specific integrated circuit (ASIC), programmable logic device, field programmable gate array, light, chemistry, biological, system quantum or nanometer engineering, assembly and mechanism to realize.In general, function of the present invention can be realized by any means known in the art.Can use distributed or networked system, assembly and circuit.The communication of data or to transmit can be wired, wireless or by any other means.

Also will recognize, according to the needs of application-specific, one or more can the realization by more separated or more integrated mode in the key element shown in accompanying drawing, or be removed even in some cases or be deactivated.The journey logic bomb that realization can be stored in machine readable media is carried out above-mentioned any method with permission computer, also within the spirit and scope of the present invention.

In addition, it is only exemplary that any signal arrows in accompanying drawing should be considered to, rather than restrictive, unless separately there is concrete indication.When term is also contemplated as when the ability of separation or combination is not known, the combination of assembly or step also will be considered to record.

Claims (16)

1. a method for choosing frame inner forecast mode, comprising:

According to the frequency characteristic of current macro, select one of 4 * 4 infra-frame predictions, 8 * 8 infra-frame predictions and 16 * 16 infra-frame predictions as the infra-frame prediction type for described current macro;

When selecting 16 * 16 infra-frame predictions as infra-frame prediction type for described current macro, utilization is carried out infra-frame prediction to described current macro respectively for the optional all intra prediction modes of 16 * 16 infra-frame prediction, and selection is used for the intra prediction mode of described current macro for a kind of conduct of predicated error minimum in the optional intra prediction mode of 16 * 16 infra-frame prediction

When selecting 8 * 8 infra-frame predictions or 4 * 4 infra-frame predictions as infra-frame prediction type for described current macro, described current macro is divided into 48 * 8 picture element matrix B0～B3 or 16 4 * 4 picture element matrix D0～D15, and by following processing selecting the intra prediction mode for any one 8 * 8 picture element matrix Bi or any one 4 * 4 picture element matrix Dj, wherein, i and j are integer, 0≤i≤3,0≤j≤15:

According to the correlation between the pixel in described picture element matrix Bi or Dj, select the prediction direction for described picture element matrix Bi or Dj,

Utilization is carried out infra-frame prediction to described picture element matrix Bi or Dj in the prediction direction for described picture element matrix Bi or Dj respectively for 8 * 8 infra-frame predictions or the optional all intra prediction modes of 4 * 4 infra-frame predictions, and the prediction direction being chosen in for described picture element matrix Bi or Dj is used for the intra prediction mode of described picture element matrix Bi or Dj for a kind of conduct of predicated error minimum in 8 * 8 infra-frame predictions or the optional intra prediction mode of 4 * 4 infra-frame predictions.

2. method for choosing frame inner forecast mode according to claim 1, is characterized in that, selection comprises for the processing of the infra-frame prediction type of described current macro:

From described current macro, gather 16 pixels and form 4 * 4 picture element matrix A0;

Described picture element matrix A0 is carried out to Hadamard transform, and the DC coefficient of the picture element matrix A0 ' after computational transformation and the ratio of ac coefficient;

When the DC coefficient of the picture element matrix A0 ' after described conversion is not less than first threshold with the ratio of ac coefficient, select 16 * 16 infra-frame predictions as the infra-frame prediction type for described current macro.

3. method for choosing frame inner forecast mode according to claim 2, is characterized in that, when the DC coefficient of the picture element matrix A0 ' after described conversion is less than described first threshold with the ratio of ac coefficient, also comprises:

Described current macro is divided into 48 * 8 picture element matrix B0～B3, and any one the picture element matrix Bi in described picture element matrix B0～B3 is carried out to following processing, wherein, i is integer, 0≤i≤3:

From described picture element matrix Bi, gather 16 pixels and form 4 * 4 picture element matrix Ci;

Described picture element matrix Ci is carried out to Hadamard transform, and the DC coefficient of the picture element matrix Ci ' after computational transformation and the ratio of ac coefficient;

If the DC coefficient of the picture element matrix Ci ' after described conversion is not less than Second Threshold with the ratio of ac coefficient, select 8 * 8 infra-frame predictions as the infra-frame prediction type for described picture element matrix Bi, otherwise select 4 * 4 infra-frame predictions as the infra-frame prediction type for described picture element matrix Bi; And

If adopt the number of 8 * 8 picture element matrixs of 8 * 8 infra-frame predictions to be not less than 2 in described current macro, select 8 * 8 infra-frame predictions as the infra-frame prediction type for described current macro, otherwise select 4 * 4 infra-frame predictions as the infra-frame prediction type for described current macro.

4. method for choosing frame inner forecast mode according to claim 2, is characterized in that, the processing that gathers 16 pixels composition 4 * 4 picture element matrix A0 from described current macro comprises:

Described current macro is divided into 16 4 * 4 picture element matrix D0～D15;

A pixel that gathers the upper left corner of each picture element matrix in described picture element matrix D0～D15 forms described picture element matrix A0.

5. method for choosing frame inner forecast mode according to claim 3, it is characterized in that, gather the pixel of first and the 4th pixel column of first and third, five, seven pixel columns of described picture element matrix Bi and the pixel of the 3rd and the 7th pixel column of second, four, six, eight pixel columns of described picture element matrix Bi forms described picture element matrix Ci.

6. method for choosing frame inner forecast mode according to claim 1, is characterized in that, selects one of vertical direction, horizontal direction, lower right and lower left as the prediction direction for described picture element matrix Bi or Dj.

7. method for choosing frame inner forecast mode according to claim 6, is characterized in that, when selecting 8 * 8 infra-frame predictions as infra-frame prediction type for described current macro, selects to comprise for the processing of the prediction direction of described picture element matrix Bi:

From described picture element matrix Bi, gather 16 pixels and form 4 * 4 picture element matrix Ci;

Calculate the poor H11 of four pixel values in four pixel value sums in four pixel values in the upper left corner of the pixel value sum of the first and second pixel columns of described picture element matrix Ci and the pixel value sum of the first and second pixel columns of poor H01, described picture element matrix Ci between the pixel value sum of the third and fourth pixel column and the poor H10 between the pixel value sum of the third and fourth pixel column and described picture element matrix Ci and the lower right corner and the upper right corner and four pixel value sums in the lower left corner;

If | H01| >=2 * | H10| and | H01| >=| H11|, select vertical direction as the prediction direction for described picture element matrix Bi, otherwise

If | H10| >=2 * | H01| and | H10| >=| H11|, selection level direction is as the prediction direction for described picture element matrix Bi, otherwise

Calculate the poor AdditionalValue2 of six pixel value sums in six pixel value sums in the upper left corner of described picture element matrix Bi and the poor AddionalValue1 of six pixel value sums in the lower right corner and the upper right corner of described picture element matrix Bi and six pixel value sums in the lower left corner

If AdditionalValue1 is greater than AdditionalValue2, selects lower left as the prediction direction for described picture element matrix Bi, otherwise select lower right as the prediction direction for described picture element matrix Bi.

8. method for choosing frame inner forecast mode according to claim 6, is characterized in that, when selecting 4 * 4 infra-frame predictions as infra-frame prediction type for described current macro, selects to comprise for the processing of the prediction direction of described picture element matrix Dj:

Calculate the poor H11 of four pixel values in four pixel value sums in four pixel values in the upper left corner of the pixel value sum of the first and second pixel columns of described picture element matrix Dj and the pixel value sum of the first and second pixel columns of poor H01, described picture element matrix Dj between the pixel value sum of the third and fourth pixel column and the poor H10 between the pixel value sum of the third and fourth pixel column and described picture element matrix Dj and the lower right corner and the upper right corner and four pixel value sums in the lower left corner;

If | H01| >=2 * | H10| and | H01| >=| H11|, select vertical direction as the prediction direction for described picture element matrix Dj, otherwise

If | H10| >=2 * | H01| and | H10| >=| H11|, selection level direction is as the prediction direction for described picture element matrix Dj, otherwise

Calculate the poor AdditionalValue2 of six pixel value sums in six pixel value sums in the upper left corner of described picture element matrix Dj and the poor AddionalValue1 of six pixel value sums in the lower right corner and the upper right corner of described picture element matrix Dj and six pixel value sums in the lower left corner

If AdditionalValue1 is greater than AdditionalValue2, selects lower left as the prediction direction for described picture element matrix Dj, otherwise select lower right as the prediction direction for described picture element matrix Dj.

9. an Intra prediction mode selection system, comprising:

Type of prediction selected cell, is configured to the frequency characteristic according to current macro, selects one of 4 * 4 infra-frame predictions, 8 * 8 infra-frame predictions and 16 * 16 infra-frame predictions as the infra-frame prediction type for described current macro;

Predictive mode selected cell, is configured to

When described type of prediction selected cell is selected 16 * 16 infra-frame predictions as infra-frame prediction type for described current macro, utilization is carried out infra-frame prediction to described current macro respectively for the optional all intra prediction modes of 16 * 16 infra-frame prediction, and selection is used for the intra prediction mode of described current macro for a kind of conduct of predicated error minimum in the optional intra prediction mode of 16 * 16 infra-frame prediction

When described type of prediction selected cell is selected 8 * 8 infra-frame predictions or 4 * 4 infra-frame predictions as infra-frame prediction type for described current macro, described current macro is divided into 48 * 8 picture element matrix B0～B3 or 16 4 * 4 picture element matrix D0～D15, and by following processing selecting the intra prediction mode for any one 8 * 8 picture element matrix Bi or any one 4 * 4 picture element matrix Dj, wherein, i and j are integer, 0≤i≤3,0≤j≤15:

According to the correlation between the pixel in described picture element matrix Bi or Dj, select the prediction direction for described picture element matrix Bi or Dj,

Utilization is carried out infra-frame prediction to described picture element matrix Bi or Dj in the prediction direction for described picture element matrix Bi or Dj respectively for 8 * 8 infra-frame predictions or the optional all intra prediction modes of 4 * 4 infra-frame predictions, and the prediction direction being chosen in for described picture element matrix Bi or Dj is used for the intra prediction mode of described picture element matrix Bi or Dj for a kind of conduct of predicated error minimum in 8 * 8 infra-frame predictions or the optional intra prediction mode of 4 * 4 infra-frame predictions.

10. Intra prediction mode selection system according to claim 9, is characterized in that, described type of prediction selected cell comprises:

The first pixel acquisition module, is configured to gather 16 pixels from described current macro and forms 4 * 4 picture element matrix A0;

The first conversion Executive Module, is configured to described picture element matrix A0 to carry out Hadamard transform, and the DC coefficient of the picture element matrix A0 ' after computational transformation and the ratio of ac coefficient;

First selects Executive Module, and when being configured to the DC coefficient of the picture element matrix A0 ' when described conversion after and the ratio of ac coefficient and being not less than first threshold, selection 16 * 16 infra-frame predictions are as the infra-frame prediction type for described current macro.

11. Intra prediction mode selection systems according to claim 10, is characterized in that, described type of prediction selected cell also comprises:

The first matrix is divided module, when being configured to the DC coefficient of the picture element matrix A0 ' after described conversion and the ratio of ac coefficient and being less than described first threshold, described current macro is divided into 48 * 8 picture element matrix B0～B3;

The second pixel acquisition module, any one the picture element matrix Bi being configured to from described picture element matrix B0～B3 gathers 16 pixels composition 4 * 4 picture element matrix Ci, and wherein, i is integer, 0≤i≤3;

The second conversion Executive Module, is configured to described picture element matrix Ci to carry out Hadamard transform, and the DC coefficient of the picture element matrix Ci ' after computational transformation and the ratio of ac coefficient;

Second selects Executive Module, if be configured to the DC coefficient of the picture element matrix Ci ' after described conversion and the ratio of ac coefficient is not less than Second Threshold, select 8 * 8 infra-frame predictions as the infra-frame prediction type for described picture element matrix Bi, otherwise select 4 * 4 infra-frame predictions as the infra-frame prediction type for described picture element matrix Bi; And

Selective type judging module, if be configured to, in described current macro, adopt the number of 8 * 8 picture element matrixs of 8 * 8 infra-frame predictions to be not less than 2, select 8 * 8 infra-frame predictions as the infra-frame prediction type for described current macro, otherwise select 4 * 4 infra-frame predictions as the infra-frame prediction type for described current macro.

12. Intra prediction mode selection systems according to claim 10, it is characterized in that, described the first pixel acquisition module is divided into 16 4 * 4 picture element matrix D0～D15 by described current macro, and a pixel that gathers the upper left corner of each picture element matrix in described picture element matrix D0～D15 forms described picture element matrix A0.

13. Intra prediction mode selection systems according to claim 11, it is characterized in that, described the second pixel acquisition module gathers the pixel of first and the 4th pixel column of first and third, five, seven pixel columns of described picture element matrix Bi and the pixel of the 3rd and the 7th pixel column of second, four, six, eight pixel columns of described picture element matrix Bi forms described picture element matrix Ci.

14. Intra prediction mode selection systems according to claim 9, it is characterized in that, described predictive mode selected cell selects one of vertical direction, horizontal direction, lower right and lower left as the prediction direction for described picture element matrix Bi or Dj.

15. Intra prediction mode selection systems according to claim 14, is characterized in that, described predictive mode selected cell comprises:

The 3rd pixel acquisition module, is configured to, when described type of prediction selected cell is selected 8 * 8 infra-frame predictions as infra-frame prediction type for described current macro, gather 16 pixels form 4 * 4 picture element matrix Ci from described picture element matrix Bi;

Calculated for pixel values module, is configured to calculate the poor H11 of four pixel values in four pixel value sums in four pixel values in the upper left corner of the pixel value sum of the first and second pixel columns of described picture element matrix Ci and the pixel value sum of the first and second pixel columns of poor H01, described picture element matrix Ci between the pixel value sum of the third and fourth pixel column and the poor H10 between the pixel value sum of the third and fourth pixel column and described picture element matrix Ci and the lower right corner and the upper right corner and four pixel value sums in the lower left corner; And

Set direction Executive Module, is configured to

If | H01| >=2 * | H10| and | H01| >=| H11|, select vertical direction as the prediction direction for described picture element matrix Bi, otherwise

If | H10| >=2 * | H01| and | H10| >=| H11|, selection level direction is as the prediction direction for described picture element matrix Bi, otherwise

Calculate the poor AdditionalValue2 of six pixel value sums in six pixel value sums in the upper left corner of described picture element matrix Bi and the poor AddionalValue1 of six pixel value sums in the lower right corner and the upper right corner of described picture element matrix Bi and six pixel value sums in the lower left corner

If AdditionalValue1 is greater than AdditionalValue2, selects lower left as the prediction direction for described picture element matrix Bi, otherwise select lower right as the prediction direction for described picture element matrix Bi.

16. Intra prediction mode selection systems according to claim 14, is characterized in that, described predictive mode selected cell comprises:

The second calculated for pixel values module, when described type of prediction selected cell is selected 4 * 4 infra-frame predictions as infra-frame prediction type for described current macro, calculate the pixel value sum of the first and second pixel columns of described picture element matrix Dj and the poor H01 between the pixel value sum of the third and fourth pixel column, poor H10 between the pixel value sum of the first and second pixel columns of described picture element matrix Dj and the pixel value sum of the third and fourth pixel column, and the poor H11 of four pixel values in four pixel value sums in four pixel values in the upper left corner of described picture element matrix Dj and the lower right corner and the upper right corner and four pixel value sums in the lower left corner,

Set direction Executive Module, is configured to

If | H01| >=2 * | H10| and | H01| >=| H11|, select vertical direction as the prediction direction for described picture element matrix Dj, otherwise

If | H10| >=2 * | H01| and | H10| >=| H11|, selection level direction is as the prediction direction for described picture element matrix Dj, otherwise

Calculate the poor AdditionalValue2 of six pixel value sums in six pixel value sums in the upper left corner of described picture element matrix Dj and the poor AddionalValue1 of six pixel value sums in the lower right corner and the upper right corner of described picture element matrix Dj and six pixel value sums in the lower left corner

If AdditionalValue1 is greater than AdditionalValue2, selects lower left as the prediction direction for described picture element matrix Dj, otherwise select lower right as the prediction direction for described picture element matrix Dj.