JP2007096679A - Method of encoding and decoding moving picture - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an encoding method of encoding a moving picture that improves encoding efficiency, and a decoding method. <P>SOLUTION: Nine prediction modes of: reverse DC component prediction (prediction mode 11) whose prediction direction is reverse to that of each of prediction modes of conventional technology; orthogonal right upper prediction (prediction mode 12); orthogonal left upper prediction (prediction mode 13); vertical left upper prediction (prediction mode 14); horizontal left upper prediction (prediction mode 15); vertical right upper prediction (prediction 16); and horizontal left lower prediction (prediction 17), including reverse vertical prediction (prediction 9) whose prediction direction is vertically reverse to that of a conventional prediction mode 0 and reverse horizontal prediction (prediction mode 10) whose prediction direction is horizontally reverse to that of a conventional prediction mode 1, are executed in addition to nine conventional prediction modes 0 to 8. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a moving image encoding and decoding method, and more particularly, to a moving image encoding and decoding method suitable for intra-frame prediction encoding and intra-frame prediction decoding.

  Predictive coding methods for moving image data are classified into intra-frame coding that uses correlation between adjacent pixels in the target frame and inter-frame coding that uses correlation between consecutive frames in the moving image. In general, inter-frame coding has higher coding efficiency than intra-frame coding. In recent years, techniques for improving the compression effect by skillfully combining intraframe prediction encoding and interframe prediction encoding have been studied.

  In Patent Document 1, in intra-frame predictive coding of H.264 / AVC (Advanced Video Coding), which is one of the next generation video coding systems, each frame is a block of 4 × 4 pixel size or 16 × 16 pixel size. (Macroblock) is divided into blocks, encoded for each block, a predicted image is generated using pixels adjacent to the target block to be encoded, and a difference value between the predicted image and the original image is converted. The compression efficiency is improved by quantizing. The prediction method used in intra-frame prediction is not one type, but nine types are prepared according to the prediction direction.

  FIG. 8 is a diagram illustrating all prediction modes in a 4 × 4 pixel block of H.264 intra-frame prediction, and FIG. 9 is a diagram schematically illustrating the prediction direction for each prediction mode.

In the prior art, prediction mode 0 (vertical prediction), prediction mode 1 (horizontal prediction), prediction mode 2 (DC component prediction), prediction mode 3 (orthogonal lower left prediction), prediction mode 4 (orthogonal lower right prediction), prediction mode Nine prediction modes of 5 (vertical lower right prediction), prediction mode 6 (horizontal lower right prediction), prediction mode 7 (vertical lower left prediction) and prediction mode 8 (horizontal upper right prediction) are prepared. Symbols A to M are reference pixel signals that have already been encoded. In prediction mode 0 (vertical prediction mode), prediction is performed from reference pixels A, B, C, and D that have already been encoded along the vertical direction. In prediction mode 2 (DC component prediction), average values of reference pixels A to D and J to M are obtained, and all pixels of a 4 × 4 block are predicted based on this average value.
Japanese Patent Laid-Open No. 2005-252679

  The above-described conventional technique is based on the premise that blocks are encoded in a raster order (upper left to lower right) within a frame. Therefore, a block referred to in predictive encoding is located above the target block (upper right and upper left). Including) and the left part. Therefore, even if the lower part and the right part of the target block are used as reference blocks, the encoding efficiency is higher, and these cannot be used as reference blocks.

  An object of the present invention is to solve the above-described problems of the prior art and to provide a moving picture encoding method and decoding method capable of improving the encoding efficiency as compared with the prior art.

  In order to achieve the above object, the moving picture coding method of the present invention includes the following procedure.

  (1) a first procedure in which all macroblocks in a frame are encoded by an intraframe prediction encoding scheme or an interframe prediction encoding scheme, and efficiency representative information representative of the encoding efficiency is registered for each macroblock; A second procedure in which attention is paid to each macroblock in the frame in a predetermined order, and the macroblock of interest is encoded by the intraframe prediction encoding method using the encoded macroblock; and the intraframe prediction encoding method A third procedure for comparing the coding efficiency according to each macroblock with the registered efficiency representative information for each macroblock, and selecting a coding method with a high coding efficiency, and based on the comparison result, the already registered A fourth procedure for updating the efficiency representative information, and a fifth procedure for repeating the second to fourth procedures for all macroblocks. , Right target macroblock, wherein the lower right side, the intraframe prediction encoding scheme as a reference macroblock to macroblock adjacent to one of the lower and left lower runs.

Furthermore, the moving picture decoding method of the present invention includes the following procedure.
(2) First procedure for decoding all macroblocks encoded by at least the interframe predictive encoding method in the frame, and paying attention to each macroblock in the frame in a predetermined order, and intraframe predictive encoding A second procedure for decoding a macroblock that has been encoded in a manner and whose reference macroblock has already been decoded, and the second procedure until all macroblocks in the frame have been decoded. And a third procedure for repeating the procedure.

  (1) According to the intra-frame prediction coding method of the present invention, each macro block is predicted not only in the macro block adjacent to the left side or the upper side thereof but also to the macro block adjacent to the right side or the lower side as a reference block. Since encoding can be performed using an encoding method, when a macro block similar to the target block to be encoded is adjacent to the right side or the lower side of the target block, higher encoding efficiency can be obtained than before. It becomes like this.

  (2) According to the intra-frame predictive decoding method of the present invention, encoded data encoded with higher encoding efficiency than before can be reliably decoded.

  DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, the best embodiment of the present invention will be described in detail with reference to the drawings. FIG. 1 is a diagram showing nine prediction modes newly added in intra-frame (intra) prediction according to the present invention, taking the case of a 4 × 4 pixel block as an example, and FIG. It is the figure shown typically.

  In the present embodiment, the prediction mode 0 includes a prediction mode 9 (reverse vertical prediction) in which the prediction direction is upside down, and the prediction mode 10 (reverse horizontal prediction) in which the prediction mode 1 and the prediction direction are right and left reversed, The prediction modes 11 (inverse DC component prediction), the prediction mode 12 (orthogonal upper right prediction), the prediction mode 13 (orthogonal upper left prediction) whose prediction directions are opposite to those of the conventional prediction modes described with reference to FIGS. Nine prediction modes of prediction mode 14 (vertical upper left prediction), prediction mode 15 (horizontal upper left prediction), prediction mode 16 (vertical upper right prediction) and prediction mode 17 (horizontal lower left prediction) are the conventional nine prediction modes 0 to 9. 8 in addition to the above.

  Again, the symbols A to M are reference pixel signals that have already been encoded. In prediction mode 9 (reverse vertical prediction), prediction is performed from reference pixels A, B, C, and D that have already been encoded along the reverse vertical direction. In the prediction mode 11 (inverse DC component prediction), the average value of the reference pixels A to D and J to M is obtained, and all pixels of the 4 × 4 block are predicted based on the average value.

  In this embodiment, when each macroblock (MB) is intra-coded, in addition to the nine conventional prediction modes 0 to 8 shown in FIGS. 8 and 9, the nine prediction modes shown in FIGS. By predicting 8 to 17 as well, prediction from all directions including the right side, lower right side, lower side and lower left side of the target MB is enabled.

  Next, a moving picture encoding procedure according to the first embodiment of the present invention will be described with reference to the flowchart of FIG.

  In step S1, it is determined whether or not the current encoding target picture (frame) is an I picture that can be encoded independently without dependency on other pictures. If it is not an I picture, the process proceeds to step S2, where all MBs are encoded and compressed by the inter-frame prediction (inter) encoding method, and the efficiency representative information representing the encoding efficiency is, for example, a cost value for each MB. Is calculated. On the other hand, if the picture to be encoded is an I picture, the process proceeds to step S3. In step S3, all MBs are encoded and compressed by intra-frame prediction (intra) encoding using any of prediction modes 0 to 8 as in the prior art, and the cost value is set for each MB. Calculated.

  As described above, when provisional encoding has been completed for all MBs, in step S4, the MB located at the upper left corner of the current encoding target picture is set as the current attention MB. Is done. In step S5, it is determined whether or not the current MB of interest is a reference MB when an intra-encoded MB adjacent to this MB (hereinafter referred to as an intra-MB) is intra-encoded. . If the target MB is a reference MB of any adjacent intra MB, the process proceeds to step S10 described later.

  On the other hand, if the target MB is not a reference MB from any adjacent intra MB, the process proceeds to step S6, where the target MB is the adjacent inter MB encoded by the inter encoding scheme or the intra encoding scheme. Using the encoded intra MB as a reference MB, intra prediction is performed from all directions using all the prediction modes 0 to 17, and a cost value is calculated for each prediction mode.

  FIG. 5 is a diagram schematically representing the determination method in step S5. The arrow in the diagram represents the prediction direction, MB21 is predictively encoded with reference to MB11 adjacent to the right side, and MB32 Indicates that prediction encoding is performed with reference to MB31 adjacent to the upper side.

  In such a state, for example, if the encoding method or prediction direction of MB31 referred to by MB32 is changed, this also affects the prediction value of MB32. Therefore, in this embodiment, in order to avoid such a situation, the procedure of step S6 is executed only for MB22, MB23, MB33, and MB43 that are not referenced from the adjacent intra MB, and the procedure of step S6 is performed for other MBs. Has been prevented from running.

  Returning to FIG. 3, in step S7, the minimum cost value is selected from the cost values calculated for each prediction mode. In step S8, the current minimum cost value selected in step S7 is compared with the already registered minimum cost value for the target MB. If the current cost value is lower than the registered minimum cost value, the current cost value is updated and registered as the minimum cost value for the target MB in step S9, and the encoding information including the prediction mode is updated and registered. The

  In step S10, it is determined whether or not the current attention MB is the terminal MB located at the lower right corner of the picture. If it is not the terminal MB, the target MB is shifted by 1 MB in the raster direction in step S11, and then the process returns to step S5, and the above-described processes are repeated for the next target MB. In contrast, if the current MB of interest is the terminal MB, the process proceeds from step S10 to step S12. In step S12, it is determined whether or not the number of iterations N has reached a predetermined upper limit number Nmax. If the upper limit number Nmax has been reached, the process proceeds to the next picture. If the number of iterations N has not reached the upper limit number Nmax, the number of iterations N is incremented in step S13, and then the process returns to step S4, and the above-described processes are repeated.

  FIG. 6 is a diagram schematically representing the function of the iterative process executed in steps S12 and S13.

  When the prediction direction at the time when the m-th (m <Nmax) iteration is completed is as shown in FIG. 5, for example, MB33 is higher than the cost value when referring to MB43 before update adjacent to the right side. Since the cost value when referring to MB32 adjacent to is smaller, MB32 is used as the reference MB. However, when MB43 is subsequently updated, in the next (m + 1) -th iterative process, MB33 is more costly when referring to the updated MB43 adjacent to the right side. May be smaller than the registered cost value when referring to. Therefore, in the present embodiment, the cost value can be further reduced by providing the above iterative procedure.

  Next, a moving picture decoding procedure according to the second embodiment of the present invention will be described with reference to the flowchart of FIG.

  In step S21, side information, DCT coefficients, and the like regarding all MBs of the current decoding target picture are acquired. At this time, if the MB is an intra MB, a prediction mode and a prediction direction are acquired as side information. In step S22, based on the side information, it is determined whether or not the current decoding target picture is an intra picture in which all MBs are intra-encoded. If it is an intra picture, the process proceeds to step S23, and only the MB located at the upper left end is intra decoded. If it is an inter picture, the process proceeds to step S24, and only MBs that are inter-coded (hereinafter referred to as inter MBs) out of all MBs are selectively inter-decoded.

  In step S25, the MB located at the upper left corner is set as the current MB of interest. In step S26, it is determined whether or not the current attention MB is an intra MB that has not been decoded yet. If the intra MB is not decoded, the process proceeds to step S27, and the prediction mode is determined based on the side information. In step S28, it is determined whether or not all reference MBs necessary for decoding the target MB have already been decoded. If it has been decoded, the process proceeds to step S29, and the MB of interest is intra-decoded according to the prediction mode.

  In step S30, it is determined whether or not the current attention MB is the terminal MB located at the lower right corner. If it is not the end MB, the target MB is shifted by 1 MB in the raster direction in step S31, and then the process returns to step S26, and the above-described processes are repeated for the next target MB. In contrast, if the current MB of interest is the terminal MB, the process proceeds from step S30 to step S32. In step S32, it is determined whether or not all MBs have been decoded. If there is an MB that has not been decoded, the process returns to step S25 and the above-described processes are repeated from the upper left MB.

  FIG. 7 is a diagram schematically showing how each MB is decoded step by step by repeating the decoding process in step S32, and the hatched MB is a decoded MB. And the arrow indicates the prediction direction.

  In the m-th process, MB23 cannot be decrypted because reference MB33 is not decrypted. Similarly, MB33 cannot be decrypted because reference MB43 is not decrypted. However, MB43 is decrypted after that in the m-th process. In the (m + 1) -th process, MB23 cannot be decrypted because reference MB33 is not decrypted, but MB33 is decrypted because reference MB43 was newly decrypted in the previous (m-th) process. The In the (m + 2) -th process, MB23 is decrypted because reference MB33 is decrypted in the previous process.

It is the figure which showed each prediction mode of the prediction in a flame | frame (4x4 pixel block) in this invention. It is the figure which showed the prediction direction of each prediction mode of the prediction in a flame | frame (4x4 pixel block) in this invention. It is the flowchart which showed the intra-frame prediction encoding procedure of this invention. It is the flowchart which showed the intra-frame prediction decoding procedure of this invention. It is the figure which represented typically the determination method in step S5 of FIG. It is the figure which represented typically the function of the iterative process performed by step S12 of FIG. 3, and S13. It is the figure which represented typically a mode that each MB was decoded in steps by repeating a decoding process by FIG.4 S32. It is the figure which showed all the prediction modes of the intra-frame prediction (4x4 pixel block) of H.264. It is the figure which showed the prediction direction of all the prediction modes of the intra-frame prediction (4x4 pixel block) of H.264.

Claims (6)

In a moving picture coding method in which each frame of a moving picture is divided into a plurality of macro blocks, and each macro block is coded by an intra-frame predictive coding scheme or an inter-frame predictive coding scheme.
A first procedure in which all macroblocks in a frame are encoded by an intraframe predictive encoding scheme or an interframe predictive encoding scheme, and efficiency representative information representative of the encoding efficiency is registered for each macroblock;
A second procedure in which attention is paid to each macroblock in the frame in a predetermined order, and the macroblock of interest is encoded by the intraframe prediction encoding method using the encoded macroblock;
A third procedure for comparing the coding efficiency according to the intra-frame prediction coding method and the registered efficiency representative information for each macroblock for each macroblock, and selecting a coding method having a high coding efficiency;
A fourth procedure for updating the registered efficiency representative information based on the comparison result;
A fifth procedure that repeats the second to fourth procedures for all macroblocks,
In the second procedure, an intra-frame prediction encoding scheme is executed using a macroblock adjacent to any of the right side, lower right side, lower side, and lower left side of the target macroblock as a reference macroblock. Image coding method.   The moving image encoding method according to claim 1, further comprising a sixth procedure in which the fifth procedure is repeated a predetermined number of times.   3. The moving picture encoding method according to claim 1, wherein the fifth procedure is skipped when the macro block of interest is a reference macro block of an intra-frame prediction encoding scheme. In a video decoding method in which each frame is divided into a plurality of macroblocks, and each macroblock is decoded by a video encoded by an intra-frame prediction encoding scheme or an inter-frame prediction encoding scheme.
A first procedure for decoding all macroblocks encoded in at least an interframe predictive encoding method in a frame;
A second macro block is a macro block which is focused on each macro block in the frame in a predetermined order and is encoded by the intra frame predictive encoding method and whose reference macro block is already decoded. Procedure and
And a third procedure for repeating the second procedure until all macroblocks in the frame are decoded.   In the first procedure, when a macro block encoded by the intra-frame prediction encoding method and a macro block encoded by the inter-frame prediction encoding method are mixed in the frame, the inter-frame prediction encoding is performed. 5. The moving picture decoding method according to claim 4, wherein only the macroblock encoded by the method is decoded.   In the first procedure, when only a macroblock encoded by an intraframe predictive encoding method is included in a frame, only a macroblock at a predetermined position is decoded. 5. The moving picture decoding method according to 4.

Images