Method and apparatus for image encoding using hold-MBs, and method and apparatus for image decoding using hold-MBs

Patent metadata

Field	Value
Publication number	US-9036702-B2
Application number	US-80764010-A
Country	US
Kind code	B2
Filing date	Sep 10, 2010
Priority date	Sep 10, 2009
Publication date	May 19, 2015
Grant date	May 19, 2015

Abstract

In video/image coding, macroblocks (MBs) are often intra coded in raster scan order, starting from one seed MB. The invention improves intra prediction for optimized usage of multi-core processors. Encoding starts from multiple intra coded seed MBs per frame, and continues with adjacent MBs. A predefined prediction spread pattern (growth pattern) is used that comprises simultaneous prediction of two or more adjacent MBs per MB. Adjacent MBs with high coding cost are called “Hold-MB” and can be held from being processed, until another neighboring edge is available for prediction. Encoding comprises marking a MB with high coding cost for deferred prediction. This MB is skipped while the other MBs are encoded. When a further adjacent MB was encoded, the marked MB is predicted based on the adjacent MBs. Since a decoder receives the deferment marks and uses the same growth pattern, it follows the encoder, and predicts and decodes correctly.

First claim

What is claimed is:

1. A method for encoding image data, comprising:
predicting macroblocks from respective adjacent macroblocks, wherein multiple macroblocks that are adjacent up, down, left or right to a previously encoded macroblock are predicted simultaneously, and wherein the predicting starts from at least one start point macroblock, and wherein the prediction direction for any one macroblock is selected according to a predefined growth pattern within a current slice to be at least one of left and up and at least two of the multiple simultaneously predicted macroblocks that are adjacent to a previously encoded macroblock use different prediction directions within said current slice, the growth pattern defining in which spatial direction the prediction spreads;
determining for at least one macroblock that the coding cost according to the selected prediction direction is higher than a threshold;
upon said determining, marking said at least one macroblock for deferred prediction and skipping the encoding of said at least one macroblock marked for deferred prediction;
continuing with encoding one or more macroblocks of said current slice of the image, except said at least one macroblock marked for deferred prediction, comprising determining residual data according to said predicting, and encoding the residual data;
predicting said at least one marked macroblock for deferred prediction after at least one further macroblock of said current slice was encoded, the at least one further macroblock being adjacent to said at least one marked macroblock for deferred prediction; and
encoding the marked macroblocks of the image, wherein the resulting encoding data include said mark.

2. The method according to claim 1 , further comprising, before said determining for a current macroblock that an adjacent macroblock was encoded, determining that the adjacent macroblock was marked for deferred prediction and that the current macroblock uses the same prediction direction as the adjacent macroblock, and upon said determining, marking the current macroblock also for deferred prediction without determining its coding cost.

3. The method according to claim 1 , further comprising calculating said coding cost threshold according to the current image, wherein if the coding cost threshold is derived on macroblock level, it is calculated from a previous coding cost threshold and the rate-distortion cost of the previous macroblock.

4. The method according to claim 1 , wherein the at least one start point macroblock is not on the edge of the image.

5. The method according to claim 1 , further comprising entropy encoding said encoded image data, and entropy decoding said entropy encoded image data before said image decoding.

6. The method according to claim 1 , wherein the encoding or decoding starts from two or more start point macroblocks within said current slice simultaneously.

7. The method according to claim 1 , wherein the predefined growth pattern is used for selecting the next macroblocks from said current slice for the predicting, and wherein the predefined growth pattern comprises selecting, for a current macroblock, three adjacent macroblocks in different of up, down, left and right directions.

8. A method for decoding encoded image data, comprising:
extracting data for at least one start point macroblock, the macroblock belonging to a current slice;
extracting for one or more further macroblocks residual data, the one or more further macroblocks also belonging to the current slice;
extracting for at least one of the further macroblocks an indication being a mark for deferred prediction;
decoding said start point macroblock and further macroblocks that are adjacent up, down, left or right to the start point macroblock and belong to the current slice, wherein the decoding starts from said start point macroblock and proceeds in the order of a predefined growth pattern that includes prediction directions of at least one of left and up, wherein the growth pattern defines in which spatial direction the prediction spreads, the decoding comprising predicting a plurality of said further macroblocks simultaneously from one or more already decoded adjacent macroblocks according to the growth pattern, wherein at least two macroblocks of the plurality of further macroblocks are predicted from any particular one previously decoded macroblock that belongs to the current slice using different prediction directions, and adding said residual data, wherein the predicting of said at least one macroblock marked for deferred prediction is skipped;
determining that, for said at least one macroblock marked for deferred prediction, a further adjacent macroblock of the current slice was decoded; and
upon said determining, decoding the at least one marked macroblock according to its residual data and the at least two available decoded adjacent macroblocks.

9. The method according to claim 8 , wherein the at least one start point macroblock is not on the edge of the image.

10. The method according to claim 8 , further comprising entropy encoding said encoded image data, and entropy decoding said entropy encoded image data before said image decoding.

11. The method according to claim 8 , wherein the encoding or decoding starts from two or more start point macroblocks within said current slice simultaneously.

12. The method according to claim 8 , wherein the predefined growth pattern is used for selecting the next macroblocks for the predicting, and wherein the predefined growth pattern comprises selecting, for a current macroblock, three adjacent macroblocks within said current slice and in different of up, down, left and right direction.

13. An apparatus for encoding image data, comprising:
first prediction means for predicting a plurality of macroblocks belonging to a current slice simultaneously from respective adjacent macroblocks also belonging to the current slice, wherein at least two macroblocks that are adjacent up, down, left or right to a previously encoded macroblock are predicted simultaneously, and wherein the predicting starts from at least one start point macroblock and wherein the prediction direction for any one macroblock is selected according to a predefined growth pattern that comprises prediction in at least one of left and up directions, wherein the growth pattern defines in which spatial direction the prediction spreads and wherein in simultaneously predicting the at least two macroblocks that are adjacent to a previously encoded macroblock different prediction directions are used;
first determining means for determining for at least one macroblock that the encoding cost according to the selected prediction direction is higher than a threshold;
marking means for marking, upon said determining, said at least one macroblock for deferred prediction;
first encoder means for encoding the macroblocks of the image, except said at least one macroblock marked for deferred prediction, comprising residual determining means for determining residual data according to said predicting, and residual encoding means for encoding the residual data, wherein the first encoder skips the at least one macroblock that is marked for deferred prediction;
second determining means for determining that at least one further adjacent macroblock of said at least one marked macroblock was encoded, said at least one further adjacent macroblock belonging to said current slice;
second prediction means for predicting said at least one marked macroblock after at least one further adjacent macroblock of said at least one marked macroblock and belonging to the current slice was encoded; and
second encoding m

Assignees

• Yang Jiheng
• Zhang Hui
• Li Yumeng
• Thomson Licensing

Inventors

• Yang Jiheng
• Zhang Hui
• Li Yumeng

Classifications

• H04N19/176

  the region being a block, e.g. a macroblock · CPC title

• H04N19/174

  the region being a slice, e.g. a line of blocks or a group of blocks · CPC title

• H04N19/46

  Embedding additional information in the video signal during the compression process (H04N19/517, H04N19/68, H04N19/70 take precedence) · CPC title

• H04N19/147

  according to rate distortion criteria (rate-distortion as a criterion for motion estimation H04N19/567) · CPC title

• H04N19/61

  in combination with predictive coding · CPC title