Transform-domain intra prediction

Patent metadata

Field	Value
Publication number	US-9344742-B2
Application number	US-201213572013-A
Country	US
Kind code	B2
Filing date	Aug 10, 2012
Priority date	Aug 10, 2012
Publication date	May 17, 2016
Grant date	May 17, 2016

Abstract

A method for encoding a frame having a plurality of blocks in a video stream using transform-domain intra prediction is disclosed. The method includes generating, using a two-dimensional transform, a set of transform coefficients for a current block; generating, using a one-dimensional transform, a set of transform coefficients for previously coded pixel values in the frame; determining, using the set of transform coefficients for the previously coded pixel values, a set of transform coefficients for a prediction block; determining a residual based on the difference between the set of transform coefficients for the current block and the set of coefficients for the prediction block; and encoding the residual.

First claim

What is claimed is:

1. A method for encoding a frame in a video stream using a computing device, the frame having a plurality of blocks including a current block, the method comprising:
generating, using a two-dimensional transform, a set of transform coefficients for the current block;
generating, using a one-dimensional transform, a set of transform coefficients for a plurality of previously coded pixel values in the frame, wherein the set of transform coefficients for the plurality of previously coded pixel values includes transform coefficients for a row above and immediately adjacent to the current block and transform coefficients for a column to the left and immediately adjacent to the current block;
determining, using the set of transform coefficients for the previously coded pixel values, a set of transform coefficients for a transform-domain prediction block;
determining a transform-domain residual based on a difference between the set of transform coefficients for the current block and the set of transform coefficients for the transform-domain prediction block, wherein DC indicates a DC coefficient for the set of transform coefficients for the prediction block; Kr indicates a first scalar value; Kc indicates a second scalar value; CT0 indicates a DC coefficient for the row above and immediately adjacent to the current block; CT1 indicates an AC coefficient for the row above and immediately adjacent to the current block; CT2 indicates an AC coefficient for the row above and immediately adjacent to the current block; CT3 indicates an AC coefficient for the row above and immediately adjacent to the current block; CL0 indicates a DC coefficient for the column left and immediately adjacent to the current block; CL1 indicates an AC coefficient for the column left and immediately adjacent to the current block; CL2 indicates an AC coefficient for the column left and immediately adjacent to the current block; CL3 indicates an AC coefficient for the column left and immediately adjacent to the current block; * indicates multiplication; and wherein determining the set of transform coefficients for the prediction block includes using the following matrix: [DC, Kr*CT1, Kr*CT2, Kr*CT3; Kc*CL1, 0, 0, 0; Kc*CL2, 0, 0, 0; Kc*CL3, 0, 0, 0];
encoding a bitstream that includes the transform-domain residual; and
performing, by the computing device, at least one of storing the encoded bitstream or transmitting the encoded bitstream.

2. The method of claim 1 , wherein Kr is equal to one of −1, 0, 1 and 2 and wherein Kc is equal to one of −1, 0, 1 and 2.

3. The method of claim 2 , wherein DC is equal to one of 2*CTO, 2*CL0, or CT0*CL0.

4. The method of claim 1 , wherein the one-dimensional transform is a 4-point discrete cosine transform and wherein the two-dimensional transform is a 4×4 discrete cosine transform.

5. The method of claim 1 , wherein the set of transform coefficients for the transform-domain prediction block represent a prediction mode that is not equivalent to any spatial-domain prediction mode.

6. A method for decoding a frame in an encoded video stream using a computing device, the frame having a plurality of blocks including a current block, the method comprising:
decoding a transform-domain residual for the current block from the encoded video stream;
generating, using a one-dimensional transform, a set of transform coefficients for a plurality of previously decoded pixel values in the frame, wherein the set of transform coefficients for the plurality of previously coded pixel values includes transform coefficients for a row above and immediately adjacent to the current block and transform coefficients for a column to the left and immediately adjacent to the current block;
generating, using the set of transform coefficients for the previously decoded pixel values, a set of transform coefficients for a transform-domain prediction block;
determining a set of transform coefficients for the current block based on a sum of the transform-domain residual and the set of transform coefficients for the transform-domain prediction block, wherein DC indicates a DC coefficient for the set of transform coefficients for the prediction block; Kr indicates a first scalar value; Kc indicates a second scalar value; CT0 indicates a DC coefficient for the row above and immediately adjacent to the current block; CT1 indicates an AC coefficient for the row above and immediately adjacent to the current block; CT2 indicates an AC coefficient for the row above and immediately adjacent to the current block; CT3 indicates an AC coefficient for the row above and immediately adjacent to the current block; CL0 indicates a DC coefficient for the column left and immediately adjacent to the current block; CL1 indicates an AC coefficient for the column left and immediately adjacent to the current block; CL2 indicates an AC coefficient for the column left and immediately adjacent to the current block; CL3 indicates an AC coefficient for the column left and immediately adjacent to the current block; * indicates multiplication; and wherein determining the set of transform coefficients for the prediction block includes using the following matrix: [DC, Kr*CT1, Kr*CT2, Kr*CT3; Kc*CL1, 0, 0, 0; Kc*CL2, 0, 0, 0; Kc*CL3, 0, 0, 0];
inverse transforming the set of transform coefficients for the current block using a second transform, wherein the second transform is a two-dimensional transform;
generating a decoded video using the inverse-transformed transform coefficients; and
performing, by the computing device, at least one of storing the decoded video or displaying the decoded video.

7. The method of claim 6 wherein Kr is equal to one of −1, 0, 1 and 2 and wherein Kc is equal to one of −1, 0, 1 and 2.

8. The method of claim 7 , wherein DC is equal to one of 2*CTO, 2*CL0, or CT0*CL0.

9. The method of claim 6 , wherein the one-dimensional transform is a 4-point discrete cosine transform and wherein the two-dimensional transform is a 4×4 discrete cosine transform.

10. The method of claim 6 , wherein the set of transform coefficients for the transform-domain prediction block represent a prediction mode that is not equivalent to any spatial-domain prediction mode.

11. An apparatus for encoding a video stream having at least one frame having a plurality of blocks including a current block, comprising:
a memory; and
a processor configured to execute instructions stored in the memory to:
generate, using a two-dimensional transform, a set of transform coefficients for the current block;
generate, using a one-dimensional transform, a set of transform coefficients for a plurality of previously coded pixel values in the frame, wherein the set of transform coefficients for the plurality of previously coded pixel values includes transform coefficients for a row above and immediately adjacent to the current block and transform coefficients for a column to the left and immediately adjacent to the current block;
determine, using the set of transform coefficients for the previously coded pixel values, a set of transform coefficients for a transform-domain prediction block;
determine a transform-domain residual based on a difference between the set of transform coefficients for the current block and the set of coefficients for the transform-domain prediction block, wherein DC indicates a DC coefficient for the set of transform coefficients for the prediction block; Kr indicates a first scalar value; Kc indicates a second scalar value; CT0 indicates a DC coefficient for the row above and immediately adjacent to the current block; CT1 indicates an AC coefficient for the row above and immediately adjacent to the current block; CT2 indicates an AC coe

Assignees

• Xu Yaowu
• Gu Qunshan
• Google Inc

Inventors

• Xu Yaowu
• Gu Qunshan

Classifications

• H04N19/61

  in combination with predictive coding · CPC title

• H04N19/593 Primary

  involving spatial prediction techniques · CPC title

• H04N19/176

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