Content adaptive transform coding for next generation video

Patent metadata

Field	Value
Publication number	US-9819965-B2
Application number	US-201314129204-A
Country	US
Kind code	B2
Filing date	Oct 29, 2013
Priority date	Nov 13, 2012
Publication date	Nov 14, 2017
Grant date	Nov 14, 2017

Abstract

Techniques related to applying content adaptive and fixed transforms to prediction error data partitions for coding video are discussed. Such techniques may include applying content adaptive transforms having content dependent basis functions to small to medium sized prediction error data partitions and fixed transforms having fixed basis functions to medium to large sized prediction error data partitions.

First claim

What is claimed:

1. A computer-implemented method for video coding, comprising:
receiving a prediction error data partition corresponding to input video for transform coding;
partitioning the prediction error data partition into a plurality of coding partitions of the prediction error data partition including a first subset of the plurality of coding partitions and a second subset of the plurality of coding partitions, wherein the first subset of the plurality of coding partitions comprises one or more coding partitions having a height of less than or equal to 16 pixels and a width of less than or equal to 16 pixels and the second subset of the plurality of coding partitions comprises one or more coding partitions having a height of greater than or equal to 16 pixels and a width greater than or equal to 16 pixels;
performing a content adaptive parametric transform on a first coding partition of the first subset of the plurality of coding partitions in response to the first coding partition of the first subset having a height of less than or equal to 16 pixels and a width of less than or equal to 16 pixels, wherein the content adaptive parametric transform comprises content dependent basis functions determined based on decoded video data neighboring the first coding partition;
performing a fixed transform on a second coding partition of the second subset of the plurality of coding partitions in response to the second coding partition of the second subset having a height of greater than or equal to 16 pixels and a width greater than or equal to 16 pixels, wherein the fixed transform comprises fixed basis functions;
encoding a bitstream with quantized transform coefficients corresponding to the content adaptive transformed first subset of coding partitions and the fixed transformed second subset of coding partitions; and
transmitting the encoded bitstream.

2. The method of claim 1 , wherein partitioning the prediction error data partition into the plurality of coding partitions comprises partitioning the prediction error data partition using a bi-tree partitioning technique.

3. The method of claim 1 , wherein the fixed transform comprises at least one of a discrete cosine transform or a discrete cosine transform approximator.

4. The method of claim 1 , wherein the content adaptive parametric transform comprise one of a parametric Haar transform, a hybrid parametric Haar transform, a parametric slant transform, or a hybrid parametric slant transform.

5. The method of claim 1 , wherein partitioning the prediction error data partition comprises partitioning the prediction error data partition into at least one coding partition of 16 by 16 pixels, the method further comprising:
selecting whether to encode the coding partition of 16 by 16 pixels based on a second content adaptive parametric transform or the fixed transform, wherein encoding the bitstream comprises encoding the bitstream with quantized transform coefficients corresponding to the selected second content adaptive parametric transform or fixed transform of the coding partition of 16 by 16 pixels.

6. The method of claim 1 , further comprising:
determining a picture type and a prediction type associated with the prediction error data partition, wherein the picture type comprises at least one of an F/B-picture or a P-picture, wherein the prediction type comprises inter prediction, and wherein partitioning the prediction error data partition into the plurality of coding partitions comprises partitioning the prediction error data partition using a bi-tree partitioning technique.

7. The method of claim 1 , further comprising:
receiving a tile for coding;
determining a picture type associated with the tile, wherein the picture type comprises an I-picture;
partitioning the tile using a k-d tree partitioning technique to generate a plurality of partitions for prediction;
performing intra-prediction to generate a plurality of prediction partitions associated with the plurality of partitions for prediction;
differencing the plurality of prediction partitions with original pixel data associated with the plurality of prediction partitions to generate a plurality of second prediction error data partitions;
performing a second content adaptive parametric transform on a third subset of the plurality of second prediction error data partitions; and
performing a second fixed transform on a fourth subset of the plurality of second prediction error data partitions.

8. A video encoder comprising:
an image buffer;
a processor communicatively coupled to the image buffer and configured to:
receive a prediction error data partition corresponding to input video for transform coding;
partition the prediction error data partition into a plurality of coding partitions of the prediction error data partition including a first subset of the plurality of coding partitions and a second subset of the plurality of coding partitions, wherein the first subset of the plurality of coding partitions comprises one or more coding partitions having a height of less than or equal to 16 pixels and a width of less than or equal to 16 pixels and the second subset of the plurality of coding partitions comprises one or more coding partitions having a height of greater than or equal to 16 pixels and a width greater than or equal to 16 pixels;
perform a content adaptive parametric transform on the first subset of the plurality of coding partitions in response to the one or more coding partitions of the first subset having a height of less than or equal to 16 pixels and a width of less than or equal to 16 pixels, wherein the content adaptive parametric transform comprises content dependent basis functions determined from decoding neighboring video data;
perform a fixed transform on the second subset of the plurality of coding partitions in response to the one or more coding partitions of the second subset having a height of greater than or equal to 16 pixels and a width greater than or equal to 16 pixels, wherein the fixed transform comprises fixed basis functions; and
encode a bitstream with quantized transform coefficients corresponding to the content adaptive transformed first subset of coding partitions and the fixed transformed second subset of coding partitions; and

a transmitter configured to transmit the encoded bitstream.

9. The video encoder of claim 8 , wherein to partition the prediction error data partition comprises the processor being configured to partition the prediction error data partition using a bi-tree partitioning.

10. The video encoder of claim 8 , wherein the fixed transform comprises at least one of a discrete cosine transform or a discrete cosine transform approximator and the content adaptive parametric transform comprise one of a parametric transform, a parametric Haar transform, a hybrid parametric Haar transform, a parametric slant transform, or a hybrid parametric slant transform.

11. The video encoder of claim 8 , wherein the processor is further configured to:
receive a tile for coding; and
partition the tile using a k-d tree partitioning technique to generate a plurality of partitions for prediction, and wherein to partition the prediction error data partition comprises the processor to partition the prediction error data partition using a bi-tree partitioning technique.

12. The video encoder of claim 8 , wherein the processor is further configured to:
receive a tile for coding; and
partition the tile using a bi-tree partitioning technique to generate a plurality of partitions for prediction, wherein to partition the prediction error data partition comprises the processor being configured to partition the predicti

Assignees

• Intel Corp

Inventors

• Puri Atul
• Gokhale Neelesh N

Classifications

• H04N19/12

  Selection from among a plurality of transforms or standards, e.g. selection between discrete cosine transform [DCT] and sub-band transform or selection between H.263 and H.264 · CPC title

• H04N19/44

  Decoders specially adapted therefor, e.g. video decoders which are asymmetric with respect to the encoder · CPC title

• H04N19/172

  the region being a picture, frame or field · CPC title

• H04N19/91

  Entropy coding, e.g. variable length coding [VLC] or arithmetic coding · CPC title

• H04N19/105

  Selection of the reference unit for prediction within a chosen coding or prediction mode, e.g. adaptive choice of position and number of pixels used for prediction · CPC title