Method for image processing and apparatus for implementing the same

Patent metadata

Field	Value
Publication number	US-12537951-B2
Application number	US-202217975904-A
Country	US
Kind code	B2
Filing date	Oct 28, 2022
Priority date	Oct 29, 2021
Publication date	Jan 27, 2026
Grant date	Jan 27, 2026

Abstract

A method of processing an image, wherein the image is divided into a plurality of blocks is provided, which comprises, for a block of the plurality of blocks: determine prediction residual data representing prediction encoding data for pixels of the block; determine coefficients based on the prediction residual data; based on a number of non-null coefficients among the coefficients, the number of non-null coefficients being at least one, determine a processing to be performed on the prediction residual data for generating a binary flow of encoded pixels of the block.

First claim

The invention claimed is:

1 . A method of processing an image, wherein the image is divided into a plurality of blocks, the method comprising, for a block of the plurality of blocks:
determining prediction residual data representing prediction encoding data for pixels of the block; determining coefficients based on the prediction residual data, wherein the determining of the coefficients comprises: determine first transform coefficients by applying a first transform to the prediction residual data; based on a number of non-null coefficients among the coefficients, the number of non-null coefficients being at least one, determining a processing to be performed on the prediction residual data for generating a binary flow of encoded pixels of the block, wherein the determining the processing comprises: based on a first number of non-null transform coefficients among the first transform coefficients being at least one, select a transform to be performed on the prediction residual data for generating a binary flow of encoded pixels of the block,
wherein the method further comprises:
performing at least one iteration of a transform test loop, an Nth iteration of the transform test loop comprising, N being a transform test loop counter initialized to one;
determining Nth test loop transform coefficients by applying a Nth test loop transform to the prediction residual data;
wherein the transform to be used for generating a binary flow of encoded pixels of the block is further based on a Nth number of non-null test loop transform coefficients among the Nth test loop transform coefficients, the Nth number of non-null test loop transform coefficients being at least one, and
wherein the Nth transform is of a type different from a type of the first transform and a type of the ith transform used during any previous iteration of the transform test loop.

2 . The method according to claim 1 , further comprising:
determining image processing context data based on a context for processing the image; determining a threshold based on the image processing context data; wherein the processing is determined based on a comparison of the number of non-null coefficients with the threshold.

3 . The method according to claim 1 , further comprising: based on the first number of non-null transform coefficients being smaller than a first transform coefficient threshold, using the first transform coefficients for generating transformed prediction residual data.

4 . The method according to claim 3 , further comprising: based on the first number of non-null transform coefficients being greater than or equal to the first transform coefficient threshold and smaller than a second transform coefficient threshold, testing transforms in a predefined first set of transforms, and selecting the transform to be performed on the prediction residual data for generating a binary flow of encoded pixels of the block among the first transform and transforms of the first set of transforms.

5 . The method according to claim 3 , further comprising: based on the first number of non-null transform coefficients being greater than a third transform coefficient threshold, testing transforms in a predefined second set of transforms, and selecting the transform to be performed on the prediction residual data for generating a binary flow of encoded pixels of the block among the first transform and transforms of the second set of transforms, wherein the second set of transforms comprises an identity transform.

6 . The method according to claim 1 , further comprising:
based on each of the first number of the non-null coefficients and the N th number of the non-null test loop transform coefficients being smaller than the first transform coefficient threshold, selecting the transform to be performed on the prediction residual data for generating a binary flow of encoded pixels of the block in a set of tested transforms comprising the first transform and transforms used during executed iterations of the transform test loop.

7 . The method according to claim 1 , further comprising: based on each of the first number of the non-null coefficients and the Nth number of the non-null test loop transform coefficients not being smaller than the first transform coefficient threshold, performing an additional iteration of the transform test loop.

8 . The method according to claim 1 , wherein the first transform is a bi-dimensional Direct Cosine Transform (DCT) transform.

9 . An image processing apparatus comprising a processor, and a memory operatively coupled to the processor, wherein the image processing apparatus is configured to perform a method of processing an image, wherein the image is divided into a plurality of blocks, the method comprising, for a block of the plurality of blocks:
determining prediction residual data representing prediction encoding data for pixels of the block; determining coefficients based on the prediction residual data, wherein the determining of the coefficients comprises: determine first transform coefficients by applying a first transform to the prediction residual data; based on a number of non-null coefficients among the coefficients, the number of non-null coefficients being at least one, determining a processing to be performed on the prediction residual data for generating a binary flow of encoded pixels of the block, wherein the determining the processing comprises: based on a first number of non-null transform coefficients among the first transform coefficients being at least one, select a transform to be performed on the prediction residual data for generating a binary flow of encoded pixels of the block,
wherein the method further comprises:
performing at least one iteration of a transform test loop, a Nth iteration of the transform test loop comprising, N being a transform test loop counter initialized to one;
determining Nth test loop transform coefficients by applying a Nth test loop transform to the prediction residual data;
wherein the transform to be used for generating a binary flow of encoded pixels of the block is further based on a Nth number of non-null test loop transform coefficients among the Nth test loop transform coefficients, the Nth number of non-null test loop transform coefficients being at least one, and
wherein the Nth transform is of a type different from a type of the first transform and a type of the ith transform used during any previous iteration of the transform test loop.

10 . The image processing apparatus according to claim 9 , wherein the method further comprises:
determining image processing context data based on a context for processing the image; determining a threshold based on the image processing context data; wherein the processing is determined based on a comparison of the number of non-null coefficients with the threshold.

11 . The image processing apparatus according to claim 9 , wherein the method further comprises: based on the first number of non-null transform coefficients being smaller than a first transform coefficient threshold, use the first transform coefficients for generating transformed prediction residual data.

12 . The image processing apparatus according to claim 11 , wherein the method further comprises: based on the first number of non-null transform coefficients being greater than or equal to the first transform coefficient threshold and smaller than a second transform coefficient threshold, test transforms in a predefined first set of transforms, and select the transform to be performed on the prediction residual data for generating a binary flow of encoded pixels of the block among the first

Assignees

• Ateme

Inventors

• Gisquet Christophe

Classifications

• H04N19/625

  using discrete cosine transform [DCT] · CPC title

• H04N19/18

  the unit being a set of transform coefficients · CPC title

• H04N19/176

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

• H04N19/124 Primary

  Quantisation · CPC title

• H04N19/12 Primary

  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