Processing device and method of compressing images

Patent metadata

Field	Value
Publication number	US-9307249-B2
Application number	US-201414310005-A
Country	US
Kind code	B2
Filing date	Jun 20, 2014
Priority date	Jun 20, 2014
Publication date	Apr 5, 2016
Grant date	Apr 5, 2016

Abstract

The present application relates an encoder. The encoder partitions an image domain into several substructures each having one of at least one size dimension, and defines at least one geometric primitive for each substructure on the basis of geometry data. The encoder also retrieves, for each substructure, a subset of image data and determines whether pixel values of the retrieved subset are the same. If the pixel values are describable by a texture mapping operation, then the encoder defines a compressed texture image and assigns texture mapping data to the geometry data. Otherwise, the encoder defines an uncompressed texture image and assigns texture mapping data to the geometry data. The compressed image includes the geometry data, the texture mapping data, and the texture image data.

First claim

The invention claimed is:

1. A processing device for compressing an image, said processing device comprises at least a processor operatively coupled to a memory and an encoder, wherein said memory is configured to provide data of an image and at least one size dimension to the encoder,
wherein said encoder is configured to receive said data of said image and said at least one size dimension from said memory and comprises a geometry engine, which is further configured
to partition the image domain into several substructures each having one of the at least one size dimension; and
to define at least one geometric primitive for each substructure on the basis of geometry data defining a positioning of the at least one geometric primitive within the image space, wherein said at least one geometric primitive represents geometrically the substructure,

wherein said encoder further comprises a texture engine, which is configured for each substructure
to retrieve a subset of data of the image corresponding to the substructure;
to determine whether pixels of the retrieved subset are reconstructable from one or more selected pixels as input parameters on the basis of a texture mapping operation in the pixel value space, wherein the selected pixels are representative of the subset of image data and input parameters to the texture mapping operation in pixel value space;
if the pixels of the retrieved subset are reconstructable to define data of a texture image in compressed form with the selected pixels;
otherwise to define data of a texture image in uncompressed form, which data corresponds to the retrieved subset of image data; and
to assign texture mapping data to the geometry data of the substructure for mapping the defined texture image onto the at least one geometric primitive of the substructure,

wherein a compressed form of said image comprises the geometry data, the texture mapping data and the texture image data for each substructure.

2. The processing device of claim 1 , wherein said encoder is further configured
to determine whether the pixel values of the retrieved subset are at least similar with respect to a distance in color space, and
to define the data of the texture image in compressed form, which has one selected pixel with a pixel value representative of the similar pixel values of the retrieved subset.

3. The processing device of claim 1 , wherein said encoder is further configured
to determine whether the pixel values of the retrieved subset change in accordance with a color gradient within a predefined variation range in color space, and
to define the data of the texture image in compressed form, the data of which has selected pixels with values representative of initial gradient values and final gradient values.

4. The processing device of claim 1 , wherein said encoder is further configured
to define a set of vertices for the at least one geometric primitive of each substructure,
wherein each vertex comprises an image coordinate defining a positioning within image space, wherein the image coordinates form the geometry data,
wherein each vertex has further assigned a texture coordinate, which defines a positioning within texture image space, wherein the texture coordinates form the texture mapping data.

5. The processing device of claim 1 ,
wherein the at least one geometric primitive defining the substructure comprises two triangle primitives, wherein said two triangle primitives are defined on the basis of four image coordinates, wherein a texture mapping coordinate is assigned to each of the image coordinates.

6. The processing device of claim 1 ,
wherein said substructure is rectangular, wherein said rectangular substructure is defined on the basis of two image coordinates, wherein a texture mapping coordinate is assigned to each of the image coordinates.

7. The processing device of claim 1 ,
wherein said image is a simply connected domain,
wherein each geometric primitive is a simply connected domain,
wherein the domains of the geometric primitive are subdomains of the image domain,
wherein the domains of the geometric primitives are disconnected,
wherein said totality of geometric primitives defines a simply connected domain corresponding to the image domain.

8. The processing device of claim 1 , wherein said encoder further comprises a merger, which is configured
to merge at least two substructures, which are adjacent to each other and which have assigned texture images in uncompressed form or which have assigned texture images in compressed form with the same pixel value.

9. A method of compressing an image, comprising:
providing data of an image;
providing at least one size dimension;
partitioning the image domain into several substructures each having one of the at least one size dimension; and
defining at least one geometric primitive for each substructure on the basis of geometry data defining a positioning of the at least one geometric primitive within the image space, wherein said at least one geometric primitive represents geometrically the substructure,
wherein said method further comprises for each substructure:
retrieving a subset of data of the image corresponding to the substructure;
determining whether pixels of the retrieved subset are reconstructable from one or more selected pixels on the basis of a texture mapping operation in the pixel value space, wherein the selected pixels are representative of the subset of image data and input parameters to the texture mapping operation in pixel value space;
if the pixels of the retrieved subset are reconstructable, then defining data of a texture image in compressed form with the selected pixels;
otherwise defining data of a texture image in uncompressed form, which data corresponds to the retrieved subset of image data; and
assigning texture mapping data to the geometry data of the substructure for mapping the defined texture image onto the at least one geometric primitive of the substructure,

wherein a compressed form of said image comprises the geometry data, the texture mapping data and the texture image data for each substructure.

10. The method of claim 9
determining whether the pixel values of the retrieved subset are at least similar with respect to a distance in color space, and
defining the data of the texture image in compressed form, which has one selected pixel with a pixel value representative of the similar pixel values of the retrieved subset.

11. The method of claim 9
determining whether the pixel values of the retrieved subset change in accordance with a color gradient within a predefined variation range in color space, and
defining the data of the texture image in compressed form, the data of which has selected pixel with values representative of initial gradient values and final gradient values.

12. The method of claim 9 , wherein said partitioning the image domain into several substructures further comprises:
defining a set of vertices for the at least one geometric primitive of each substructure,
wherein each vertex comprises an image coordinate defining a positioning within image space, wherein the image coordinates form the geometry data,
wherein each vertex has further assigned a texture coordinate, which defines a positioning within texture image space, wherein the texture coordinates form the texture mapping data.

13. The method of claim 9 ,
wherein the at least one geometric primitive defining the substructure comprises two triangle primitives, wherein said two triangle primitives are defined on the basis of four image coordinates, wherein a texture mapping coordinate is assigned to e

Assignees

• Krutsch Robert Cristian
• Gancev Valentin-Adrian
• Tewell Thomas Richardson
• Freescale Semiconductor Inc

Inventors

• Krutsch Robert Cristian
• Gancev Valentin-Adrian
• Tewell Thomas Richardson

Classifications

• H04N19/119

  Adaptive subdivision aspects, e.g. subdivision of a picture into rectangular or non-rectangular coding blocks · CPC title

• H04N19/436

  using parallelised computational arrangements · CPC title

• H04N19/167

  Position within a video image, e.g. region of interest [ROI] · CPC title

• G06T9/20

  Contour coding, e.g. using detection of edges · CPC title

• H04N19/14

  Coding unit complexity, e.g. amount of activity or edge presence estimation (H04N19/146 takes precedence) · CPC title