Geometric transforms for image compression

Patent metadata

Field	Value
Publication number	US-11412260-B2
Application number	US-201916662818-A
Country	US
Kind code	B2
Filing date	Oct 24, 2019
Priority date	Oct 29, 2018
Publication date	Aug 9, 2022
Grant date	Aug 9, 2022

Abstract

A decoder may perform a method of decompressing images that include texture features that are not aligned with an axis of the image being compressed. In some example implementation, the method may include receiving a block of geometrically transformed pixel values and performing an inverse geometric transformation on the block of geometrically transformed pixel values to generate a first block of pixel values. The geometrically transformed pixel values represent texture features of an image that are non-parallel with a vertical axis or a horizontal axis of the image and the first block of pixel values being one of a plurality of blocks of the image. The example method may further include generating at least a portion of the image based on the first block of pixel values.

First claim

What is claimed is:

1. A computer-implemented method, comprising:
receiving, by a decoder, a compressed bit stream including geometric transformation values;
performing entropy decoding on the compressed bit stream to define quantized discrete cosine transform (DCT) coefficients;
performing dequantization of the quantized DCT coefficients to define DCT coefficients;
performing inverse discrete cosine transform on the DCT coefficients to generate a block of geometrically transformed pixel values;
performing, by the decoder, an inverse affine transformation on the block of geometrically transformed pixel values to generate a block of pixel values of an image such that a texture feature is restored to an original position in the block of pixel values of the image based on the geometric transformation values,
the block of geometrically transformed pixel values representing the texture feature of a block of the image that is geometrically rotated from being non-parallel to being parallel with a vertical axis or a horizontal axis of the image by applying an affine transformation; and
generating, by the decoder, at least a portion of the image based on the block of pixel values.

2. The computer-implemented method of claim 1 , wherein the geometric transformation values are stored on a basis of a tile.

3. The computer-implemented method of claim 2 , wherein a size of the tile is 64×64 pixels.

4. The computer-implemented method of claim 1 , wherein the block of pixel values is a first block of pixel values, the computer-implemented method further comprising:
performing, by the decoder, an inverse color space conversion on the first the block of pixel values to convert the first block of pixel values in a first color space to a second block of pixel values in a second color space.

5. The computer-implemented method of claim 4 , wherein the first color space is a YC b C r color space and the second color space is an RGB color space.

6. A non-transitory computer-readable storage medium having stored thereon computer executable program code which, when executed on a computer system, causes the computer system to perform operations, comprising:
receiving a compressed bit stream including geometric transformation values;
performing entropy decoding on the compressed bit stream to define quantized discrete cosine transform (DCT) coefficients;
performing dequantization of the quantized DCT coefficients to define DCT coefficients;
performing inverse discrete cosine transform on the DCT coefficients to generate a block of geometrically transformed pixel values;
performing an inverse affine transformation on the block of geometrically transformed pixel values to generate a first block of pixel values of an image such that a texture feature is restored to an original position in the first block of pixel values of the image based on the geometric transformation values,
the block of geometrically transformed pixel values representing the texture feature of a block of the image that is geometrically rotated from being misaligned to being aligned with a horizontal axis or a vertical axis of the image by applying an affine transformation;
performing an inverse color space conversion on the first block of pixel values to convert the first block of pixel values that are in a first color space to a second block of pixel values in a second color space; and
generating at least a portion of the image based on the second block of pixel values in the second color space.

7. The non-transitory computer-readable storage medium of claim 6 , the operations further comprising:
receiving, by a decoder, a compressed bit stream of the image; and
performing, by the decoder, entropy decoding on the compressed bit stream of the image to generate a block of quantized DCT coefficients.

8. The non-transitory computer-readable storage medium of claim 6 , wherein the geometric transformation values are stored on a basis of a tile.

9. The non-transitory computer-readable storage medium of claim 8 , wherein a size of the tile is 64×64 pixels.

10. The non-transitory computer-readable storage medium of claim 6 , wherein the first color space is a YC b C r color space and the second color space is an RGB color space.

11. A method, comprising:
receiving a compressed bit stream including geometric transformation values;
performing entropy decoding on the compressed bit stream to define quantized discrete cosine transform (DCT) coefficients;
performing dequantization of the quantized DCT coefficients to define DCT coefficients;
performing inverse discrete cosine transform on the DCT coefficients to generate a block of geometrically transformed pixel values;
performing an inverse affine transformation on the block of geometrically transformed pixel values to generate a block of pixel values of an image such that a texture feature is restored to an original position in the block of pixel values of the image based on geometric transformation values,
the block of geometrically transformed pixel values representing the texture feature of a block of an image that is geometrically rotated from being diagonally aligned, when at the original position within the image, to being vertically or horizontally aligned by applying an affine transformation; and
performing inverse color space conversion on the block of pixel values to convert the block of pixel values from a first color space to a second color space.

12. The method of claim 11 , further comprising:
receiving, by a decoder, a compressed bit stream of the image; and
performing, by the decoder, entropy decoding on the compressed bit stream of the image to generate a block of quantized DCT coefficients.

13. The method of claim 11 , wherein the geometric transformation values are stored on a basis of a tile.

14. The method of claim 13 , wherein a size of the tile is 64×64 pixels.

15. The computer-implemented method of claim 1 , wherein the horizontal axis includes an x-axis of the image.

16. The computer-implemented method of claim 1 , wherein the vertical axis includes a y-axis of the image.

Assignees

• Google Llc

Inventors

• Alakuijala Jyrki Antero
• Wassenberg Jan

Classifications

• G06T3/60

  Rotation of whole images or parts thereof · CPC title

• H04N19/625 Primary

  using discrete cosine transform [DCT] · CPC title

• H04N19/176

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

• H04N19/14

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