Video encoding and decoding methods based on scale and angle variation information, and video encoding and decoding apparatuses for performing the methods

Patent metadata

Field	Value
Publication number	US-9693076-B2
Application number	US-201514591417-A
Country	US
Kind code	B2
Filing date	Jan 7, 2015
Priority date	Jan 7, 2014
Publication date	Jun 27, 2017
Grant date	Jun 27, 2017

Abstract

A video encoding and decoding methods and apparatuses for performing the methods thereof are provided. The video encoding method includes: determining a region in a current image having a change in at least one of a scale and an angle with respect to a reference image; generating variation information representing a variation in at least one of the scale and the angle between the region in the current image and a reference region in the reference image corresponding to the region in the current image; and encoding residual information representing a difference between a predicted sample value from the reference image and a sample value of the current image based on the generated variation information.

First claim

What is claimed is:

1. A video encoding method comprising:
determining, by at least one processor, a region in a current image which has a change in at least one of a scale and an angle with respect to a reference image;
generating, by the at least one processor, variation information including at least one of difference component in the scale and difference component in the angle between the region in the current image and a reference region in the reference image corresponding to the region in the current image;
encoding based on the generated variation information, by the at least one processor, residual information representing a difference between a predicted sample value from the reference image and a sample value of the current image by performing transformation and quantization on the residual information; and
generating a bitstream including the variation information and the encoded residual information.

2. The video encoding method of claim 1 , wherein the determining the region in the current image comprises:
determining whether a size a current block in the current image is a predetermined size or is larger than the predetermined size; and
determining the region in the current block in response to the determining that the size of the current block is the predetermined size or is larger than the predetermined size.

3. The video encoding method of claim 1 , wherein the determining the region in the current image comprises:
generating a scale space of the current image; determining a region candidate based on the generated scale space; and
determining the region in the current image based on the determined region candidate having a change in at least one of a scale and an angle with respect to the reference image.

4. The video encoding method of claim 3 , wherein the determining the region candidate comprises:
comparing a Hessian determinant value of a predetermined region in the scale space with a Hessian determinant value of a region adjacent to the predetermined region; and
determining the predetermined region as the region candidate when the predetermined region has a local maximum value with respect to a value of a sample in the predetermined region.

5. The video encoding method of claim 4 , wherein a region interpolated in the predetermined region having the local maximum value is determined to be the region candidate.

6. The video encoding method of claim 1 , wherein the generating the variation information between the region and the reference region comprises:
calculating a Haar wavelet response vector of the region and a Haar wavelet response vector of a region adjacent to the region and a Haar wavelet response vector of a Haar wavelet response vector of the reference region and a Haar wavelet response vector of a region adjacent to the reference region;
determining a reference direction of the region and a reference direction of the reference region based on directions of the calculated Haar wavelet response vectors, respectively; and
generating variation information representing an angle variation based on the determined reference direction of the region and the determined reference direction of the reference region.

7. The video encoding method of claim 6 , wherein the determining the region in the current image comprises:
setting a predetermined region comprising the region based on the determined reference direction of the region;
determining a predetermined region comprising a first region based on a determined reference direction of the first region;
calculating a gradient vector of a predetermined region of the region and a gradient vector of a predetermined region of the first region; and
determining the first region as the reference region based on a comparison result between the calculated gradient vector of the region and the calculated gradient vector of the first region.

8. The video encoding method of claim 1 , wherein the generating the variation information comprises generating variation information representing a scale variation between the region in the current image and the reference region based on a coordinate value in a scale space representing scale information of the region in the current image and a coordinate value in a scale space representing scale information of the reference region.

9. A video encoding apparatus comprising an encoder configured to:
determine a region in a current image which has a change in at least one of a scale and an angle with respect to a reference image,
generate variation information including at least one of difference component in the scale and difference component in the angle between the region in the current image and a reference region in the reference image corresponding to the region in the current image,
encode, based on the generated variation information, residual information representing a difference between a predicted sample value from the reference image and a sample value of the current image by performing transformation and quantization on the residual information, and
generate a bitstream including the variation information and the encoded residual information.

10. A video decoding method comprising:
obtaining, by at least one processor, encoded information of a current image, variation information, and residual information about the current image from a bitstream;
determining, by the at least one processor, a reference region in a reference image based on the obtained encoded information;
predicting, by the at least one processor, the current image by using the determined reference region and the obtained variation information; and
restoring, by the at least one processor, the current image from the predicted current image by performing inverse-quantization and inverse-transformation on the obtained residual information,
wherein the variation information comprises at least one of difference component in scale and difference component in angle between a region in the current image and the reference region in the reference image.

11. The video decoding method of claim 10 , wherein the predicting the current image comprises:
determining whether a size of a current block in the current image is of a predetermined size or larger than the predetermined size; and
predicting the current image utilizing the variation information in response to the size of the current block being the predetermined size or being larger than the predetermined size.

12. The video decoding method of claim 10 , wherein a scale space of the current image is generated, a region candidate is determined by analyzing the generated scale space, a region of the determined region candidate having a change in the scale and the angle with respect to the reference image is determined as a region, and the encoded information comprises information about the reference region in the reference image determined corresponding to the region in the current image.

13. The video decoding method of claim 12 , wherein a Hessian determinant value of a predetermined region in the scale space is compared with a Hessian determinant value of a region adjacent to the predetermined region, the predetermined region is determined as the region candidate when the predetermined region has a local maximum value with respect to a value of a sample in the predetermined region, and
the encoded information comprises information about the reference region in the reference image determined corresponding to the determined region candidate.

14. The video decoding method of claim 13 , wherein a region interpolated in the predetermined region having the local maximum value is determined

Assignees

• Samsung Electronics Co Ltd
• Industry-Academic Cooperation Foundation Yonsei Univ

Inventors

• Jeong Seung-Soo
• Ro Won-Woo
• Wey Ho-Cheon
• Kim Kyung-Ah
• Kim Deok-Ho
• Kim Min-Woo
• Kim Chan-Yul

Classifications

• H04N19/537

  Motion estimation other than block-based · CPC title

• H04N19/52

  by predictive encoding · CPC title

• H04N19/567 Primary

  Motion estimation based on rate distortion criteria · CPC title

• H04N19/51

  Motion estimation or motion compensation · CPC title