Method for image processing and apparatus for implementing the same

Patent metadata

Field	Value
Publication number	US-11700397-B2
Application number	US-202117390509-A
Country	US
Kind code	B2
Filing date	Jul 30, 2021
Priority date	Jul 30, 2020
Publication date	Jul 11, 2023
Grant date	Jul 11, 2023

Abstract

A method of processing an image is proposed, which includes: determining, based on the image, one or more noise templates, wherein each of the one or more noise templates includes noise pixels representing noise contained in the image; calculating one or more first autocovariance values, based on the noise pixels of at least one of the one or more noise templates; based on the one or more first autocovariance values, selecting an entry of a noise model database among database entries which respectively include values of noise model parameters corresponding to a noise model.

First claim

The invention claimed is:

1. A method of processing an image, the method comprising:
determining, based on the image, one or more noise templates, each of the one or more noise templates including noise pixels representing noise contained in the image;
calculating one or more first autocovariance values, based on the noise pixels of at least one of the one or more noise templates;
based on the one or more first autocovariance values, selecting an entry of a noise model database among database entries which respectively comprise values of noise model parameters corresponding to a noise model;
using the values of the noise model parameters to characterize the noise contained in the image to be used for further processing of the image; and
obtaining, for each entry of the database to be searched, one or more second autocovariance values, based on noise pixels representing one or more database noise templates corresponding to the values of noise model parameters comprised in the entry,
wherein the one or more second autocovariance values are pre-calculated, and comprised in data stored in the entry.

2. The method according to claim 1 , wherein the selecting the entry of the noise model database is based on a comparison of the one or more first autocovariance values with the one or more second autocovariance values.

3. The method according to claim 2 , further comprising:
comparing the one or more first autocovariance values with the one or more second autocovariance values based on a distance between the one or more first autocovariance values and the one or more second autocovariance values.

4. The method according to claim 1 , performed at an encoder configured for encoding the image, the method further comprising: generating a set of denoised image pixels, based on the noise model parameters comprised in the entry selected in the noise model database.

5. The method according to claim 1 , wherein the one or more noise templates are determined for pixels of the image corresponding to respective values of planes (Y, Cb, Cr), each representing luminance or chrominance of pixels.

6. The method according to claim 1 , further comprising:
normalizing one or more of the noise templates based on which first autocovariance values are calculated, prior to calculating the first autocovariance values, by dividing their noise pixels by a standard deviation calculated on their basis.

7. The method according to claim 1 , wherein each of the database entries comprises values of noise model parameters associated with one or more of (Y, Cb, Cr) planes representing luminance or chrominance of pixels.

8. The method according to claim 1 , wherein the noise model is an auto-regressive (“AR”) parametric noise model, and the noise model parameters comprise AR parametric noise model parameters, wherein each of AR parametric noise model noise pixels representing a database noise template among the one or more database noise templates corresponding to the values of the AR parametric noise model parameters is computable based on a combination of a linear combination of previous noise pixels in a causal neighborhood of the AR parametric noise model noise pixel weighted by respective AR model linear combination parameters (φ 1 , . . . , φ P ) with a generated noise sample corresponding to an additive Gaussian noise of AR model variance parameter (σ) and using a seed parameter of a pseudo-random number generator.

9. The method according to claim 1 ,
wherein respective second autocovariance values for the database noise template are obtained for one or more of (Y, Cb, Cr) planes representing luminance or chrominance of pixels, and
wherein the entry of the noise model database is selected based on similarity estimates respectively representing, for each of the one or more of (Y, Cb, Cr) planes representing luminance or chrominance of pixels, a similarity between the one or more first autocovariance values and the one or more second autocovariance values.

10. An apparatus, comprising:
a processor, and a memory operatively coupled to the processor,
the apparatus configured to perform an image processing method that comprises steps of:
determining, based on the image, one or more noise templates, each of the one or more noise templates including noise pixels representing noise contained in the image;
calculating one or more first autocovariance values, based on the noise pixels of at least one of the one or more noise templates;
based on the one or more first autocovariance values, selecting an entry of a noise model database among database entries which respectively comprise values of noise model parameters corresponding to a noise model;
using the values of the noise model parameters to characterize the noise contained in the image to be used for further processing of the image; and
obtaining, for each entry of the database to be searched, one or more second autocovariance values, based on noise pixels representing one or more database noise templates corresponding to the values of noise model parameters comprised in the entry,

wherein the one or more second autocovariance values are pre-calculated, and comprised in data stored in the entry.

11. A video encoder, configured to encode video content comprising a plurality of images, the video encoder comprising the apparatus according to claim 10 , configured to process images of an input video.

12. The apparatus of claim 10 , wherein the selecting the entry of the noise model database is based on a comparison of the one or more first autocovariance values with the one or more second autocovariance values.

13. A non-transitory computer-readable medium encoded with executable instructions which, when executed by an apparatus comprising a processor operatively coupled with a memory, causes the processor to perform an image processing method that comprises steps of:
determining, based on the image, one or more noise templates, each of the one or more noise templates including noise pixels representing noise contained in the image;
calculating one or more first autocovariance values, based on the noise pixels of at least one of the one or more noise templates;
based on the one or more first autocovariance values, selecting an entry of a noise model database among database entries which respectively comprise values of noise model parameters corresponding to a noise model;
using the values of the noise model parameters to characterize the noise contained in the image to be used for further processing of the image; and
obtaining, for each entry of the database to be searched, one or more second autocovariance values, based on noise pixels representing one or more database noise templates corresponding to the values of noise model parameters comprised in the entry,
wherein the one or more second autocovariance values are pre-calculated, and comprised in data stored in the entry.

14. The non-transitory computer-readable medium of claim 13 , wherein the selecting the entry of the noise model database is based on a comparison of the one or more first autocovariance values with the one or more second autocovariance values.

Assignees

• Ateme

Inventors

• Guionnet Thomas
• Herbreteau Sébastien

Classifications

• G06T2207/10016

  Video; Image sequence · CPC title

• G06T5/002

  Physics · mapped topic

• H04N19/85 Primary

  using pre-processing or post-processing specially adapted for video compression · 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

• H04N19/86

  involving reduction of coding artifacts, e.g. of blockiness · CPC title