Method for processing image noise in compression process and circuit system

Patent metadata

Field	Value
Publication number	US-12094090-B2
Application number	US-202117493963-A
Country	US
Kind code	B2
Filing date	Oct 5, 2021
Priority date	Oct 7, 2020
Publication date	Sep 17, 2024
Grant date	Sep 17, 2024

Abstract

A method for processing image noise in a compression process and a circuit system thereof are provided. The circuit system receives an image that is compressed in a specific compression format, and obtains pixel values after the image is decompressed. Differences of pixel values between pixels adjacent to each other can be calculated. A specific area of the image can be divided into multiple sections. A maximum pixel value and a minimum pixel value for each area are used to obtain a regional maximum difference. The regional maximum difference is used to determine whether or not to perform an image-denoising process on the areas such as a flat area or a high-frequency area. Differences among the adjacent pixels can be adjusted according to a weighted average brightness of the adjacent pixels. The image-denoising process is performed on the areas, and an image with reduced noise is finally outputted.

First claim

What is claimed is:

1. A method for processing image noise in a compression process, performed in an image processor of a video camera, comprising:
receiving an image compressed in a compression format;
decompressing the image for obtaining pixel values of pixels of the image;
calculating pixel-value differences between adjacent ones of the pixels, wherein one of a plurality of areas of the image is divided into multiple sections, a maximum pixel value and a minimum pixel value for each of the sections are obtained, and a regional maximum difference of the section is also obtained; wherein a maximum difference of the area is obtained by comparing the regional maximum differences;
comparing the maximum difference of the area with a lower threshold and an upper threshold for dividing the image into at least one of a flat area, an ordinary texture area or a high-frequency area; and
performing an image-denoising process on the flat area and the high-frequency area, but not on the ordinary texture area, such that an output image with reduced noise in the compression process is then rendered.

2. The method according to claim 1 , wherein, when the areas that require performing the image-denoising process are obtained, a difference (D i,j ) between a pixel value of each of the pixels and pixel values of the pixels that are adjacent thereto in the areas are compared with two thresholds (EPF_THD0,EPF_THD1) so as to decide a weight (W i,j ) of each of the pixels, and the weight of each of the pixels is obtained through an equation as follows:

W

i
,
j

=

{

max
⁢

⁢
W

,

⁢

if
⁢

⁢

D

i
,
j

<
EPF_THD0

(

D

i
⁢
j

-
EPF_THD0

)

×
max
⁢
W

EPF_THD1
-
EPF_THD0

⁢

,

⁢

if
⁢

⁢
EPF_THD0

≤

D

i
,
j

≤

EPF_THD1
.

0
,

⁢

if
⁢

⁢

D

i
,
j

>
EPF_THD1

.

3. The method according to claim 1 , wherein, a weighted average brightness is calculated based on a pixel value of each pixel of the image and pixel values of the pixels that are adjacent thereto, and the weighted average brightness of the pixel is used to calculate an average brightness of each area, so as to adjust the maximum difference of the area.

4. The method according to claim 3 , wherein, a brightness adjustment curve is introduced for obtaining a brightness coefficient according to the average brightness of each area, and the brightness coefficient is used to adjust the maximum difference of the area.

5. The method according to claim 4 , wherein, when the areas that require performing the image-denoising process are obtained, a difference (D i,j ) between a pixel value of each of the pixels and pixel values of the pixels that are adjacent thereto in the areas are compared with two thresholds (EPF_THD0,EPF_THD1) so as to decide a weight (W i,j ) of each of the pixels, and the weight of each of the pixels is obtained through an equation as follows:

Assignees

• Realtek Semiconductor Corp

Inventors

• Chen Yan-Fong
• Huang Wen-Tsung

Classifications

• H04N19/86

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

• G06T2207/20021

  Dividing image into blocks, subimages or windows · CPC title

• G06T5/70 Primary

  Denoising; Smoothing · CPC title