Granted Patent B2

US 12,651,317 · App. 18/607,014 · Granted Jun 9, 2026

Inventors: Moonhee Kim (Suwon-si, KR); Bongjoe Kim (Suwon-si, KR); Sanghun Kim (Suwon-si, KR)

Assignee: SAMSUNG ELECTRONICS CO., LTD.

G06T5/70G06F18/15G06T5/50G06T5/60G06T7/0002G06T7/13G06V10/30G06V10/44G06V10/77G06V10/82G06T2207/20012G06T2207/20081G06T2207/20084G06T2207/20172G06T2207/20192G06T2207/20212G06T2207/30168

Abstract

Claims (48)

1 . An image processing method comprising:

obtaining a noise-removed image from an image;

obtaining removed noise from a difference between the image and the noise-removed image;

obtaining a statistical feature of the noise-removed image and a statistical feature of the removed noise;

identifying a noise feature based on the statistical feature of the noise-removed image and the statistical feature of the removed noise, wherein the noise feature indicates whether the removed noise of a pixel of the image is perceptible noise or perceptible detail; and

removing perceptible noise from the image based on the noise feature.

2 . The image processing method of claim 1 , wherein the obtaining the noise-removed image comprises filtering the image based on a predetermined rule.

3 . The image processing method of claim 1 , wherein the obtaining the noise-removed image comprises applying, to the image, an artificial intelligence model configured to remove noise.

4 . The image processing method of claim 1 , wherein the obtaining the statistical feature of the noise-removed image and the statistical feature of the removed noise comprises:

identifying the pixel in the image;

identifying an area in the noise-removed image corresponding to a position of the pixel and an area in the removed noise corresponding to the position of the pixel; and

obtaining a statistical feature of the area in the noise-removed image and a statistical feature of the area in the removed noise.

5 . The image processing method of claim 4 , wherein the identifying the noise feature comprises:

obtaining directionality of the pixel by analyzing the area in the noise-removed image and the area in the removed noise; and

identifying the noise feature based on the directionality of the pixel.

6 . The image processing method of claim 1 , wherein the identifying the noise feature comprises:

based on a difference between the statistical feature of the noise-removed image and the statistical feature of the removed noise being within a predetermined range, identifying the noise feature of the pixel as perceptible detail; and

based on the difference between the statistical feature of the noise-removed image and the statistical feature of the removed noise being outside the predetermined range, identifying the noise feature of the pixel as perceptible noise.

7 . The image processing method of claim 1 , wherein the identifying the noise feature comprises obtaining a feature parameter indicating a degree to which the noise feature of the pixel is perceptible noise or perceptible detail, based on the statistical feature of the noise-removed image and the statistical feature of the removed noise.

8 . The image processing method of claim 7 , wherein the removing perceptible noise comprises combining the image with the noise-removed image based on the feature parameter.

9 . The image processing method of claim 7 , wherein the removing perceptible noise comprises:

obtaining a first modified image by applying, to the image, a first artificial intelligence model configured to process perceptible noise;

obtaining a second modified image by applying, to the image, a second artificial intelligence model configured to process perceptible detail; and

combining the first modified image with the second modified image based on the feature parameter of the pixel.

10 . The image processing method of claim 1 , wherein the identifying the noise feature comprises:

identifying whether the pixel corresponds to an edge, based on an edge map of the image; and

based on the pixel corresponding to the edge, identifying the noise feature of the pixel as perceptible detail.

11 . The image processing method of claim 1 , wherein the identifying the noise feature comprises identifying the noise feature based on at least one of a size of a screen on which the image is reproduced, a brightness of the screen on which the image is reproduced, or a viewing distance of the image.

12 . The image processing method of claim 1 , wherein the removing perceptible noise comprises:

based on identifying the noise feature as perceptible noise, replacing the pixel with a corresponding pixel of the noise-removed image.

13 . The image processing method of claim 1 , wherein the removing perceptible noise comprises:

based on identifying the noise feature of the pixel as perceptible noise, applying, to the pixel, a first artificial intelligence model configured to process perceptible noise; and

based on identifying the noise feature of the pixel as perceptible detail, applying, to the pixel, a second artificial intelligence model configured to process perceptible detail.

14 . An image processing device comprising:

at least one memory storing one or more instructions; and

at least one processor configured to execute the one or more instructions,

wherein the at least one processor is configured to execute the one or more instructions to cause the image processing device to:

obtain a noise-removed image from an image,

obtain removed noise from a difference between the image and the noise-removed image,

obtain a statistical feature of the noise-removed image and a statistical feature of the removed noise,

identify a noise feature based on the statistical feature of the noise-removed image and the statistical feature of the removed noise, wherein the noise feature indicates whether the removed noise of a pixel of the image is perceptible noise or perceptible detail, and

remove perceptible noise from the image based on the noise feature of the pixel.

15 . A non-transitory computer readable medium having instructions stored therein, which when executed by at least one processor cause the at least one processor to execute an image processing method, the image processing method comprising:

obtaining a noise-removed image from an image;

obtaining removed noise from a difference between the image and the noise-removed image;

obtaining a statistical feature of the noise-removed image and a statistical feature of the removed noise;

identifying a noise feature based on the statistical feature of the noise-removed image and the statistical feature of the removed noise, wherein the noise feature indicates whether the removed noise of a pixel of the image is perceptible noise or perceptible detail; and

removing perceptible noise from the image based on the noise feature.

Assignments (1)

Priority Claims (1)

Continuity (2)

References Cited (31)