IP Library › Granted Patent US 12664406

Granted Patent B2

US 12664406 · App. 18/220,008 · Granted Jun 23, 2026

Method of correcting image by using neural network model, and computing device for executing neural network model for image correction

Inventors: Pilkyu Park (Suwon-si, KR); Jaehwan Kim (Suwon-si, KR); Heechul Yang (Suwon-si, KR); Jongseok Lee (Suwon-si, KR); Chaeeun Lee (Suwon-si, KR); Youngo Park (Suwon-si, KR); Kwangpyo Choi (Suwon-si, KR)

Assignee: SAMSUNG ELECTRONICS CO., LTD.

G06N3/045 G06T5/60 G06T2207/20081 G06T2207/20084

View Patent ↗

Loading inventors, assignments & file history…

Monitor This Case

Get email alerts when status or documents change.

Order Certified Copies

Most orders are placed with the USPTO same day — all within 24 business hours.

Order via The Patent Place →

Pre-filled with this patent's details

Quick Facts

Patent No.: US 12664406
App. No.: 18/220,008
Granted: Jun 23, 2026
Kind: B2

External Links

USPTO Patent Center↗ Full Patent PDF↗

Abstract

A method of correcting an image includes: inputting an input image and a correction parameter to a first neural network model; and obtaining an inference image from the first neural network model, the inference image corresponding to the correction parameter. The first neural network model is a model trained to minimize a difference between the inference image that is output after the input image and the correction parameter are input and a label image corresponding to the correction parameter. The label image is an image obtained by, when the correction parameter is applied to at least one image correction algorithm, correcting the input image by using the at least one image correction algorithm.

Claims (107)

1 . A method of correcting an image, the method comprising:

controlling, by a processor of a computing device, the computing device to display an input image and a user interface;

identifying, by the processor, a correction parameter based on a first input received through the user interface, wherein the first input indicates a correction corresponding to the input image;

inputting, by the processor, the input image and the correction parameter to a first neural network model; and

obtaining, by the processor, an inference image from the first neural network model, the inference image corresponding to the correction parameter,

wherein the first neural network model is a model trained to minimize a difference between the inference image that is output after the input image and the correction parameter are input and a label image corresponding to the correction parameter,

wherein the label image corresponding to the correction parameter is an image obtained by, when the correction parameter is applied to at least one image correction algorithm, correcting the input image by using the at least one image correction algorithm, and

wherein the first neural network model is trained on a relationship between the correction parameter and the label image so that the inference image output from the first neural network model is controllable by changing the correction parameter.

2 . The method of claim 1 ,

wherein a first inference image output from the first neural network model when the input image and a first correction parameter are input to the first neural network model is different from a second inference image output from the first neural network model when the input image and a second correction parameter are input to the first neural network model.

3 . The method of claim 1 ,

wherein the correction parameter is a first correction parameter,

wherein the inputting the correction parameter comprises:

inputting, by the processor, the input image to a second neural network model;

obtaining, by the processor, the first correction parameter that is inferred by the second neural network model; and

inputting, by the processor, the first correction parameter together with the input image to the first neural network model, and

wherein the second neural network model is a model trained to minimize a difference between the first correction parameter that is inferred by the second neural network model after the input image is input to the second neural network model and a second correction parameter that allows the label image to have preset image characteristics.

4 . The method of claim 3 , wherein the second neural network model is trained by:

inputting the input image to the second neural network model;

obtaining the correction parameter output from the second neural network model;

generating the label image by correcting the input image by applying the correction parameter to the at least one image correction algorithm;

obtaining a measured characteristic value that quantifies image characteristics of the label image; and

updating the second neural network model to minimize a difference between the measured characteristic value and a preset target characteristic value.

5 . The method of claim 1 , wherein the inputting the correction parameter comprises:

receiving, by the processor, an input for adjusting the correction parameter through an input/output interface of the computing device;

changing, by the processor, a preset first correction parameter into a second correction parameter, based on the input; and

inputting, by the processor, the second correction parameter together with the input image to the first neural network model.

6 . The method of claim 5 ,

wherein the obtaining the inference image comprises obtaining, by the processor, an inference image corresponding to the second correction parameter from the first neural network model,

wherein the first neural network model is a model trained to minimize a difference between the first inference image output after the input image and the first correction parameter are input and the first label image corresponding to the first correction parameter, and minimize a difference between the second inference image output after the input image and the second correction parameter are input and the second label image corresponding to the second correction parameter,

wherein the first label image is an image obtained by, when the first correction parameter is applied to the at least one image correction algorithm, correcting the input image by using the at least one image correction algorithm, and

wherein the second label image is an image obtained by, when the second correction parameter is applied to the at least one image correction algorithm, correcting the input image by using the at least one image correction algorithm.

7 . The method of claim 1 , wherein the first neural network model is trained by:

generating the label image by correcting the input image by applying the correction parameter to the at least one image correction algorithm;

inputting the correction parameter and the input image to the first neural network model;

obtaining the inference image output from the first neural network model; and

updating the first neural network model to minimize a difference between the label image and the inference image.

8 . The method of claim 1 , wherein the inputting the input image and the correction parameter to the first neural network model comprises:

converting, by the processor, the input image into a first gray scale image;

generating, by the processor, a second gray scale image by applying a first parameter included in the correction parameter to the first gray scale image;

generating, by the processor, a third gray scale image by applying a second parameter included in the correction parameter to the first gray scale image; and

inputting, by the processor, the second gray scale image and the third gray scale image to the first neural network model.

9 . The method of claim 1 , wherein the obtaining the inference image comprises:

obtaining, by the processor, at least one of a filter or map information from the first neural network model; and

converting, by the processor, the input image into the inference image by using the at least one of the filter or the map information.

10 . The method of claim 1 , wherein the inputting the input image and the correction parameter to the first neural network model comprises:

receiving, by the processor, from a camera included in the computing device, raw images obtained through an image sensor of the camera;

generating, by the processor, the input image by using the raw images; and

inputting, by the processor, the input image and the correction parameter to the first neural network model.

11 . The method of claim 1 , wherein the inputting of the input image and the correction parameter to the first neural network model comprises:

receiving, by the computing device, raw images from an external device;

generating, by the processor, the input image by using the raw images; and

inputting, by the processor, the input image and the correction parameter to the first neural network model.

12 . A non-transient computer-readable recording medium having recorded thereon a program for performing a method comprising:

controlling, by a processor of a computing device, the computing device to display an input image and a user interface;

identifying, by the processor, a correction parameter based on a first input received through the user interface, wherein the first input indicates a correction corresponding to the input image;

inputting, by the processor, the input image and the correction parameter to a first neural network model; and

obtaining, by the processor, an inference image from the first neural network model, the inference image corresponding to the correction parameter,

wherein the label image is an image obtained by, when the correction parameter is applied to at least one image correction algorithm, correcting the input image by using the at least one image correction algorithm, and

13 . A computing device comprising:

an input/output interface through which an input is received from a user;

a memory in which a program for performing a method of correcting an input image is stored; and

a processor configured to execute the program,

wherein the method includes:

controlling the computing device to display the input image and a user interface;

identifying a correction parameter based on a first input received through the user interface, wherein the first input indicates a correction corresponding to the input image;

inputting the input image and the correction parameter to a first neural network model, and

obtaining an inference image from the first neural network model, the inference image corresponding to the correction parameter,

14 . The computing device of claim 13 ,

15 . The computing device of claim 13 ,

wherein the correction parameter is a first correction parameter,

wherein, the inputting the correction parameter includes:

inputting, by the processor, the input image to a second neural network model,

obtaining, by the processor, the first correction parameter that is inferred by the second neural network model, and

inputting the first correction parameter together with the input image to the first neural network model, and

wherein the second neural network model is a model trained to minimize a difference between the first correction parameter that is inferred by the second neural network model after the input image is input and a second correction parameter that allows the label image to have preset image characteristics.

16 . The computing device of claim 15 , wherein the second neural network model is trained by:

inputting the input image to the second neural network model,

obtaining the correction parameter output from the second neural network model,

generating the label image by correcting the input image by applying the first correction parameter to the at least one image correction algorithm,

obtaining a measured characteristic value that quantifies image characteristics of the label image, and

updating the second neural network model to minimize a difference between the measured characteristic value and a preset target characteristic value.

17 . The computing device of claim 13 , wherein the inputting the correction parameter includes:

receiving, by the processor, an input for adjusting the correction parameter through the input/output interface,

changing, by the processor, a preset first correction parameter into a second correction parameter based on the input, and

inputting, by the processor, the second correction parameter together with the input image to the first neural network model.

18 . The computing device of claim 17 ,

wherein the obtaining the inference image includes obtaining by the processor, an inference image corresponding to the second correction parameter from the first neural network model,

19 . The computing device of claim 13 , wherein the first neural network model is trained by:

generating the label image by correcting the input image by applying the correction parameter to the at least one image correction algorithm,

inputting the correction parameter and the input image to the first neural network model,

obtaining the inference image output from the first neural network model, and

updating the first neural network model to minimize a difference between the label image and the inference image.

20 . The computing device of claim 13 , wherein the inputting the input image and the correction parameter to the first neural network model includes:

converting, by the processor, the input image into a first gray scale image,

generating, by the processor, a second gray scale image by applying a first parameter included in the correction parameter to the first gray scale image,

generating, by the processor, a third gray scale image by applying a second parameter included in the correction parameter to the first gray scale image, and

inputting, by the processor, the input image, the second gray scale image and the third gray scale image to the first neural network model.