Image Artifact Removal Using GANs for OCR-Ready Images

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing image editing methods require multiple instances of user intervention and are ineffective in removing artifacts like glare, degrading image quality and hindering tasks such as optical character recognition.

Innovation Solution

A method and electronic device using generative adversarial networks (GANs) to classify and manage artifacts like shadows and glare with minimal user intervention, intelligently determining whether to retain or remove them based on their impact on image details.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If multiple instances of user intervention are used to manage artifacts, then the user can control artifact removal, but the process becomes complex and time-consuming

Engineering Contradiction:
Improveuser control over artifact removalVSAvoidmultiple intervention steps
Core Design Contradiction:
Ease of operationVSDevice complexity

Solution Approach 1:

The system performs automatic artifact detection and classification without requiring multiple user interventions. The GAN-based model autonomously identifies artifacts, classifies them as wanted or unwanted, and applies appropriate processing, enabling the system to serve itself rather than requiring continuous user input.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The system changes the operational parameters from manual slider adjustments to automated classification categories. By transforming the control mechanism from continuous user adjustment to discrete automated classification (wanted/unwanted), the system reduces operational complexity while maintaining user control through high-level intent selection.

Inventive Principle:
Principle #35Parameter changes

2Ease of operation

If traditional methods like shadow sliders are used to remove artifacts, then the operation is simple, but the artifact removal is ineffective and degrades image quality

Engineering Contradiction:
Improvesimplicity of artifact removalVSAvoidartifact removal effectiveness
Core Design Contradiction:
Ease of operationVSManufacturing precision

Solution Approach 1:

The patent replaces mechanical slider-based contrast adjustment with an intelligent GAN-based classification system. Instead of using manual mechanical controls that merely adjust overall image contrast, the system employs machine learning models to intelligently identify and selectively process artifacts, achieving effective artifact removal while preserving image quality.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system introduces an intermediary classification layer between the user's simple artifact selection and the actual image processing. The GAN-based classifier acts as an intelligent mediator that determines whether artifacts should be removed or retained, enabling effective artifact management without requiring complex user intervention or sacrificing image quality.

Inventive Principle:
Principle #24Intermediary (Mediator)

3Manufacturing precision

If all artifacts are removed from images, then image quality improves, but wanted artifacts that enhance image details are lost

Engineering Contradiction:
Improveimage qualityVSAvoidwanted artifact details
Core Design Contradiction:
Manufacturing precisionVSLoss of information

Solution Approach 1:

The system applies different processing qualities to different regions of the image based on artifact classification. Wanted artifacts are preserved with their original quality, while unwanted artifacts are selectively removed. This local differentiation ensures that image quality is improved only where necessary without losing important detail information.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The system uses color and texture analysis as part of the artifact classification process. By examining visual properties such as color composition and texture patterns, the GAN model can distinguish between wanted and unwanted artifacts, preserving those that enhance image details while removing those that degrade quality.

Inventive Principle:
Principle #32Color changes

4Ease of operation

If manual artifact management methods are used, then the process is controllable, but productivity is reduced due to multiple user interventions

Engineering Contradiction:
Improvecontrollability of artifact managementVSAvoidimage processing speed
Core Design Contradiction:
Ease of operationVSProductivity

Solution Approach 1:

The system performs preliminary automatic artifact detection and classification before the user makes any decisions. By pre-processing the image to identify and categorize artifacts, the system reduces the number of user interventions needed, thereby improving productivity while maintaining controllability through the user's final selection of artifact management preferences.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP4189638B1Method and electronic device for managing artifacts of image
Publication Date: 2025.12.31 SAMSUNG ELECTRONICS CO LTD
  • EP4189638B1 patent drawingFigure 1
  • EP4189638B1 patent drawingFigure 2
  • EP4189638B1 patent drawingFigure 3

AI summary

A method and an electronic device for managing artifacts of an image includes: receiving an input image and extracting multiple features from the input image. The multiple features include a texture of the input image, a color composition of the input image and edges in the input image. Further, the method includes determining a region of interest (RoI) in the input image including an artifact based on the features and generating an intermediate output image by removing the artifact using multiple generative adversarial networks (GANs). Further, the method includes generating a binary mask using the intermediate output image, the input image, an image illustrating edges in the input image and an image illustrating edges in the intermediate output image and obtaining a final output image by applying the generated binary mask to the input image.