Image Restoration via Dilated Mask Overlay

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing image restoration methods struggle with poor restoration effects due to the complexity and irregularity of original masks, especially when the patterns of masks in photos are greatly different.

Innovation Solution

The method processes the original mask into a dilated mask with a regular shape, overlays it in the original image to create a mask enhancement feature map, and then uses an image restoration network to restore the dilated mask, improving the restoration effect.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If the original mask is directly used for restoration, then the restoration method is simple, but the restoration effect is poor due to complexity and irregularity of the mask

Engineering Contradiction:
Improverestoration effectVSAvoidmask processing complexity
Core Design Contradiction:
Manufacturing precisionVSDevice complexity

Solution Approach 1:

The patent divides the mask restoration task into separate modules: mask detection module, mask dilation module, and image restoration module. The mask is processed in segments through detection, dilation, and restoration operations, allowing each module to focus on specific aspects and improving overall restoration effectiveness while managing complexity through modular architecture.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent applies preliminary actions by first detecting the mask position and then dilating the mask before restoration. This preliminary processing prepares the mask data in advance, transforming the irregular mask into a dilated version that is easier to restore, thereby improving restoration effect without increasing the core restoration algorithm complexity.

Inventive Principle:
Principle #10Preliminary action

2Adaptability or versatility

If the mask pattern is highly variable, then the method can handle diverse masks, but the restoration quality deteriorates due to lack of adaptability

Engineering Contradiction:
Improvemask pattern adaptabilityVSAvoidrestoration quality
Core Design Contradiction:
Adaptability or versatilityVSManufacturing precision

Solution Approach 1:

The patent changes the parameter of mask dilation to handle variable mask patterns. By adjusting the dilation radius and other parameters in the mask dilation module, the system can adapt to different mask shapes and patterns. This parameter adjustment allows the restoration network to maintain high restoration quality across diverse mask types without requiring complex pattern-specific processing.

Inventive Principle:
Principle #35Parameter changes

3Device complexity

If the mask is dilated to simplify texture, then the restoration difficulty is reduced, but the mask area increases

Engineering Contradiction:
Improverestoration difficultyVSAvoidmask area
Core Design Contradiction:
Device complexityVSArea of stationary object

Solution Approach 1:

The patent performs mask dilation as a preliminary action before restoration. The mask is dilated in advance to simplify its texture and make it easier to restore, then the restored result is used to generate the final output. This preliminary dilation reduces restoration difficulty while the increased mask area is temporary and necessary for achieving good restoration quality.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentUS20250029369A1Image restoration method and device, and non-transitory computer storage medium
Publication Date: 2025.01.23 BOE TECHNOLOGY GROUP CO LTD
  • US20250029369A1 patent drawing
  • US20250029369A1 patent drawing
  • US20250029369A1 patent drawing

AI summary

Disclosed are an image restoration method and device, and a non-transitory computer storage medium. The image restoration method includes: an original mask in an original to-be-restored image is processed into a dilated mask with a regular shape, the dilated mask is overlaid at the location of the original mask in the original to-be-restored image to acquire a mask enhancement feature map, and then the dilated mask in the mask enhancement feature map is restored using an image restoration network to acquire a restored image corresponding to the original to-be-restored image.