Image Sharpening Weight Maps to Reduce Ringing Artifacts
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing methods using filters for blur correction suffer from adverse effects such as ringing and undershoot in luminance saturation portions of images, leading to a decrease in the effectiveness of blur correction.
Innovation Solution
A method involving sharpening an image using a machine learning model, generating two weight maps based on pixel values, and combining them to generate a corrected image, which reduces adverse effects while maintaining blur correction efficacy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If blur correction processing is performed using a filter (e.g., Wiener filter or inverse filter), then the deterioration (blur) of a captured image due to optical system aberration is corrected, but adverse effects such as ringing and undershoot occur in luminance saturation portions and their peripheral portions
Solution Approach 1:
The patent applies local quality by generating a weight map that assigns different weights to different regions of the image. Specifically, pixels in luminance saturation portions and their peripheral portions are assigned lower weights, while other pixels are assigned higher weights. This allows the blur correction to be applied more strongly in regions where it is beneficial and less strongly in regions where it causes adverse effects, thereby resolving the contradiction between correction quality and harmful artifacts.
Solution Approach 2:
The patent uses partial action by selectively applying blur correction to only those regions of the image where it is appropriate. Through the weight map mechanism, the correction is partially applied (with full strength) to non-saturation regions and partially suppressed (with reduced strength) to saturation regions and their peripheries. This selective partial application eliminates ringing and undershoot while preserving sharpening effects in appropriate areas.
2Object-affected harmful factors
If the correction amount of blur correction is reduced to reduce adverse effects in luminance saturation portions, then ringing and undershoot are reduced, but the effects of blur correction decrease overall
Solution Approach 1:
The patent resolves this contradiction by making the correction amount spatially variable through the weight map. Instead of uniformly reducing the correction amount across the entire image, the weight map assigns high weights to pixels outside luminance saturation portions (maintaining strong correction effects) and low weights to pixels in and near luminance saturation portions (reducing adverse effects). This local differentiation allows the system to maintain strong blur correction where needed while minimizing artifacts where they occur.
Solution Approach 2:
The weight map serves as an intermediary mechanism that mediates between the need for strong blur correction and the need to avoid adverse effects. By introducing this intermediate weighting layer, the patent enables differential control of the correction amount across different image regions, allowing simultaneous achievement of strong correction in appropriate areas and artifact reduction in problematic areas.
Data Source
AI summary
A method includes sharpening a first image to generate a second image, generating a first weight map based on the first image, generating a second weight map based on the second image, and generating a third image based on the first image, the second image, the first weight map, and the second weight map, wherein each of the first weight map and the second weight map indicate a weight of one of the first image and the second image in the generation of the third image.