Blur Removal Without Ringing Artifacts Using Directional Weights
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for removing blur from images often introduce ringing-artifacts, which reduce image quality, due to the amplification of high-frequency components and the difficulty in determining an optimal regularization parameter that balances blur removal and artifact reduction.
Innovation Solution
An image processing method that detects blur, segments images based on blur amount, applies high-pass filtering in horizontal and vertical directions, calculates directional weights, and uses these weights in iterative or closed-form image restoration to reduce ringing-artifacts, thereby improving image quality without introducing artifacts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If image restoration using inverse matrix of blur is applied, then blur removal effectiveness is improved, but ringing-artifact increases due to excessive amplification of high frequency components
Solution Approach 1:
The patent applies different regularization parameters to different regions of the image based on local blur characteristics. By segmenting the image into multiple regions and determining region-specific regularization parameters, the method achieves effective blur removal in each local area while controlling ringing-artifact generation, rather than applying a uniform regularization parameter across the entire image.
Solution Approach 2:
The patent dynamically adjusts the regularization parameter based on local image characteristics and blur severity. By changing the regularization parameter value according to local conditions, the method optimizes the balance between blur removal effectiveness and ringing-artifact suppression for different regions of the image.
2Object-generated harmful factors
If regularization parameter is increased to reduce ringing-artifact, then ringing-artifact is suppressed, but blur removal effectiveness decreases
Solution Approach 1:
The patent determines different regularization parameters for different image regions based on local blur characteristics. Regions with severe blur use smaller regularization parameters to maximize blur removal, while regions prone to ringing-artifact use larger regularization parameters to suppress artifacts, achieving optimal performance in each local area.
Solution Approach 2:
The regularization parameter is not fixed but dynamically determined based on local image characteristics and blur severity. This dynamic adjustment allows the system to adapt the regularization strength to local conditions, optimizing both blur removal and artifact suppression for each region.
3Device complexity
If conventional image restoration is applied uniformly to entire image, then processing simplicity is maintained, but statistical features of partial images are not reflected properly
Solution Approach 1:
The patent divides the image into multiple regions based on blur characteristics and applies region-specific restoration parameters. This segmentation allows the method to capture statistical features of different partial images while maintaining a relatively simple processing framework through automated region classification and parameter determination.
Data Source
AI summary
Provided is an image processing method and apparatus for removing blur in a screen image, including detecting blur in an input image, segmenting the image according to an amount of the detected blur, performing a high pass filtering of the segmented image in a horizontal direction and a vertical direction, detecting corresponding weights by using corresponding coefficients of each direction obtained from filtering; and restoring the image by applying the detected weights to one of: an iterative form or a closed form of an image restoration.