Spatial Mask Deconvolution for Ringing Artifact Suppression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current deconvolution methods struggle with accurately recovering latent sharp images from blurry images due to ill-conditioned problems, noise, and artifacts, especially in complex scenes with varying depth and bright areas, leading to ringing artifacts and inefficient processing.
Innovation Solution
Incorporating a spatial mask in the fidelity term of both latent sharp image and PSF estimation cost functions, along with variable splitting techniques, allows for region-specific deblurring, masking out difficult areas, and suppressing artifacts, and utilizing fast Fourier transforms for efficient minimization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If deconvolution methods are applied to recover latent sharp images, then image sharpness is improved, but ringing artifacts appear around edges and boundaries
Solution Approach 1:
The patent introduces a spatial mask that selectively applies deconvolution to specific regions of the image. The mask allows different processing behavior for different spatial locations, enabling sharpness enhancement in regions where it is needed while suppressing ringing artifacts in regions where deconvolution would be harmful, such as around strong edges and boundaries.
2Adaptability or versatility
If blind deconvolution is used to recover both PSF and latent sharp image, then versatility is improved, but computational complexity increases significantly
Solution Approach 1:
The patent segments the deconvolution process into two separate optimization problems: one for estimating the latent sharp image and another for estimating the PSF. This segmentation allows each problem to be solved independently with appropriate cost functions, reducing the overall computational complexity compared to simultaneous blind deconvolution while maintaining the ability to handle both image recovery and PSF estimation.
3Productivity
If standard deconvolution cost functions are used, then processing speed is improved, but image quality deteriorates due to saturation and bright areas
Solution Approach 1:
The patent modifies the cost function parameters by introducing a spatial mask that changes the weighting of different image regions. This allows the optimization to focus on regions with reliable information while downweighting or excluding saturated and extremely bright areas that would otherwise degrade image quality, maintaining processing speed while improving output quality.
4Loss of time
If deconvolution is applied to entire image, then processing time is reduced, but processing efficiency decreases due to unnecessary computation in certain regions
Solution Approach 1:
The spatial mask enables selective deconvolution by allowing the algorithm to skip computation in regions where the mask value indicates that deconvolution would not be beneficial or would introduce artifacts. This local adaptation reduces unnecessary computational effort while maintaining processing time efficiency, improving overall processing efficiency.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach improves the quality of deblurred images by reducing ringing artifacts, allowing for faster deblurring on less powerful processors and enabling selective deblurring of specific image regions while maintaining image integrity.
Implementation Method 1
utilizing fast Fourier transforms for efficient minimization
Data Source
AI summary
A method for deblurring a blurry image (18) includes utilizing a spatial mask and a variable splitting technique in the latent sharp image estimation cost function. Additionally or alternatively, the method can include the utilizing a spatial mask and a variable splitting technique in the PSF estimation cost function. The spatial mask can be in a fidelity term in either or both the latent sharp image estimation cost function and the PSF cost function. The latent sharp image estimation cost function can be used for non-blind deconvolution. Alternatively, one or both cost functions can be used for blind deconvolution.