Image Deconvolution Using Bilateral Filtering to Suppress Ringing Artifacts
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Image deconvolution processes often introduce ringing artifacts and amplified noise due to the ill-posedness of the deconvolution process, especially when dealing with bandlimited blur kernels, making it difficult to produce clear and artifact-free images.
Innovation Solution
The method involves an iterative non-blind deconvolution using a Richardson-Lucy deconvolution process with a bilateral range/spatial filter, which includes edge preservation regularization and the use of a guide image in a progressive deconvolution approach to form an image pyramid, reducing ringing artifacts and noise amplification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If deconvolution is performed to remove blur from an image, then image clarity is improved, but ringing artifacts and noise amplification occur
Solution Approach 1:
The patent applies different processing strategies to different regions of the image. A guide image is used to identify edge regions versus non-edge regions. Inverse filtering is applied selectively - more aggressively in non-edge regions where it effectively removes blur, while being constrained in edge regions to prevent ringing artifacts. This local differentiation resolves the contradiction by allowing high clarity improvement in safe regions while preventing harmful artifacts in sensitive edge regions.
Solution Approach 2:
The patent introduces a guide image as an intermediary element that mediates between the blurred image and the deconvolution process. The guide image, obtained through edge detection and smoothing, serves as a reference to control where inverse filtering should be applied and with what strength. This intermediary structure enables the system to achieve clear image restoration while suppressing ringing artifacts and noise amplification by using the guide image to modulate the filtering operation.
2Manufacturing precision
If inverse filtering is applied to deconvolute an image, then blur removal is achieved, but noise is excessively amplified
Solution Approach 1:
The patent modifies the inverse filtering operation by applying spatially varying constraints based on the guide image. Regions identified as edges in the guide image receive suppressed filtering to prevent noise amplification, while non-edge regions receive full inverse filtering for effective blur removal. This local quality approach resolves the contradiction by adapting the filtering strength to the local image characteristics.
Solution Approach 2:
The patent performs preliminary processing to create a guide image that identifies potential problem areas before applying inverse filtering. By pre-detecting edges and creating a smoothed guide image, the system prepares a map of where noise amplification is likely to occur and takes preventive action by constraining the inverse filtering in those regions. This preliminary anti-action prevents noise amplification before it can occur during the main deconvolution process.
Data Source
AI summary
Embodiments related to the removal of blur from an image are disclosed. One disclosed embodiment provides a method of performing an iterative non-blind deconvolution of a blurred image to form an updated image. The method comprises downsampling the blurred image to form a blurred image pyramid comprising images of two or more different resolution scales, downsampling a blur kernel to form a blur kernel pyramid comprising kernels of two or more different sizes, and deconvoluting a selected image in the blurred image pyramid according to a Richardson-Lucy deconvolution process in which a bilateral range/spatial filter is employed.