Edge Smoothing Block Filter for Noise Reduction and Texture Preservation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current digital image processing algorithms face challenges in effectively filtering noise while preserving edges and textures, particularly in handling mixed digital and temporal noise, as existing filters like FIR and bilateral filters either smooth out image contrast or lose low-level textures due to unpredictable artifacts.
Innovation Solution
The Edge Smoothing Block (ESB) filter, combined with Motion Compensated Temporal Filtering (MCTF), uses a weight function that calculates similarity measures based on pixel neighborhoods and adaptive thresholds to selectively filter noise, preserving edges and textures by weighting pixel values according to their similarity and position, thereby addressing the limitations of existing filters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If traditional filters (FIR or bilateral) are used to reduce noise, then noise reduction is achieved, but image contrast is smoothed out or low-level textures are lost
Solution Approach 1:
The filter applies different processing to different regions by evaluating local neighborhood similarity. Pixels in homogeneous regions receive stronger filtering while pixels near edges or textures preserve their original values, achieving spatially adaptive noise reduction that preserves important image features
Solution Approach 2:
The filter dynamically adjusts its behavior based on local image characteristics. The similarity metric and adaptive threshold allow the filter to automatically adapt to different regions (smooth areas vs. edge areas), changing the filtering strength according to local content rather than applying a fixed operation
2Object-affected harmful factors
If stronger filtering is applied to reduce noise, then noise reduction performance improves, but edges and fine textures are blurred or lost
Solution Approach 1:
The filter uses feedback from the local neighborhood comparison to control filtering strength. The similarity metric provides feedback about whether a pixel is in an edge region or homogeneous region, and this feedback dynamically adjusts the filtering applied to that pixel, preventing edge blurring while maintaining noise reduction in appropriate areas
Solution Approach 2:
The filter changes the effective filtering parameter (filtering strength) based on local image content. By using an adaptive threshold and similarity metric, the filter automatically adjusts its behavior from strong filtering in homogeneous regions to weak or no filtering near edges, achieving content-adaptive noise reduction
Data Source
AI summary
Aspects of an edge smoothing block filter and combinations of the filter with a motion compensated temporal filter are described. In one embodiment, edge smoothing block filtering includes selecting a current pixel to be filtered and selecting a candidate pixel within a search area about the current pixel. The edge smoothing block filtering generally seeks to identify candidate pixels having surrounding pixels similar to pixels surrounding the current pixel. The edge smoothing block filtering further computes a cost difference between pixels within a candidate pixel neighborhood and pixels within a current pixel neighborhood, and filters the current pixel based on the cost difference. Aspects of the filters and filter element combinations may preserve edges and textures adaptively based on image content. For example, diagonal or curved image edges may be filtered along edges while texture is preserved along the edges.