Adaptive Image Smoothing via Local Correlation and Basis Vectors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image processing techniques struggle to simultaneously reduce noise and maintain image sharpness, as they use a fixed smoothing strength for all pixels, leading to either noise reduction at the cost of blur in texture parts or unevenness in flat parts.
Innovation Solution
An image processing apparatus that calculates correlations and feature quantities using basis vectors to dynamically adjust weights for pixel values, allowing for adaptive smoothing strength based on local image structures, thereby reducing noise while preserving sharpness.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If a high smoothing strength is set to reduce noise, then noise reduction is improved, but blur is generated in texture parts and sharpness is lost
Solution Approach 1:
The patent applies different smoothing strengths to different regions of the image based on local characteristics. Texture parts are identified and assigned lower smoothing strengths to preserve sharpness, while flat parts are assigned higher smoothing strengths to effectively reduce noise. This local differentiation resolves the contradiction by allowing noise reduction where needed without compromising texture sharpness.
Solution Approach 2:
The patent dynamically adjusts the smoothing strength for each pixel based on local image characteristics such as texture density and gradient information. Rather than using a fixed smoothing strength, the system adapts the smoothing parameter locally, enabling the image processing to respond to varying content requirements and achieve both noise reduction and sharpness preservation.
2Shape
If a low smoothing strength is set to maintain sharpness, then sharpness is maintained, but noise cannot be sufficiently reduced and unevenness is generated in flat parts
Solution Approach 1:
The patent identifies flat regions in the image and applies higher smoothing strengths specifically to these areas, while maintaining lower smoothing strengths in texture regions. This selective application allows sufficient noise reduction in flat parts without generating unevenness, while preserving sharpness in texture parts through locally adapted smoothing parameters.
3Ease of operation
If a fixed smoothing strength is used for one image, then processing simplicity is maintained, but both noise removal and sharpness maintenance cannot be achieved simultaneously
Solution Approach 1:
The patent performs preliminary analysis of image characteristics (texture density, gradient information) before applying smoothing. This pre-processing step identifies regions that require different smoothing treatments, allowing the system to automatically determine appropriate local smoothing strengths without requiring complex manual intervention, thus maintaining ease of operation while achieving superior noise reduction and sharpness preservation.
Data Source
AI summary
According to one embodiment, an image processing apparatus includes following units. The correlation calculation unit calculates correlations between a first region and predetermined first basis vectors. The distance calculation unit calculates distances between the first region and second regions on a subspace generated by the second basis vectors selected from the first basis vectors. The feature quantity calculation unit calculates a feature quantity based on the correlations. The weight calculation unit calculates weights based on the distances and the feature quantity. The pixel value calculation unit calculates a weighted average of pixel values according to the weights to generate an output pixel value.