Cross-Kernel Median Filter for Image Noise Removal
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional median filters struggle to effectively remove noise from image edges and corners, leading to deteriorated image quality and reduced compression efficiency, especially in low illuminance conditions where noise is increased relative to the original image.
Innovation Solution
A cross-kernel type median filter is employed, which determines whether a target pixel is noise and corrects the image by applying a cross-kernel median filter based on peripheral pixel information, selecting medians from pixels arranged in '+' or 'X' shapes to efficiently remove noise without distorting the image outline.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If a conventional median filter is used to remove noise, then noise removal capability is improved, but image quality in edge and corner regions deteriorates
Solution Approach 1:
The patent divides the filtering process into two distinct stages: first applying a conventional median filter to remove noise, then applying a cross-kernel median filter specifically to edge and corner regions to restore image quality. This segmentation allows each filter to optimize for its specific function without compromising the other.
Solution Approach 2:
The patent applies different filtering strategies to different regions of the image. The cross-kernel median filter is specifically designed to process edge and corner pixels differently from regular interior pixels, using directional kernels that adapt to local image structures to preserve edges while removing noise.
2Object-affected harmful factors
If a conventional median filter is used to remove noise, then noise removal capability is improved, but compression efficiency deteriorates
Solution Approach 1:
By segmenting the filtering process into conventional median filtering followed by cross-kernel filtering, the patent achieves effective noise removal that improves compression efficiency. The cross-kernel filter's ability to preserve edges and reduce artifacts creates better-compressible image data.
3Reliability
If noise removal processing is applied to improve image quality, then signal to noise ratio improves, but image details are lost
Solution Approach 1:
The cross-kernel median filter uses directionally-adapted kernels that preserve local image structures such as edges and corners while removing noise. By adjusting kernel orientation and size based on local image characteristics, it maintains image details that conventional filters would blur or remove.
Solution Approach 2:
The filter dynamically adapts its kernel parameters based on local image content, detecting edge orientations and adjusting filtering strength accordingly. This dynamic adaptation allows the filter to be aggressive in noise removal where appropriate while being conservative in preserving important image features.
Data Source
AI summary
A device for removing a noise on an image using a cross-kernel type median filter includes a target pixel determination unit configured to determine whether a target pixel is a noise or not; a peripheral pixel determination unit configured to determine degrees of the noise on a peripheral pixel information of a peripheral pixel of the target pixel; and a noise removing unit configured to remove the noise of the target pixel and corrects the image by applying a cross-kernel type median filter based on the peripheral pixel information.