Patch Shift Texture Filtering for Edge Preservation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image texture removal methods struggle to effectively separate textures from structural edges due to computational complexity and limited explicit texture processing, leading to smudging of edges and difficulties in algorithm acceleration, expansion, and adaptation.
Innovation Solution
A patch-based texture filtering method that determines structural edges using modified relative total variation (mRTV) and performs patch shifting to select representative patches, generating guide images through joint bilateral filtering to preserve edges and remove textures.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If patch-based texture removal method is used to explicitly process textures, then texture removal effectiveness is improved, but computational complexity increases making acceleration and adaptation difficult
Solution Approach 1:
The image is divided into multiple patches, and processing is performed independently on each patch. This segmentation allows the complex texture removal problem to be broken down into smaller, more manageable sub-problems that can be processed in parallel, reducing overall computational complexity while maintaining texture removal effectiveness.
Solution Approach 2:
Instead of processing the entire image with full computational complexity, the method applies processing selectively to specific patches that contain texture. By identifying and processing only the necessary regions, the computational burden is reduced while still achieving effective texture removal where needed.
2Reliability
If conventional smoothing techniques are used to preserve edges, then edge preservation is improved, but explicit texture removal capability is limited
Solution Approach 1:
Different processing strategies are applied to different regions of the image based on local characteristics. Patches containing edges are processed differently from patches containing texture, allowing the method to preserve edges where needed while removing texture in appropriate regions, thus achieving both edge preservation and explicit texture removal.
Solution Approach 2:
The method changes processing parameters based on patch characteristics. By analyzing patch properties and adjusting processing intensity and type accordingly, the system can preserve edges in regions requiring structural integrity while applying stronger texture removal in regions where texture is the primary concern.
3Productivity
If patches at edge boundaries are processed using conventional methods, then processing completeness is improved, but structural edges become smudged
Solution Approach 1:
The method takes preliminary actions to prevent edge smudging before it occurs. By detecting patches that contain or are near edge boundaries and applying protective processing strategies to these patches, the system prevents the smudging effect that would otherwise occur during standard texture removal processing.
Solution Approach 2:
Before performing texture removal processing, the method preliminarily identifies patches containing structural edges and marks them for special handling. This preliminary action ensures that edge-preserving processing is applied to the correct regions, preventing smudging while maintaining processing completeness for the overall image.
Data Source
AI summary
Disclosed is a method and device for filtering texture, using a patch shift. A patch-based texture filtering method for removing texture from an image comprises the steps of: distinguishing a structure edge from patches for each pixel of an input image; and performing a patch shift for the patches of each pixel, on the basis of structure edge information of each of the patches to select a representative patch of each pixel. Accordingly, the method can generate a result image in which texture has been automatically and completely removed from a piece of the input image.