Core Concepts
Proposed TEP method efficiently detects texture edges using patch consensus.
Abstract
The Texture Edge Detection by Patch Consensus (TEP) method is a training-free approach that utilizes local patches to capture similarities and differences between textures. By analyzing patch responses within a larger domain, the proposed model effectively identifies texture boundaries without detecting edges within textures. TEP demonstrates robustness against noise, multiple junctions, and varying scales of textures. The edge function V generated by TEP enables color segmentation based on chromaticity and brightness components, providing sharp texture edge detection while diffusing within regions.
Stats
Various experiments are presented to validate the proposed model.
Parameters used: r = 5, R = 20, λ = 0.018 for noise robustness experiments.
Parameters used: r = 5, R = 20, λ = 0.018 for Salt and Pepper noise experiments.
Parameters used: r = 5, R = 20, λ = 0.018 for multiple junctions experiments.
Parameters used: r = 5, R = 20, λ = 0.018 for image segmentation experiments.