Texture Classification and Segmentation using Roughness Feature Extraction through Edge Detection

Abstract

In this paper, we propose a new rotational invariant roughness feature extraction method for texture image classification and segmentation. Roughness feature extraction is carried out through edge detection, followed by roughness value calculation. We propose an edge descriptor that extracts edge information from an image. We also propose a method for computing the roughness value so to estimate the roughness level of an image. We used the k-nearest neighbor classifier for classification and segmentation. The proposed approach has been tested for rotational variant and illumination variant image classification and texture image segmentation. A few of the existing methods are compared with the proposed method, and it is found that this method outperforms traditional roughness estimation methods.