Color Image Decomposition and Fuzzy Clustering for Dermoscopic Image Segmentation

Abstract

This paper proposes a new approach for automatic and effective segmentation of dermoscopic images. The method of segmentation is based on a pre-processing using the color structure-texture image decomposition based on Partial Differential Equations, and a step of segmentation by fuzzy clustering based algorithm. The proposed approach has been implemented and compared with other recent algorithms for dermoscopic image segmentation in order to demonstrate sufficiently good results, the proposed technique was be evaluated as performance for lesion detection in dermoscopic images, and the results are 78% of specificity, 99% of sensitivity and 0.99 of the AUC metric.