Texture and color feature extraction for classification of melanoma using SVM

Abstract

Feature plays a vital in the domain of image processing. The various features of an image are color, texture, shape or domain specific features. Texture is considered as one of the main feature of any image. The second order statistical features for an image is obtained by Gray level co-occurrence matrix (GLCM) and it operates on spatial domain. The haralick texture features are energy, entropy, homogeneity, correlation, contrast, dissimilarity and maximum probability. The aim of the paper is to classify the dermoscopy images into melanoma and non-melanoma by considering the texture and color features of an image. GLCM is used to extract the texture features of an image. Color histograms are used to extract the color features in three color spaces namely RGB, HSV and OPP. Support vector machine (SVM) is used for the process of classification. The performance of the proposed system is evaluated by the metrics sensitivity and specificity. The experimental result shows that the texture combined with RGB color space provides better classification accuracy.