A Hybrid Diagnosis System for Malignant Melanoma Detection in Dermoscopic Images

Among the generally occurring common skin cancer, Melanoma is said to be the most dangerous type of cancer. Many of the Computer vision techniques have adapted to detect the disease early days. In the similar way this paper proposes an image pattern classification to identify skin disease in images with a combination of texture and color feature extraction. The main aim of this paper is to find appropriate features that can identify skin disease. Initially, normal and diseased images are collected and pre-processed by converting the images into Grayscale by PCA and multilevel Otsu thresholding. In addition the post processing includes dilation and erosion techniques and canny edge detection for quantization. Then features of shape, color and texture are extracted from the images and these images are classified by support vector machine classifier. A combination of several features is used to evaluate the appropriate features to find distinctive features for identification of disease. When a single feature is used, shape feature has the lowest accuracy of and texture feature has the highest accuracy. A combination of texture and color feature extraction results highest classification accuracy.