Iterative Threshoding and Morphology Operation based Melanoma Image Segmentation

Abstract

Dermoscopy is a suitable diagnostic technique for biology observation of pigmented skin lesions used in dermatology. Nowdays there is great interest in the prospects for methods of automatic image analysis for dermoscopy image, it Provide quantitative information about the lesion, which can be of link the doctor, and as a standalone early warning tool. This paper presents a good method of melanoma images segmentation. It based on threshoding as segmentation and mathematical morphology used to remove unwanted part in order to obtain a better segmentation. The proposed method is compared with the famous method of segmentation of skin lesions in images dermoscopic such adaptive thresholding and fuzzy K-means clustering for the segmentation and evaluated with two metrics, False Positive Rate (FPR) and the False Negtive Rate (FNR) , using the segmentation results obtained by a dermatologist experienced and ground truth.