Malignant Melanoma, essentially the most deadly type of dermis cancer, has a good prognosis if dealt with in the curable early levels. Early prognosis and surgical excision is probably the most robust treatment of melanoma. In this article, we use some data set clinical dermatoscopic images of skin lesions, in which some images are of malignant lesion and some images are of benign lesion, image pre-processing and segmentation techniques is used to distinguish melanoma from benign pigmented lesions. This study reward Asymmetry, Border Irregularity, color variation and Diameter characteristic and extraction of Dermatoscopic images for melanoma dermis cancer prognosis. This selection is used to calculate Total Dermatoscopic Score (TDS) for melanoma epidermis melanoma analysis. The pertinent proves that Havrda Entropy and Harris Corner Detector based melanoma diagnosis approach achieve more sensitivity with respect to Otsu and Harris based combined approach. While as compared to prior arts the sensitivity performance of proposed scheme outperforms the standard arts of Zagrouba and Joanna. The targeted model achieved an specificity and sensitivity of 27.48% and 92.45% respectively.
Keywords—malignant melanoma; image segmentation; entropy; dermatoscopic.