A combined segmentation approach for melanoma skin cancer diagnosis

Abstract

This paper illustrates the methodology for melanoma diagnosis, a deadly skin cancer using computer aided image processing techniques. Here, the combination of Median filtering and Karhunen-Loeve transform are used in preprocessing part and also a combination of Active contours and Watershed Transformation algorithms are used in Segmentation process. When this combined method is applied on skin lesion images, the problems like over smoothing and over segmentation were solved.