Otsu's Multi-Thresholding and Active Contour Snake Model to Segment Dermoscopy Images

Melanoma is one of the deadliest form of skin cancer widely which affects human beings. Due to its growing occurrence rates, it is essential to develop an assisting methodology to support the present clinical detection procedures. Digital dermatoscopy is a clinically proven process to record and analyse the suspicious regions of skin. Extracting the uncertain region from dermoscopy image is generally preferred to have a clear idea about skin infections. The success of skin infection forecast relies primarily on the chosen image segmentation and analysing tool. In this paper, heuristic algorithm assisted multi-thresholding and Active Contour Snake Model (ACSM) based segmentation is proposed to extract the cancerous section from the image dataset. The experimental results obtained with ACSM are then validated with the segmentation results of Localized Active Contour (LAC) and Regularized Level Set (RLS) approaches. The results confirm that proposed approach offers better values of image similarity index and statistical measures compared with the alternatives.