An Improved FFCM Based Segmentation Method for Dermoscopic Images

Dermatoscopic image segmentation is the most important part to diagnose the skin cancer. In this paper, an improved method is proposed for attaining the accurate segmentation of dermatoscopic images. In the proposed method, the U channel of converted dermatoscopic image is enhanced after filtering and intensity adjustment. Then, it is classified in two classes using statistical histogram based fast version of fuzzy c-mean. Subsequently, Otsu thresholding is applied to class one, for obtaining a suitable binary mask of image. An improved segmented dermatoscopic image will be achieved after performing morphological reconstruction algorithms on the binary image. The experimental results show that the proposed method satisfies sensitivity and specificity more than the other earlier works.

Keywords:

Dermatoscopic image segmentation , U channel , Intensity adjustment , Fast fuzzy c-mean , Otsu thresholding