An effective segmentation method for dermoscopy images

Abstract

In this paper, an improved segmentation method of the pigmented skin lesions have been proposed in which to achieve the high characteristics of segmentation, converting the RGB images to U channel of YUV color space and noise reduction with fourier-domain filtering properties is done in pre-processing step. Also, Otsu thresholding and morphological reconstruction algorithms are used respectively in segmentation and post-processing steps. The experimental results reveal that the proposed scheme not only attains satisfactory sensitivity, but also has a higher accuracy (about 96.13%) and specificity (about 97.74%) compared to the related existing methods.