Preprocessing Methods for Dermoscopic skin cancer images
Skin cancer is à significant disease prevailing in several countries and is a cause for frequent deaths among majority of age groups. Early diagnosis of skin cancer is important for reducing the mortality and for avoiding the cost and stress related with surgeries, biopsies and unnecessary therapies. Automated methods for early detection of skin cancer using skin images includes preprocessing of skin lesion images, segmentation and classification of lesions. The success of segmentation and classification algorithms depends mainly on the preprocessing techniques as they help in removal of several kinds of noise and imperfections from the skin images. In this paper, we have applied preprocessing techniques on skin image sample taken from PH2 skin cancer data set. For removing noise, median and gaussian filters have been used and their performance is compared using metrices like Mean Square error (MSE), Peak signal to noise ratio (PSNR) and Structural Similarity Index Measure (SSIM). Several Morphological operations are also implemented for removal of hair from skin cancer images.