Removal of Hair Particles from Skin Disease Images Using Pixel Based Approach

Abstract

A methodology for digital hair removal from clinical dermoscopic images, based on mathematical operations is proposed in this paper. The approach used in case of grayscale images, is based on edge property of the images, where a circular mask is used for removal of non-skin element and reconstructing the image by normalized pixel values. In case of RGB images, based on histogram values, frequency of occurrence is measured through individual components. A list is deduced from this measurement that best defines the image. Neighborhood operation based on minimum distance approach between each pel component is the key strategy. It preserves the border properties. This method is applied on enormous number of image data and is observed to be produced suitable results