Segmentation and classification of burn color images

The aim of the algorithm described in this paper is to separate burned skin from normal skin in burn color images and to classify them according to the depth of the burn. The segmentation procedure consists of an elaborated treatment of color representation, followed by a grayscale segmentation algorithm based on the stack mathematical approach. The proposed algorithm has been developed to be applied to skin wound images, but it works properly as a general segmentation approach. In the classification part, we take advantage of color information by clustering, with a vector quantization algorithm, the color centroids of small squares, taken from the burnt segmented part of the image, in the (V₁, V₂) plane into two possible groups, where V₁ and V₂ are the two chrominance components of the CIE Lab representation.