Recognition of Skin Diseases Using Curvelet Transforms and Law’s Texture Energy Measures

This work presents an automated system to recognize human skin disease. In many computer vision and pattern recognition problems, such as our case, considering only a single descriptor to mine one sort of feature vector is not enough to attain the entire relevant information from the input data. Therefore, it is required to apply more than one descriptor to extract more than one feature vector categories with different dimensions. In this paper, for the purpose of skin disease classification, we propose a new hybrid method which is the combination of two methods to proficiently classify different types of feature vectors in their original form, dimensionality. The first one uses the Curvelet transform method in spatial and frequency viewpoint and the second one uses the set of energy measures to define textures had been formulated by Law’s texture energy measure. Minimum euclidean distance of the Law’s texure energy measures between different species are calculated for discrimination.