Skin Classification And Feature Extraction Based On Neural Network Analysis Tool

With the progress of science and technology, people’s living standard have improved greatly, and have paid more and more attention to pursue the beauty because of skin as the reference of the external performance. So, how to analyze the skin with objective and quantitative methods and the texture recognition have widely discussion and application in many regions. In this paper, the image processing technology, such as a tool based on neural network approach was used to apply for the classification and feature extraction.Firstly, the problems of current research and future trends were presented. And then, the study of the spatial gray level co-occurrence matrix and fractal dimension method were analyzed and comprised to determine the feasible method. The result indicated that the spatial gray level co-occurrence matrix is the appropriate method.Secondly, the effective image preprocessing method was designed to character the skin through analysis of the existing image preprocessing methods.Lastly, the NNTool of MATLAB was used to train in the network in the experiment of skin texture image. and the simulation of mice skin texture was carried out. The results indicated that the classification accuracy reach90%. Experimental results proved that the analysis tool based on neural network applied the skin of classification and feature extraction method is feasible.