Extraction And Analysis Of Skin Texture Image Feature

With the development of computer image processing technology, people began to discuss extracting the feature of skin image by digital image processing technology. By means of digital image processing technology, people could make estimated methods of skin surface from traditional qualitative analysis to accurate quantitative analysis, so the precision of measure would be improved greatly. The purpose of this paper was analyzing the texture of skin surface quantitatively by digital image processing technology. It provided evidence for quantitative evaluation of skin surface, and it also provided technical basis for developing and exploiting quantitative analysis of skin image.This paper devoted to analyze the feature of skin texture completely. First, based on feature of skin texture it proposed the image preprocessing method of Gaussian filtering and Wiener filtering to remove the noise of skin image, and it made texture image beneficial to analysis and treatment; Then it introduced the Sobel operator, which was a method of image sharpening. This method was applied to transform texture image into gradient image. It was necessary procedure of carrying on watershed image segmentation; This paper also proceeded to binary converting in image processing for skin texture by using of Ostu algorithm, it would prepare for skin texture from the view of morphologic particle size analysis.Second, this paper introduced common method of texture analysis, it analyzed texture image by using of simple statistical analysis method such as gray histogram and gray-level interpolation histogram. It mainly extracted characteristic parameters of skin texture image through gray level co-occurrence matrix. It reflected the character of skin texture through the statistical numerical value of gray level co-occurrence matrix, such as moment of the inertia, contrast ratio and correlation.Finally, this paper gave grain size analysis method based on mathematical morphology and also gave the method of watershed texture segmentation. The former combined grain size analysis with mathematical morphology, it would obtain the total area, the maximum area, the minimum area and those corresponding numbers of texture image. This method also could provided the basis for recognition and retrieval of texture image. In this part it also extracted skin cleavage lines with the method of watershed based on marker, then it calculated out the characteristics of texture lines quantitatively such as transverse and longitudinal tendency, the number of crossing points.This paper had studied on several methods of texture analysis, it made feature analysis of skin texture at different angles. It found the features of skin texture at different ages. It compared and analyzed characteristic value such as the statistics of gray level co-occurrence matrix, the size of texture blocks, and the number of texture crossing points, which existed in the aged people, the middle age people and the young people. It obtained very good achievements.