Texture analysis method using regional rank coding

This paper presents a method of texture analysis based on regional rank. The method namely regional rank coding is to first determine the rank of the gray level of each pixel in a region whose size and shape depend on the gray level of the pixel being processed, the code of the regional rank is then calculated according to the found rank and also the gray level of processed pixel. This encoding allows highlighting the texture information while reducing the number of gray level. Texture is then characterized by attributes representing the frequency codes. Tests performed including image segmentation, show a good performance with an estimated classification accuracy exceeding 93%. This confirms the relevance of the proposed attributes and suggests good prospects for our method.