A Segmentation Algorithm on Color Images Using Extenics Theory

Extenics theory was introduced into color image segmentation. Similar pixels were classified into a same area by their pixels´ RGB values which were regarded as eigenvectors of them. Classical and extensional ranges of different areas in the color image were determined by several images taken in different scenes, then dependent degrees between pixels and different areas of the color image were computered, and the same work to weight-added coefficients of RGB values of pixels, finally, weight-added dependent degrees between every pixel and every area were calculated, the maximum could tell us which area a pixel belonging to. We emulated by processing a traffic image using this method, the result showed the validity of it.