An Image Segment Algorithm Based on Concave Points and Skeleton Features

In order to identify the different components in a hybrid fiber image, this paper develops a new image segmentation algorithm. Two principles are defined in this paper to judge whether the Public Intersected Line Segment (PILS) is pseudo or not. After PILS judgment, the concave points are classified according to the corresponding relationship between the PILS in an overlapped-fiber image, or cross points in an adhesive-fiber image, which can be used to split different fibers in the binary image. The experiment result shows that the new algorithm is have a good segment effect on the situation that several fibers overlapped or adhered to each other. Compared with the existing research, the new segmentation method establishes a mapping relationship between the concave points in the binary image and the intersected points or PILS in the skeleton image.