A novel image segmentation algorithm based on convex polygon edge detection
Lian Yishao, Jiajun Bu, Chun Chen, Song Mingli · 2004
A novel image segmentation technique: closed edge circular scan (CECS) is proposed to locate the convex polygon edge of the central object in an image. Conventional edge detection approaches are employed to trace object boundaries. One key to this approach is a circular edge scan algorithm with a dynamic radius, which connects separate edges into a semiclosed edge. A semiseed-fill algorithm is performed to finish the segmentation of the image. Test has been done on numerous images of various kinds. The experiment result shows good accuracy. This algorithm is efficient and robust in segmentation of object with convex polygon object.