A Multi-stage Active Contour Algorithm

A multi-stage active contour algorithm is present in this paper. The original image is analyzed with one-level wavelet, and then an approximate image can be obtained. In the growing stage, active contour is attracted to the vicinity of the approximate image by using greedy method. In adult stage, the B-spline shape is used for active contour in the original image. And the dynamic programming algorithm makes it the global best. Experiment proves that active contour could converge fast to the real contour, and fits it well.