Abstract: Surface reconstruction from contours is an important research aspect in reverse engineering, in which image segmentation is a key step. A new image segmentation method base on B-spline active contour model is proposed, and Greedy algorithm is adopted to solve this model, which can produce fast and stable segmentation results and is fit for surface reconstruction from contours application. Series magnetic resonance image is successfully segmented by this model, and from the results, surface reconstruction is realized, which demonstrate the feasibility of the method.