Shape sharing initialized active contour model for image segmentation

Initial contour learning plays a very important role in the active contour model segmentation. In this paper, a novel shape sharing based method is proposed for initial shapes learning. The main insight of the method is that shapes are often shared between objects of different categories. To exploit the “shape sharing” phenomenon, the local shapes of the test images are firstly extracted. Then the matched local shapes set in the exemplar database is found. The object shapes from the exemplar database are subsequently transferred to the test image based on the size and relative location of the local shapes. Finally, the initial shapes are obtained in accordance with the global shape coverage. We regard these initial shapes as the initial involution functional of the active contour model. In addition, the active contour model integrates the boundary of the color gradient with the region information. The results show that our scheme of initial shape learning could express the shape information more efficient and the segmentation results are more accurate.