A Two-Way Active Contour Model for Incomplete Contour Segmentation

Abstract

A two-way segmentation model is proposed in this article. The model is used to solve the problem that the objective contour can not be completely extracted from image due to occlusion between objects within similar image groups or image sequences. The proposed model first decomposes incomplete contours into sub-segments using local features identified by seed points set along each path. Then, locate the occluded part of the target object and reconstruct the target. Finally, a new vector field is generated based on the reconstructed object from the proposed model, followed by iterative evolution. The experimental results show that the proposed algorithm can better handle the problem of occlusion or misleading features of targets in composite images and medical images. Not only does it facilitate subsequent measurement and analysis, but it also preserves the original shape of the object during the segmentation process without prior information. It is worth noting that the accuracy of the proposed model is robust to our initialization strategy.