A Fast Level Set Method Without Solving PDEs
Yonggang Shi, William Clem Karl · 2006
In this paper, we propose a novel and fast level set method without the need for solving PDEs (partial differential equations) while preserving the advantages of level set methods, such as the automatic handling of topological changes. The foundation of our method is the direct use of an optimality condition for the final curve location based on the speed field. By testing this condition, only simple operations like insertion and deletion on two lists of boundary points are needed to evolve the curve. Our method is suitable for a set of general evolution speeds that are composed of two parts: an external speed derived from the image data and a speed term imposing boundary smoothness or regularization. In our experiments, we demonstrate that our algorithm is approximately two orders of magnitude faster than previous optimized narrow band algorithms for image segmentation tasks.