An efficient multiscale approach to level set evolution

In this paper, an efficient multiscale scheme for level set evolution is proposed. First, we are addressing the problem of passing the solution from the coarser scale to the finer one. Inspired by the idea of the entropy condition and its extension, an efficient passing solution method is presented, where neither extrapolation nor complex computation is needed. Thus it could induce fast convergence rate. Furthermore, an improved Hermes algorithm, called fast Hermes, is developed to fast implement the level set evolution on each scale by further loosening the constraint in the intermediate levels. Our approach is evaluated and compared to the existing algorithm. The experimental results are very promising.