Shape Recovery Using Improved Fast Marching Method for SEM Image

摘要

Fast Marching method provides a solution to the Eikonal equation, and it also can be used to solve the Shape from Shading problem. But it still has some limitations. This paper proposes a new approach to recover 3-D shape by using improved Fast Marching method. The second-order finite difference, the diagonal grid points, and new update mode is used to improved FMM and the method can recover 3-D shape for Lambert image. Then we propose a method to modify the original Scanning Electron Microscope (SEM) image with intensity modification by using affine transform and NN learning, thus the improved FMM can recover 3-D shape from SEM image. Finally, the results were compared between the proposed method and previous method. Experiment includes numerical experiments, computer simulation experiments and real image. The results show the method is satisfied with Lambert and SEM image, and both robust and accurate.