提高水平集方法初始化计算速度的研究

Speed up Initialization Time for the Level Set Method

本文对水平集方法的初始化步骤进行了改进。在水平集方法中，初始化是一个很重要的步骤。它包括计算图像平面中每个点的初始水平集距离，同时找到与当前点距离最近的曲线上的点。这是一个费时的过程，影响到整个水个集方法的计算速度。本文基于快速推进法提出了一种快速初始化方法。这种方法只需要Ｏ（Ｎ）的时间就能够遍历所有的图像点，完成初始化。最后，对三种不同的初始化方法：直接法、快速推进法和本文的方法在计算速度上作了对比。实验表明，最终得到理想的分割结果的情况下，本文的方法在时间上优于其他两种方法。

In the paper, we modify the initialization step for the level set method. The initialization is an important step in the level set method. The main task of this step is calculating the level set distance of a point in the image plane and finding the nearest point on the curve of the current point. However, this step is a computational time-consuming step. Our modified initialization step is based on the fast marching method. It takes only time O (N ) to accomplish the whole initialization process. Finally, we compare the modified initialization step with the fast marching method and the direct method. The results show that the modified method takes less time to access all the data than the other two methods while still obtains the ideal segmentation results.