基于快速水平集算法的图像分割方法

针对水平集算法计算量大,本文提出了一种使用改进的快速水平集算法的图像分割方法。本文方法是一种极限的窄带法;首先水平集的曲线进化用两个序列元素之间的交换实现;其次提出了平滑过程处理进化后的曲线;本文方法不需要解偏微分方程,运算速度很快,而且能够得到和标准算法相近的结果;实验结果证明本文算法速度快,适合于实时图像系统中的应用。

Image Segmentation based on Fast Level Set Method

An improved fast level set method was proposed to reduce large computation load of level set method.Our method was a kind of extreme narrow band one.Firstly,curve evolution process was replaced by exchanging elements of two sequence.Then a smooth stage was approached to process the evolutional curves.Speed of our method was faster than previous methods because of not requiring solving Partial Differential Equations(PDEs),while similar results could be achieved.Experiments show that this method is fast and ap- plicable for real time application of image processing.