该文对Chan—Vese提出的水平集图像分割算法进行了改进，提出了分段光滑的Mumford—Shah全局优化的水平集图像分割模型，并对偏微分方程进行了修正，以提高模型的图像分割能力。实验表明，该方法不但解决了C—V方法对于灰度值渐进图像无法正确分割的问题，同时可更精确地描述原图像，是一种高效、稳定的图像分割模型。另外，针对水平集方法中符号距离函数构造计算量大的问题，还提出一种全邻域源点扫描法，以便通过对图像平面网格点的扫描来实现距离函数的快速计算，这种方法不仅计算性能稳定，而且速度快、精度高。

In this paper a new image segmentation model based on techniques of curve evolution, piecewise smooth Mumford Shah functional for segmentation and level sets was proposed as improvement of C V method. New method shows global optimization and less insensibility of initialization and can detect objects whose boundaries are not necessarily defined by gradient and solves the problem of locating the edges on images with non uniform brightness, for which the previous methods based on piecewise constant Mumford Shah model, including the C V method, are not applicable. Besides, the model was improved for the location of subtle and complicated edges of target objects by the modification of PDE. In order to further stabilize and fasten the level set evolution procedures, the paper addresses an improved approach to construction of the signed distance function using new Voronoi source scanning method, which needs simple comparison and few multiplication operations, faster than the traditional approaches. Finally, various experimental results for synthesized and real images will be presented to prove the proposed model efficiency and stabilized.