【作者】 李建国； 潘振宽； 魏伟波； 石洁； 李华；

【Author】 Li Jianguo,Pan Zhenkuan,Wei Weibo,Shi Jie,Li Hua (College of Information Engineering,Qingdao University,Qingdao 266071)

【机构】 青岛大学信息工程学院；

【摘要】 本文针对隐式曲面上图像分割的问题,提出了一种基于变分水平集方法的隐式曲面上图像分割的Chan-Vese模型。该模型将平面图像上两相图像分割的Chan-Vese模型推广到隐式曲面上的图像分割。为定义隐式曲面上图像区域划分的动态轮廓线,引入另一个动态水平集函数,其零水平集与隐式曲面的交线即为曲面上区域划分的轮廓线。所研究模型的能量泛函的数据项即为曲面上两分割区域的图像强度与对应区域平均图像强度的差的平方,其轮廓线长度项为两水平集函数的零水平集交线的长度。通过变分方法得到图像分割空间轮廓线演化的梯度降方程。通过显式差分格式对演化方程进行离散。实验结果表明,该模型对复杂形状上的复杂图像具有良好的分割效果,并且在图像分割的同时能够完全保持曲面固有的几何特征。更多还原

【Abstract】 A new model to segment the images on implicit surface based on the Chan-vese model and variational level set method was proposed.The classic Chan-Vese model for two phases segmentation of planar images is extended to image segmentation on implicit surfaces.The spatial contour for image segmentation is defined by means of intersection of the above implicit surface and the zero set of another dynamic level set function.The novel energy functional includes two parts,the first one is the sum of square of the difference between pixel intensity and mean values in two regions on the surface,the second part is the length of spatial contour.The evolution equation of level set function is obtained using variation technique and is discretized by simple explicit finite difference method.Experimental results show that the model has promising effect on image segmentation on implicit surface and preserves salient features of the surface.更多还原