基于推广的长程外力场的库仑Snake算法应用于MR心室图像的分割  

A Coulomb Snake Model Based on Generalized Long-range External Force Field Applied to MRI Ventricle Segmentation

作　　者：王蓓[1] 张立明[1] 

机构地区：[1]复旦大学电子工程系图像与智能实验室,上海200433

出　　处：《复旦学报（自然科学版）》2004年第1期21-26,共6页Journal of Fudan University：Natural Science

基　　金：国家自然科学基金资助项目(60171036)

摘　　要：提出了一种Snake算法的新的外部能量模型,用基于库仑定律的图像库仑场代替传统Snake中的外力场,并对该外部能量场进行了推广,根据图像的边缘特征自动设定参数,使其外力具有真正的长程性.实验结果表明,该方法提高了分割精度,减少了外部能量场中存在的局部最小,从而大大减少了对初始轮廓的限制,同时减少了人工干预.A new external energy field for Snake model is proposed; the traditional external force fields are replaced by a new generalized external force based on Coulombs Law. According to the edge property of the input image,the parameters are set automatically to give external force the real long-range property. Simulation results show that with the method proposed, the accuracy of segmentation is improved and the local minima in the external energy field are reduced, then the constrictions on the initial curve are relaxed, and the manual operations are also reduced.

关 键 词：图像处理 主动轮廓线模型 库仑定律 边缘提取 图像分割 

分 类 号：TN911.73[电子电信—通信与信息系统]