基于活动轮廓模型的脑部医学图像弹性配准  

Brain medical image elastic registration based on the active contour model

作　　者：唐祚[1] 闫德勤[1] 

机构地区：[1]辽宁师范大学计算机与信息技术学院,辽宁大连116081

出　　处：《微型机与应用》2015年第14期39-41,共3页Microcomputer & Its Applications

摘　　要：针对传统互信息弹性配准方法在医学图像应用上计算量大、处理速度慢的问题提出了一种基于活动轮廓模型(CVL-BFGS)医学图像配准方法。该算法结合了图像局部轮廓信息和全局变化信息,通过提取图像的边缘轮廓,可以有效地挖掘轮廓信息,并克服了弹性配准算法容易陷入局部极值问题,使图像配准的结果更加稳定。同时该算法为全局互信息配准提供一个通过局部配准得到的更优初始值,从而降低了整体配准的迭代次数,提高图像配准效率,并证明了该算法的鲁棒性和有效性。To solve the problems that the traditional elasticity of mutual information registration method in medical image application has large amount of calculation and slow processing speed, we put forward a method based on active contour model （CVL-BFGS） medical image registration method. The algorithm combines the local image contour information and global change information. By extracting the edge contours of the image, it can effectively tap the contour information, and can overcome the shortage that the elastic registration algorithm is easy to fall into local optima problem, so image registration results are more stable. The algorithm in the global mutual information registration provides better initial values obtained by a local registration, thereby reducing the number of iterations of overall registration, improving the efficiency of image registration, and proving the robustness and effectiveness of the algorithm.

关 键 词：流行学习 线性化 局部线性嵌入 降维 稀疏数据 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]