Image registration based on mutual information (MI) has become an increasing popular match metric duing to it doesn' t relay on the image' s gray level and can realize automatic standerdising, multimodality medical image registration based on mutual information is deeply discussed mainly from the famework of registration, we have done the main reasearch of the method of image registration based on maximization of mutual information and used partial volume interpolation counted the value of column diagram, calculated the value of mutual information, analyzed its advantages and disadvantages. To speed up the rate of image registration and overcome the large calculation and the presence of local extremum of mutual information, we have focused on optimization strategy, based on discussing and analysing the common optimization strategy,we have put forward the improved optimization strategy. Because the common powell strategy dosen' t regard to the problem of linear independence, this design used the improved powell strategy that can make the direction of search linear independence and increase the degree of adjoint. This design also make the PSO optimization strategy compare with powell strategy,finaly make the analysis through simulation.%基于互信息的配准方法有不依赖于图像本身灰度,可实现自动校准等优点,针对基于互信息的多模态医学图像配准方法进行深入研究,从图像配准的框架入手,着重研究了基于最大互信息的配准方法,用PV插值法统计联合直方图的值计算出互信息值,分析了互信息作为配准的测度函数具有的优点和存在的缺点.为了加快配准速度,针对互信息计算量大和存在局部极值的问题,集中于优化策略的研究,在一般优化算法的讨论分析基础上,提出了改进的优化算法,针对一般Powell法不考虑线性无关问题,采用了改进后的Powell法,可以使搜索方向线性无关,共轭程度增加,并且引入了PSO优化搜索算法与Powell算法进行比较,最后通过仿真做定性和定量分析.结果表明,Powel1优化算法明显优于PSO优化算法.
展开▼
