摘要
互信息作为多模医学图像配准的测度函数被广泛应用,但是当配准图像不理想时,互信息配准方法可能会失效.为了提高配准的精度及稳定性,提出了一种改进加权互信息的医学图像配准方法.首先,由图像直方图辅助的方法选取一定的感兴趣区域;其次,分别计算感兴趣区域的互信息和整体图像的互信息;最后,由两个不同的互信息加权组合构造出两个新的测度函数并作为图像配准的依据.对3个测度函数的特性、配准误差和配准时间进行实验比对,结果表明改进的方法具有较高的精度和较好的鲁棒性.
The mutual information is widely used as the measure function in multimodality medical image registration. But this method might fail when registering images are not ideal. In order to im- prove the stability, an improved method of weighted mutual information is proposed. Firstly, a cer- tain region of interest is selected by the method of histogram assisted. Then, the mutual information of the local region of interest and the global mutual information are calculated respectively. Finally, by combining two different weighting mutual information, two new registration measure functions are constructed as the basis of image registration. The experimental results show that the improved meth- od has higher accuracy and better robustness by the comparison of characteristics, registration errors and registration times for three measurement functions.
出处
《测试技术学报》
2014年第2期137-142,共6页
Journal of Test and Measurement Technology
关键词
加权互信息
图像配准
测度函数
感兴趣区域
加权概率互信息
weighted mutual information
image registration
measure function
region of interest
probability of weighted mutual information