基于简单Schur凹函数的图像配准测度研究  

Novel Measures Based on Simple Schur Concave Function for Image Registration

作　　者：杨静[1] 胡顺波[2,3] 刘常春[2] 杨金宝[2] 

机构地区：[1]山东大学计算机科学与技术学院,山东济南250101 [2]山东大学控制科学与工程学院,山东济南250061 [3]临沂师范学院物理系,山东临沂276005

出　　处：《电子学报》2008年第12期2328-2332,共5页Acta Electronica Sinica

基　　金：国家"863"计划资助项目(No.2006AA02Z4D9);山东省自然科学基金资助项目(No.Z2006C05)

摘　　要：使用互信息或归一化互信息进行图像配准时,由于噪声、模态、插值等影响,测度函数存在许多局部极值,收敛范围较窄,有可能导致误配准.该文根据一个简单的Schur凹函数,充分利用它的特殊上凸性来消除噪声等引起的小概率分布,并由Jensen-Schur测度、广义距离测度和f信息测度的定义,构造了六种新测度.从运算时间、收敛性能、抗噪鲁棒性方面,对这六种测度、互信息和归一化互信息进行了比较和分析.实验结果表明,Jensen-Schur-beta和D-beta测度的收敛性能优于其它测度,抗噪声能力强于其它测度,运算速度快于互信息和归一化互信息.When the influence of noise,interpolation and image modality is considered, the image registration method based on mutual information or normalized mutual information may cause local extrema, small convergence area, and even inaccurate registration. According to a simple Schur concave function and the definition of Jensen-Schur measure, generalized divergence measure and f information measure, six new measures were constructed. The Schur function with special concave characteristics can filter the small probability distribution caused by noise, interpolation and so on. The characters of six new measures, mutual information and norrnalized mutual information are analyzed and compared by applying them to rigid registration. The results of tests show that Jensen-Schur-beta and D-beta measures outperform other measures in convergence performance and noise ilmnunity, and they are time saving in comparison with mutual information and normalized mutual information measures.

关 键 词：图像配准 Schur凹函数 Jensen-Schur测度 互信息 

分 类 号：R318[医药卫生—生物医学工程]