一种新的多模态图像集成配准方法  

A new synthetic registration principle for multi-modal images

作　　者：常晓丽[1] 李金屏[1] 

机构地区：[1]济南大学信息科学与工程学院模式识别与智能系统研究所,山东济南250022

出　　处：《山东大学学报（理学版）》2009年第9期35-39,共5页Journal of Shandong University(Natural Science)

基　　金：国家自然科学基金资助项目(60873089);山东省研究生教育创新计划资助项目(SDYY08032);山东省教育科学规划课题重点资助项目(2008ZK0007)

摘　　要：提出一种新的多模态图像集成配准方法。该算法是在对齐度和归一化互信息2种方法的基础上,引入图像边缘检测和均衡化,从而综合利用6种配准算法进行集成多模态图像配准。使用少数服从多数和基于可信度的加权2种集成准则,对红外热像仪图像和相应可见光图像进行实验分析,结果证明该集成配准方法准确性高,鲁棒性强,较之非集成配准方法具有明显的优势。A novel synthetic registration method is put forward. Based on alignment metric and normalized mutual information, image edge detection and image equalization are employed, and six registration algorithms are integrated to operate synthetic multi-modal image registrations. Two criteria are that the minority should subordinate to the majority and the weight be based on the credibility. Results of infrared image and the corresponding visible light image registrations are analyzed, which indicate that the synthetic registration method has high accuracy and robustness, and has an obvious comparative advantage to non-synthetic registration methods.

关 键 词：多模态图像配准 对齐度 归一化互信息 集成配准准则 

分 类 号：TP751.1[自动化与计算机技术—检测技术与自动化装置]