首页 | 本学科首页   官方微博 | 高级检索  
     

基于MMI的存在旋转的异类图匹配方法
引用本文:孙卜郊,周东华.基于MMI的存在旋转的异类图匹配方法[J].红外与激光工程,2006,35(6):742-745.
作者姓名:孙卜郊  周东华
作者单位:清华大学自动化系,北京,100084
基金项目:国家自然科学基金;国家重点基础研究发展计划(973计划)
摘    要:为了解决存在旋转的异类图匹配问题,在互信息(MMI,Maximum of Mutual Information)匹配方法的基础上,利用穷举法在参考图中找出实时图的对应匹配位置,并作为控制点,利用模型以及最小二乘方法找出两幅图像的对应关系。利用参考图的旋转变换较好地解决了存在旋转的异类图匹配问题;把控制点理论应用于灰度匹配得到的匹配点;把景象匹配的新成果(MMI)用于解决旋转异类图匹配。仿真实验表明此方法可较好地解决存在旋转的异类图匹配问题。

关 键 词:异类图    互信息    景象匹配    坐标转换
文章编号:1007-2276(2006)06-0742-04
收稿时间:2006-06-27
修稿时间:2006-07-06

Rotated image registration method based on MMI
SUN Bo-jiao,ZHOU Dong-hua.Rotated image registration method based on MMI[J].Infrared and Laser Engineering,2006,35(6):742-745.
Authors:SUN Bo-jiao  ZHOU Dong-hua
Affiliation:Department of Automation, Tsinghua University, Beijing 100084, China
Abstract:In order to match the multisensor images with rotation,a method based on the MMI(Maximum of Mutual Information) is proposed.Control points are obtained by finding the matching position of sensed image in the reference image exhaustively.Then,transformation parameters are obtained by Least Square Method.The innovations are as follows.Firstly,reference image is rotated to suit for the rotated multisensor image registration.Secondly,control points(CPs) theory is used in points got by intensity-based method.Thirdly,the new method(MMI) is used in rotated multisensor image registration.Experimental results verify its effectiveness on rotated image registration.
Keywords:Multisensor images  Maximum of Mutual Information  Image registration  Coordinate transformation
本文献已被 CNKI 维普 万方数据 等数据库收录!
点击此处可从《红外与激光工程》浏览原始摘要信息
点击此处可从《红外与激光工程》下载全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号