计算机工程与应用 ›› 2011, Vol. 47 ›› Issue (9): 160-162.
• 图形、图像、模式识别 • 上一篇 下一篇
陈庆芳,吴小俊
收稿日期:
修回日期:
出版日期:
发布日期:
CHEN Qingfang,WU Xiaojun
Received:
Revised:
Online:
Published:
摘要: 针对传统互信息缺乏利用空间信息而容易导致误配的缺点,提出了基于分块互信息的多模图像配准方法,并运用于可见光与红外图像之间的配准。该方法首先将可见光与红外图像分块,求得每个可见光与红外图像块对的互信息,并由块对中可见光与红外图像的质心间的距离为参数,确定块对的配准系数,求得每个块对的互信息与配准系数的乘积的和,定义为分块互信息,并以此为配准准则。实验表明,该方法运用与可见光与红外光配准,在配准精度上优于传统互信息方法。
关键词: 互信息, 多模图像配准, 分块互信息, 配准参数, 质心
Abstract: It would cause mistakes in the process of image registration due to absence of spatial information in the conventional registration method based on mutual information.A multi-modality image registration method is proposed based on block mutual information,which is applied to the registration of visual and infrared image.Firstly,both the infrared and visual images are divided into parts,based on which,the mutual information of each block-pairs is computed.Secondly,the distance between the centroid of the infrared and visual images in each block-pairs is computed,and the parameters of registration are obtained.The sum of the products between the mutual information and the registration parameters for each block pairs is computed,which is used as the criterion of image registration.Experiment results show that,image registration based on block mutual information will get more accurate result than traditional method.
Key words: mutual information, multi-modality image registration, block mutual information, registration coefficient, centroid
陈庆芳,吴小俊. 基于分块互信息的图像配准[J]. 计算机工程与应用, 2011, 47(9): 160-162.
CHEN Qingfang,WU Xiaojun. Image registration based on block mutual information[J]. Computer Engineering and Applications, 2011, 47(9): 160-162.
0 / 推荐
导出引用管理器 EndNote|Ris|BibTeX
链接本文: http://cea.ceaj.org/CN/
http://cea.ceaj.org/CN/Y2011/V47/I9/160