月球CCD影像的快速匹配算法  

Fast Matching Algorithm of Lunar CCD Image

作　　者：王君铎[1,2] 汪日伟[1,2] 温显斌[1,2] 邹永廖[3] 郑永春[3] 王玲[1,2] 

机构地区：[1]天津理工大学天津市智能计算及软件新技术重点实验室,天津300384 [2]天津理工大学计算机视觉与系统教育部重点实验室,天津300384 [3]中国科学院国家天文台,北京100012

出　　处：《激光与光电子学进展》2012年第1期63-70,共8页Laser & Optoelectronics Progress

基　　金：国家863计划重点项目(2010AA122201);国家自然科学基金(60872064)资助课题

摘　　要：影像匹配是基于CCD数据获得月球高程数据的基础和关键。研究了月球CCD影像的小波金字塔分层匹配算法。利用小波分析的多分辨率特性,对CCD影像进行小波分解,构造了金字塔顶层影像误匹配剔除策略及金字塔底层影像匹配点的快速搜索策略,实现了自粗尺度到细尺度的匹配算法。该算法通过减少非同名像点的匹配时间和搜索时间,大大降低了运算量,从而提高了匹配效率和配准精度。实验结果表明,该算法在匹配正确率和匹配效率方面都获得了很好的效果。Image matching is the foundation and key for obtaining lunar elevation information based on CCD data. We study the wavelet pyramid layered matching algorithms of lunar CCD image. CCD image is decomposed by wavelet according to the wavelet multiresolution analysis. Mismatch elimination strategy is constructed on the top of the pyramid level, and rapid searching strategy for matching points is constructed on the bottom of the pyramid level. Image matching is completed from low resolution to high resolution. Matching time and searching time for non- corresponding image points are reduced in this algorithm, accordingly, the calculation complexity is reduced greatly. So the searching efficiency and the registration precision are improved. The experimental results show that this algorithm has high precision and matching efficiency.

关 键 词：成像系统 影像匹配 特征点提取 误匹配剔除 小波金字塔 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]