光栅立体视觉三维表面重建系统及其图像处理方法(二)  

A3D Surface Reconstruction System Based on Grid Stereo Vision and Its Image Processing Techniques(II)

作　　者：吴剑波[1] 骆清铭[1] 谭玉山[2] 

机构地区：[1]华中科技大学生物医学光子学教育部重点实验室,武汉430074 [2]西安交通大学激光红外研究所,西安710049

出　　处：《计算机工程与应用》2004年第13期11-13,19,共4页Computer Engineering and Applications

基　　金：国家科委863高技术研究发展计划项目(编号:2001AA234071);华中科技大学博士后基金项目资助

摘　　要：该文针对网格立体视觉三维表面重建系统提出了一种由粗到精的网格立体图像匹配方法,该方法将基于区域的图像相关匹配技术与基于特征的匹配技术相结合,充分利用了前者的精确性和后者的简洁性等优点。在粗匹配过程中,将特征点的提取过程与匹配过程统一起来,在提取特征的同时对特征点进行必要标记,无需进行特征点的属性分析和设置,减小了特征匹配的难度,避免了匹配歧义性,匹配由粗到精进行,既保证了匹配的准确度,又保证了匹配的精确性。An image matching method for grid stereo images is presented.From coarseness to precision,the method is contrived especially for the3D surface reconstruction system based on the stereo vision principle and the raster projec-tion.By integrating the area-based image correlation technique(ABIC)and the feature-based image matching technique(FBIM),the method described here utilizes the accuracy of ABIC and the conciseness of FBIM,and also it unites the feature points extracting process and the image matching process together by labeling the feature points in the extract-ing process,which can both reduce the matching complexity and avoid the mismatching difficulty.

关 键 词：图像匹配 立体视觉 光栅投影 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]