摘要
针对双目立体匹配中存在的误匹配、视差场稀疏等不足,提出了一种将区域匹配和特征匹配相结合的块匹配算法。对匹配图像对进行极线校正和限定搜索区域,采用SAD算法进行相似度测量。对小型无人机动平台的双目系统采集的任意一幅图像对用不同大小的匹配窗进行匹配。对结果进行了分析,结果表明了该方法的良好效果及实用性。
A new stereo patch matching algorithm combining region matching and feature matching is proposed to solve the deficiency about error matching and sparse disparity, polar rectification（is carried out and match searching region is limited to before image pair matching, then SAD is adopted in similarity measurement. It is experimented by an arbitrary image pair acquired from small unmanned mobile platform＇ s binocular stereo system, the matching results show that this matching method is effective and practicable.
出处
《计算机工程与设计》
CSCD
北大核心
2012年第1期273-277,共5页
Computer Engineering and Design
