单目视觉定位方法研究综述  

Research Overview of Location Method for Monocular Vision

作　　者：李荣明[1] 芦利斌[1] 金国栋[1] 

机构地区：[1]第二炮兵工程学院602教研室,西安710025

出　　处：《现代计算机》2011年第11期9-12,共4页Modern Computer

摘　　要：根据单目视觉定位所用图像帧数不同把定位方法分为基于单帧图像的定位和基于双帧或多帧图像的定位两类。单帧图像定位常利用已知的点特征、直线特征或曲线特征与其在图像上的投影关系进行定位,其中用点特征和直线特征的定位方法简单有效而应用较多;基于双帧或多帧图像的定位方法因其操作复杂、精度不高而研究的还较少。通过对各方法的介绍和评述,为单目视觉定位问题的研究提供参考。The location methods based on monocular vision fall into two classes,which are the location method from single image and two or more images, based on the different number of the images needed. The location methods from single image use the projection of the point features, straight line features, curved line features to estimate the pose of the camera.The location algorithm based on point features and straight line features is simple, effective and greatly used. There are few studies on the location method from two or more images, because the location algorithm is complex in computing and less exact. Introduces and reviewes all kinds of location methods, which provides reference for the research of location method based on monocular vision.

关 键 词：单目视觉 视觉定位 位姿估计 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]