摘要
基于多摄像位的空间定位即从来自多个摄像位的视频截图中获取物体的空间三维几何信息。本系统主要由三个部分构成,分别是:摄像机标定部分、立体匹配部分、基于点的三维重建部分。本文主要围绕这几个部分展开研究:首先,在摄像机参数标定中,采用正友的基于平面标定模板的摄像机标定方法,通过实验和计算获取摄像机的参数;其次,立体匹配部分采用了Burchfield算法,试验结果得到了视差图;最后,最后根据立体视觉基本原理,完成了三维坐标计算,得到了一个特定区域的点集信息(如球类)。本文最终设计并实现了一个应用于体育节目包装的基于多摄像位的空间定位系统。
Multi-position based space position setting system obtains the three-dimensional geometric objects information from multiple video screenshot of the camera.The system mainly consists of three parts：camera calibration,3-D matching and 3-D reconstruction.This paper mainly studies the followings：firstly,using the Zhang＇s Camera calibration technique to obtain the camera parameters;secondly,using the Burchfield＇s algorithm to obtain the disparity map;at last,according to the basic principles of stereoscopic vision,obtain the 3-D information of the object which we want to position.The implementation programs and the experimental results have been given in the paper.
出处
《微计算机信息》
2010年第7期208-210,共3页
Control & Automation
关键词
三维重建
立体视觉
摄像机标定
立体匹配
3-D reconstruction
three-dimensional vision
camera calibration
3-D matching
