基于极少信息的任意照片目标点定位算法  

Target points positioning algorithm based on rare information in uncalibrated camera's photos

作　　者：刘吟啸 韦巍[1] 

机构地区：[1]浙江大学电气工程学院,杭州310027

出　　处：《计算机应用研究》2015年第1期272-275,共4页Application Research of Computers

基　　金：国家"863"计划重大项目(2011AA050204);中央高校基本科研业务费专项资助项目(2013QNA4021);杭州市重大科技创新产业链项目

摘　　要：为了达到通过任意摄像头拍摄照片就可对目标点进行定位的目的,针对如何用最少信息得到空间中目标点坐标问题,基于摄像机的成像模型、空间中固定点之间的几何约束,以及坐标系变换的基本原理,推导出一种类似于P4P的无标定照片的目标点定位方法。该方法可由观测到的两幅图像中已知世界坐标系中位置坐标的任意四点在图像中的位置,通过计算得到图像中目标点的三维坐标。此方法需要的已知信息极少,对于拍摄图像以及拍摄所用相机没有要求,并通过实验验证可行,精度相比传统标定方法没有明显损失,且较其他自标定定位方法更高,实用性较强。In order to achieve the purpose of positioning the target point from any camera photos, for information on how to use the least get the coordinates of the target point in space, based on the the camera imaging model, geometric constraints of the fixed point in space, and the basic principles of the coordinate system transformation, this paper obtained a target point positio- ning method through uncalibrated photographs similar to P4P. This method could calculate the three-dimensional world coordi- nates of the target point by observing two images＇ position coordinates of four points whose world coordinate was already known. Very minimal information was needed in this method, and it could be used with any cameras of photos. Experiment has been made to verify the feasibility of the algorithm. The accuracy compared to traditional calibration methods has no signifi- cant losses, and is higher than other self-calibration positioning methods.

关 键 词：机器视觉 无标定定位 P4P 双目视觉 几何约束 坐标系转换 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TP301.6[自动化与计算机技术—计算机科学与技术]