一种基于模型的单相机姿态估计方法  

A Method for Pose Estimation Based on Single Camera and Object Model

作　　者：蔡伟 柯文贡 坚习飞 

机构地区：[1]国防科学技术大学电子科学与工程学院ATR重点实验室,长沙410073

出　　处：《中国图象图形学报》2008年第3期513-518,共6页Journal of Image and Graphics

基　　金：国家自然科学基金项目(40571103)

摘　　要：利用单幅影像测量目标在空间的位置和姿态参数(Pose)在目标定位、场景监测与更新、目标识别、自动导航等领域有着广泛的应用。目前有很多方法用于解决该问题,这些方法按照目标模型点与对应成像点的对应关系是否已知而分为两大类:一类是对应性已知条件下求解Pose;另一类是对应性未知条件下同时求解Pose与对应性问题。该文提出了一种新的采用单幅影像求解目标位置与姿态参数的方法。与已有的方法不同,该方法完全回避了与Pose问题并存的对应性问题。它利用目标模型投影与真实影像中目标区域的重合程度建立一个包含Pose参数的能量函数,将Pose求解转换为能量最优化问题,并采用了Hooke-Jeeves算法求解这个最优问题。多组仿真数据的仿真实验结果表明,该方法是正确和有效的。The problem of Pose estimation based on single camera arises in many areas of computer vision, including object recognition ,site inspection and updating, object tracking and autonomous navigation. Many methods are used to solve this problem nowadays. These methods can be categorized into two groups depending on whether correspondence is known：one is to estimate the Pose parameters if the correspondence has been solved; the other is to solve the Pose and correspondence problem simultaneously. A new method based on single camera for estimating the Pose parameters of the object is presented in this paper. Different from the existing methods, the correspondence problem which is coupled with Pose problem is completely avoided in the proposed approach. An energy function including the Pose parameters is defined to the degree of superposition between the model projection and the object area in the image. The optimal Pose parameters are found by optimizing the energy function. Hooke-Jeeves Algorithm is used to minimize the energy function. The performance of the method is evaluated in a variety of simulations on synthetic data. These.tests show that the method is accurate and effective.

关 键 词：单相机 姿态估计 能量函数 Hooke—Jeeves算法 

分 类 号：P231.5[天文地球—摄影测量与遥感]