一种全局指数收敛的摄像机内参数观测器

针对摄像机标定问题,本文从控制理论角度出发设计了一种具有指数收敛特性的摄像机内参数观测器.当摄像机随移动机器人等运动平台一起旋转时,该观测器使用其运动信息和实时拍摄得到的特征点图像信息,可以实现对摄像机内参数的在线估计.具体而言,论文分析了云台摄像机的运动约束,建立了特征点图像坐标变化的运动学模型,随后基于4个或4个以上特征点（其中任意3点不共线）,构造了一种内参数非线性观测器,并通过理论分析证明了其状态估计指数收敛于摄像机的相应内参数,仿真结果验证了这种非线性观测器的良好特性.

A global exponential observer for camera calibration

This paper presents a global exponential observer to solve the camera calibration problem on-line. While the camera is rotating with the mobile robot platform, the observer estimates online the intrinsic parameters of a camera from the motion information and the online information of the characteristic point image obtained by the camera. We analyze the motion constraints of the pan-tilt camera, and build the kinematics model for the coordinate variations of the characteristic point image. The nonlinear observer of the intrinsic parameters is built based on no less than four image points(any three of them are not collinear); the parameters of state-estimation are proved theoretically to be convergent to the intrinsic parameters of the camera exponentially. Simulation results demonstrate superior performance of the proposed nonlinear observer.