Abstract: A new method of linear movement camera calibration based on particle filter is proposed in order to achieve online dynamic calibration in the case that intelligent robot AS-R is moving. State-space method is used to describe the linear movement camera model. Taking the intrinsic camera parameters and location parameters as the state vector and taking the image coordinates of feature points as the observation vector, the optimized values of the intrinsic camera parameters and location parameters are obtained with particle filter. And the dual threads are used to achieve the whole calibration process. The experiment results of online dynamic calibration when robot AS-R is moving linearly show that this method is reasonable and feasible, with high accuracy and good robustness. In addition, this method is suitable for all types of system noise.

Key words: camera calibration, linear movement, particle filter, camera parameters

摘要： 为实现AS-R智能机器人在运动情况下摄像机在线动态标定，提出一种新的基于粒子滤波的直线运动摄像机标定方法。用状态空间方法描述直线运动摄像机模型，把摄像机内参数和位置运动参数作为状态量，特征点图像坐标作为观测量，根据粒子滤波算法求得摄像机内参数和位置运动参数的最优估计，并用双线程实现整个标定过程。AS-R机器人在直线运动情况下的摄像机在线动态标定实验结果表明：该算法是合理可行的，并且具有很高的标定精度和良好的鲁棒性。该方法适用于各种类型的系统噪声。

关键词: 摄像机标定, 直线运动, 粒子滤波, 摄像机参数