车载摄像机外参数的自动标定

Automatic Extrinsic Parameter Calibration for Camera-on-Vehicle

摄像机外参数自动标定的目的是通过自动的方法获得车载摄像机部分外参数。研究基于自主驾驶汽车的运动学模型、高速公路的道路模型以及车载摄像机的成像模型,进行了车载摄像机外参数的在线自动标定方法研究,并利用车体运动学模型作为扩展Kalman滤波(EKF)的状态模型,并将图像坐标系下与车体坐标系下道路参数间的非线性关系,作为EKF的观测模型进行扩展Kalman滤波。仿真与实车试验证明,方法实时检测了滤波结果的正确性,自动标定了最新的摄像机外参数,使得视觉导航系统能够自动适应摄像机姿态变化带来的影响。

The aim of camera extrinsic parameters calibration is to get part of extrinsic parameters of the vehicle - board camera by automatic means. Based on the kinematic model of autonomous vehicle, highway model as well as camera imaging model, an online automatic extrinsic parameter calibration method is studied. The kinematical model of the vehicle is used as Extended Kalman Filter（EKF） state model, at the same time, the nonlinear relationships between the road parameters in the image coordinate and the vehicle coordinate are utilized as the EKF observation model. The simulation and real experiments show that this algorithm checks the filter results and automatically calibrates the latest camera extrinsic parameters in real - time, so the navigation system will automatically adapt to the effects brought about by pose changing of the camera.