机器视觉导引系统中摄像机的标定方法

针对无人机(UAV)自主起飞与着陆机器视觉导引系统中摄像机的标定问题,讨论了标定摄像机内参数的重要意义,选定了包含畸变因素的摄像机透视投影模型。采用改进两步法进行摄像机内参数标定,结果中主要参数相对误差均小于1%,位置误差小于0.2像素,证明了所选择模型和标定方法的合理性。分析并验证了影响标定精度的因素,包括模型选择、图像拍摄质量和标定靶精度。针对机器视觉导引的应用条件,基本解决了准确、快速标定摄像机内参数的实际问题。

Camera Calibration Algorithms in Vision Based Navigation Systems

Calibration of intrinsic camera parameters is fundamental to vision based autonomous unmanned aerial vehicle(UAV) launching and landing.A pinhole camera perspective projection model is proposed.Both radial and tangential distortions are modeled.An improved two-step approach is examined by actual calibration experiments.Relative errors of parameters were all below 1%,and position accuracy is within 0.2 pixel.The proposed camera model and calibration method are appropriate for this application.Three major factors influence the accuracy of cali-bration process.They are camera model,precision of calibration pattern and quality of images.A method of practical camera calibration is realized.