摘要
特征点三维坐标的测量精度直接影响目标位姿的解算。相机安装校准过程中内部图像传感器产生的装调误差会导致实际成像点位置发生偏离,为消除定位偏差而最终提高三维测量精度,经过理论分析将装调误差分解为夹角误差、位移误差、旋转误差并分别建立不同的误差模型。首先将目标函数设置为成像点坐标差方程,利用多目标不等式的非线性最优化模型从理论上定量分析了装调误差,绘制各误差的三维分布图,定性分析了装调误差的发生趋势。选取某一型号相机进行物理实验,首先测定出主点坐标偏移量Δx=0.665 08 mm,Δy=0.936 68 mm,然后根据上述方法获得装调误差大小,得到相机偏移夹角为0.92°,偏移距离为1.41 pixel,偏移角度为0.047°,与所给真实值范围进行对比,验证了所提出标定方法的有效性。
The measurement accuracy of 3 D coordinate of feature points directly affects the calculation of target pose. In the process of camera installation calibration,the adjusting error caused by the internal image sensor will cause the actual imaging point position to deviate. In order to eliminate the adjusting errors and improve the 3 D measurement accuracy,the theoretical analysis is used to decompose the adjusting errors into the included angle error,the displacement error and the rotation error,and the error model is established respectively. Taking the coordinate difference equation of the imaging point as the objective function,the positioning error is quantitatively analyzed theoretically by using the nonlinear optimization model of multi-objective inequality. By drawing the 3 D distribution of each error,the trend of location error is qualitatively analyzed. A camera is selected for physical experiment. Firstly,the principal point coordinates offset Δx = 0. 665 08 mm,Δy = 0. 936 68 mm are determined. Then,the location deviation size is obtained according to the method described above. It obtains that the camera offset angle is equal to 0. 92°,the offset distance is equal to 1. 41 pixel,the offset rotation angle is equal to 0. 047°. And the validity of the proposed calibration method is verified by comparing the real values of the given camera parameters.
出处
《电子测量与仪器学报》
CSCD
北大核心
2018年第2期30-35,共6页
Journal of Electronic Measurement and Instrumentation
基金
国家自然科学基金(51305107)资助项目
关键词
图像传感器
装调误差
标定
坐标差方程
image sensor
adjusting error
calibration
coordinate difference equation
