摄像机参数的三线性插值误差补偿标定方法

Camera Calibration Method of Tri-linear Interpolation Error Compensation

提出了一种摄像机标定方法并应用于某轿车翼子板的非接触检测。该方法通过虚拟靶标构建空间点阵,应用SUNSAN算子准确提取角点坐标,并代入摄像机模型,逆向求解空间点的坐标及与真实坐标的误差值,以此误差作为判断依据,通过迭代更新求解摄像机参数,直到平均误差小于某一预定值,然后基于空间三线性插值算法进行误差补偿,从而求解待测点的误差。实验证明此方法计算迅速、稳定、准确,检测精度达到亚像素级。

A camera calibration method was put forward and applied to non-contacting detection for the front fender of a car. The spatial lattice has been constructed by using virtual target and the corner coordinates were extracted with SUSAN operator. In order to acquire the camera parameters, the corner coordinates have been brought into the camera model and the coordinates and error of these points were obtained. An iterative algorithm was presented to correct all the parameters till the error within the permission. The error compensation was performed by using tri-linear interpolation. The experiment showed that the method is rapid and has high stability and accuracy, and its precision is up to subpixel.