双目视觉测量系统的标定及3维测量

Calibration of Binocular Vision System and Its Application in 3D Measurement

针对结构光形位公差视觉测量系统,提出了一阶径向畸变的摄像机成像模型和标定方法,并对传统的直接线性变换(DLT变换)标定算法做了些改进。即先针对没有畸变的线性模型,利用传统的标定算法,解线性超定方程组求解摄像机的全部参数,再针对引入一阶径向畸变的成像模型,以线性模型的参数为初值,通过非线性迭代优化摄像机的图像中心,等效焦距,倾斜因子,畸变系数等内部参数。实验结果表明,该方法无需预标定,精度适中,是相对简单实用的标定方法。

A camera imaging model with radial distortion, used for a structured-light vision system to measure geometric tolerance, is introduced in this paper, together with an improved calibration of Direct Linear Transformation （ DLT）. First, with traditional calibration technique used on linear model, all the parameters can be worked out through solving super-linear equations; second, with the results of first step as initial value, the inner parameters, such as Center Of Image, Equivalent Foci, Slope Factor, Coefficient Of Distortion, can be worked out through non-linear iteration. Pre-calibration is not necessary in calibration with this method. The method performs with suitable accuracy, and is proved to be simple and practical.