针对标准遗传算法在摄像机标定参数非线性优化过程中, 易出现过早收敛和停滞现象的问题, 提出一种新的摄像机标定参数优化方法。首先, 采用非线性的归一化几何排名函数与随机遍历抽样法混合作为选择方法, 对遗传算法进行改进; 然后, 采用改进的遗传算法对摄像机标定参数进行非线性优化; 最后, 将其与标准遗传算法进行标定参数非线性优化对比实验, 实验结果表明: 算法平均绝对误差低于标准遗传算法, 且图像主点坐标更接近参考值, 能较好地缓解过早收敛和停滞现象, 提高了标定精度。

For the problem of simple genetic algorithm prone to premature convergence and stagnation phenomena in the process of camera calibration parameters’ nonlinear optimization, a new camera calibration parameter optimization method was proposed. Firstly, the genetic algorithm was improved by using nonlinear normalized geometric ranking function and stochastic universal sampling mixed as the selection operator. Then, the improved genetic algorithm was adopted to do the nonlinear optimization for the camera calibration parameters. Finally, in the nonlinear optimization comparison experiment, the mean absolute error of the algorithm is lower than simple genetic algorithm, and the image coordinates of the principal point is closer to the reference value. It is concluded that the improved genetic algorithm can alleviate the premature convergence, and the stagnation phenomena and the calibration accuracy are both improved.