基于全参数自适应变异粒子群算法的单目相机标定

Monocular camera calibration based on particle swarm algorithm with all parameter adaptive mutation mechanism

针对传统粒子群方法求解相机内参时的局部最优解问题,提出一种基于全参数自适应调节和变异机制的粒子群单目相机内参优化方法.首先,基于向量约束关系对单应性矩阵进行变形,利用最小二乘法求得相机的初始内参.然后,考虑在迭代过程中局部最优粒子、全局最优粒子对各个粒子的作用不同,分别给出了基于粒距的自适应的局部因子学习调节策略和全局因子学习调节策略;同时,设计了基于粒子群平均粒距的改进的粒子自适应变异率.最后,给出了基于全参数的自适应变异机制的粒子群相机内参优化算法.实验结果表明,与张正友标定方法、传统粒子群优化标定方法相比,该方法具有较好的标定精度和收敛速度.

To solve the problem of a local optimal solution in traditional particle swarm camera calibration,an internal parameter optimizing method for the camera calibration was proposed based on a particle swarm algorithm with all parameter adaptive regulation and mutation mechanism. Firstly,by the modification of the homography matrix based on vector restrictions,the initial value of intrinsic parameter was obtained with the least square method. Secondly,according to different influences of the local optimal particle and global optimal particle on each individual particle in the process of the iterations,the adaptive regulation strategies of the local learning factor and the global learning factor based on the particle distances were used,respectively. Meanwhile,an adaptive mutation rate based on the average particle distance was designed. Finally,the particle swarm algorithm with all parameter adaptive mutation mechanism for the monocular camera calibration was given. Experimental results demonstrate that,compared with Zhang Zhengyou's calibration method and the traditional particle swarm calibration method,the proposed method has better calibration accuracies and convergence speeds.