基于改进遗传算法的立体视觉系统标定

立体视觉系统的摄像机标定是一个多参数、非线性的复杂函数优化问题,传统优化方法很难解决。论文对标准遗传算法的编码方式进行了改进,经过改进后的遗传算法具有变量搜寻区间的自适应调整能力,在保持染色体编码长度不变的情况下,能同时满足变量搜索空间大小和编码精度的要求。利用改进了的遗传算法对双目视觉系统摄像机进行标定的结果表明,该算法能有效地实现高维寻优空间的近优解搜索。

Camera Calibration of Stereo Vision System Based on Genetic Algorithms

The camera calibration of stereo vision system is a complex nonlinear multi-parameter function optimization problem.It is very difficult to solve this problem for conventional optimization techniques.In this paper,the encoding method of standard genetic algorithm is modified and the improved genetic algorithm has the ability to adjust parameter search interval adaptively.The method of adaptive adjustment of parameter search interval not only provides enough search space,but also ensures search resolution without altering the length of the chromosome string.The experiments on the camera calibration of the binocular vision system indicate that the modified genetic algorithm can find the near optimum in the optimization spaces of high dimension effectively.