在足球机器人视觉系统中,由于摄像机是普通的非量测摄像机,所以实际空间点的三维几何位置与其在图像中对应点之间是一种复杂的非线性映射关系,必须进行摄像机标定。由于基于精确模型的各种标定方法比较繁琐,而径向基函数网络可以实现从输入到输出的任意的非线性映射,论文将径向基函数网络应用于足球机器人视觉系统的标定,仿真实验结果表明该算法是确实可行的。 In the soccer robot vision system, because the camera is a common non-measurement camera, the three-dimensional geometric position of the actual space point and its corresponding point in the image is a complex non-linear mapping, and camera calibration must be performed. Since various calibration methods based on the exact model are cumbersome, and radial basis function networks can achieve arbitrary nonlinear mapping from input to output, the paper applies radial basis function network to the calibration of soccer robot vision system. The simulation results It shows that the algorithm is feasible.