论文部分内容阅读
摄像机标定是视觉定位控制中的重要步骤。针对现有标定方法存在的不足,采用最小二乘支持向量机和遗传算法,设计了摄像机标定方法。以系列标定点的图像亚像素坐标作为输入,世界坐标作为输出,建立最小二乘支持向量机的预测模型,表征世界坐标与图像像素坐标间的非线性函数关系。利用遗传算法在参数取值范围内,寻找预测模型正规化参数和RBF核函数参数的最优值,进而提高预测结果的速度和精度。实验结果表明,测试集中标定点的均方根误差,以及x和y坐标轴的均方根误差均较小。基于遗传算法和最小二乘支持向量机建立的摄像机模型预测能够以一定精度逼近世界坐标与图像像素坐标间复杂的非线性映射关系。
Camera calibration is an important step in visual positioning control. Aiming at the shortcomings of the existing calibration methods, the least squares support vector machine and genetic algorithm are used to design the camera calibration method. Taking the image subpixel coordinates of the series of calibration points as input and the world coordinates as output, a prediction model of least squares support vector machine is established to represent the nonlinear function relationship between world coordinates and pixel coordinates. Genetic algorithm is used to find out the optimal value of the normalization parameter and the RBF kernel function parameter in the parameter range, so as to improve the speed and accuracy of the prediction result. The experimental results show that the root mean square error of the calibration points in the test set and the root mean square error of the x and y axes are small. The camera model prediction based on genetic algorithm and least squares support vector machine can approach complex nonlinear mapping relationship between world coordinates and image pixel coordinates with a certain precision.