基于分割区间LS-SVM的摄像机标定

利用最小二乘支持向量机（LS-SVM）可以不考虑摄像机具体的内部参数和外部参数实现摄像机的标定。由于镜头的畸变主要由径向畸变引起,根据摄像机畸变特点对畸变区域进行划分,提出一种基于分割区间LS-SVM的摄像机标定法,对不同的畸变区域进行单独处理。该方法与BP神经网络和基本LS-SVM预测结果对比表明,分割区间LS-SVM摄像机标定法误差小、速度快、标定精度高。

Camera Calibration Based on Divided Region LS-SVM

By using Least Squares Support Vector Machines(LS-SVM), it need not consider internal and external parameters to achieve the camera calibration. Because the lens distortion is mostly caused by radial distortion, according to the camera distortion characteristic, it divides the distortion region. A new method of camera calibration based on divided region LS-SVM is proposed, and the distortion of different regions deals with separately. The comparison with other methods including BP Neural Network(BPNN) and LS-SVM shows that the calibration accuracy is improved by using the divided LS-SVM method, and the speed is higher.