摘要
利用最小二乘支持向量机来直接学习图像信息与三维信息之间的关系,不需确定摄像机具体的内部参数和外部参数。在双目视觉的情况下,两摄像机的位置关系不需具体求出,而是隐含在映射关系中。根据最小二乘支持向量机与摄像机标定的特点,提出了基于最小二乘支持向量机的双目立体摄像机标定方法。将摄像头采集到的图像的像素坐标作为输入,将世界坐标作为输出,用最小二乘支持向量机使网络实现给定的输入输出映射关系。该方法同BP神经网络预测结果对比表明:基于最小二乘支持向量机的双目视觉标定方法速度快,实时性好,能有效提高标定精度。
Least Squares Support Vector Machines （LS-SVM） are used to learn the relationships between the image information and the 3D information, which doesn＇t need to confirm intemal and external parameters of the camera. In the case of binocular vision, the location relationship between the two cameras is implicated in the map reIafions. According to the LS-SVM and camera calibration characteristic, a new method of camera calibration for stereo vision based on the LS-SVM is proposed. The image pixel coordinate which gathers from the camera is made as input, the world coordinate is be used as output, and the given mapping relation of network realization is completed by the LS-SVM. The comparison with BP neural network shows that the calibration accuracy can be improved by using the LS-SVM. The real-time performance is good and response velocity is quick.
出处
《光电工程》
EI
CAS
CSCD
北大核心
2008年第10期21-25,47,共6页
Opto-Electronic Engineering
基金
黑龙江省自然科学基金(200419)资助项目
关键词
摄像机标定
最小二乘支持向量机
立体视觉
BP神经网络
camera calibration
least squares support vector machines
stereo vision
BP neural network
