摘要
为了求得摄像机的投影矩阵,在进行标定时,以采样点组合矢量坐标到由投影矩阵决定的超平面距离平方和为目标函数,设计有侧向连接的Sanger网络;采用自适应次分量的提取方法,以自相关矩阵最小特征值所对应的特征向量为超平面的拟合系数,据此求得摄像机的投影矩阵,完成摄像机的标定。根据摄像机的数学模型得到的投影点坐标与实际图像检测结果的图像残差均方值作为标定性能指标,进行精度分析。实验表明,所提出的方法是自适应Sanger算法在摄像机标定中的新应用。
In order to obtain camera′s projection matrix, while it is calibrated,the sum of square distances from vector coordinates combined with sample points to the hyperplane decided by projective matrix is taken as the objective function.Then a Sanger neural network with lateral connection is designed,where the self-adaptive minor component extracting method is adopted.And taking the eigenvector of minimum eigenvalues as fitted coefficients of the hyperplane,the projective matrix is obtained and the camera is calibrated.At the same time the mean square of image errors between projective point coordinates from camera model and the actual image results is taken as the performance index of camera calibration to analyze the precision.The proposed method is a novel application of the self-adaptive Sanger operator in camera calibration for experiment.
出处
《光电子.激光》
EI
CAS
CSCD
北大核心
2010年第11期1720-1724,共5页
Journal of Optoelectronics·Laser
基金
国家高技术研究发展计划资助项目(2007AA04Z111)
湖南省教育厅重点资助项目(07A062)
湖南省教育厅优秀青年资助项目(09B092)
湖南省自然科学基金资助项目(09JJ6092)