一种基于正交直线构造的摄像机畸变标定方法  

Camera distortion calibration method based on orthogonal lines

作　　者：徐嵩[1] 孙秀霞[1] 刘树光[1] 刘希[1] 杨朋松[1] 

机构地区：[1]空军工程大学航空航天工程学院,西安710038

出　　处：《仪器仪表学报》2013年第9期2118-2123,共6页Chinese Journal of Scientific Instrument

基　　金：航空科学基金(20121396008);国家级大学生创新训练项目(201390052011)资助

摘　　要：提出了一种不需要利用共线特征点就可以消除非线性迭代优化过程中参数耦合问题的畸变参数标定方法,只需一幅图像即可获得高精度标定结果。首先设计了可避免非线性优化不稳定的畸变校正模型,然后直接利用4点求得的单应性矩阵并与另一点构建平面内正交线段,并结合对偶圆环点构造了正交代价函数。通过Nelder-Mead非线性单纯型法对正交代价函数中的畸变参数进行寻优,即可完成畸变参数的标定。仿真和实际图像实验表明了该方法的有效性。A new method of camera distortion calibration is proposed, which can eliminate the parameter coupling in the nonlinear iterative optimization process without utilizing collinear mark points, and obtain high precision calibra- tion result only with one image. In this method, a camera distortion correction model is designed to avoid the unstable phenomena in nonlinear optimization. Then, the homography matrix solved with four points and another point are used directly to construct the orthogonal lines in a plane. An orthogonal cost function is created with these lines and the dual circular points. The distortion parameters in the orthogonal cost function are optimized with Nelder-Mead nonlin- ear simplex method, and the distortion parameter calibration is achieved. The simulation and experiment on real ima- ges prove the effectiveness of the proposed method.

关 键 词：相机标定 畸变校正 正交 单应矩阵 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]