摘要
首先研究探讨了基于绝对二次曲线 ( the absolute conic)进行摄像机自标定鲁棒性差的内在原因 .研究发现 ,该类方法鲁棒性不足的原因主要有三个方面 :1 )在目标函数的全局最小点处存在大范围的平坦区域 ,使得任何数值优化算法难以达到全局最小点 ;2 )当存在噪声时 ,上述平坦区域内会出现大量局部极小值 ,这样数值优化算法就非常容易收敛到靠近初值的局部极小值 ,使得算法对初始值的选取十分敏感 ;3)当有噪声时 ,目标函数的全局最小值极易偏离正确值 .这样 ,即使数值算法找到了全局最小值 ,该最小值也不再对应正确的摄像机内参数值 .鉴于上述情况 ,探讨了如何通过平面场景来确定内参数矩阵的初始值 ,而后进一步利用
It is well recognized that the IAC (Image of the Absolute Conic) based camera calibration techniques are not quite robust. We find that the following three sources largely contribute to the non-robustness of the IAC based techniques: 1) The global minimum of the cost function lies on a large flat area. 2) With noise, many local minimums could appear in the above flat area. 3) With noise, the point corresponding to the global minimum of the cost function could deviate significantly from the one sought for. In addition, we explore a two-step calibration technique. In this new technique, firstly a plane, which is distant from the origin, is used to obtain an initial solution, then Kruppa equations are used to refine this initial estimation. The experiments on simulated data as well as on real images validate our new technique.
出处
《自动化学报》
EI
CSCD
北大核心
2001年第5期621-630,共10页
Acta Automatica Sinica
基金
国家自然科学基金 ( 69975 0 2 1
60 0 75 0 0 4
60 0 330 1 0 )
国家"973"计划 ( G1 980 30 5 0 2 -3)
中国科学院机器人学开放实验室 ( RL2 0 0 0 1 0 )资助
关键词
KRUPPA方程
摄像机自标定
计算机视觉
数值算法
Calibration
Computer simulation
Functions
Image processing
Matrix algebra
Optimization
Robustness (control systems)
Spurious signal noise