基于附加噪声预测器无迹卡尔曼滤波的摄像机标定算法  

The Additional Noise Predictor UKF Based Camera Calibration Algorithm

作　　者：陈琳[1] 卢荣胜[1] 

机构地区：[1]合肥工业大学仪器科学与光电工程学院,合肥230009

出　　处：《科学技术与工程》2015年第2期124-130,共7页Science Technology and Engineering

基　　金：国家科技部科技支撑计划项目(2011BAK15B07)资助

摘　　要：由于环境等因素影响摄像机拍摄过程中存在可变噪声,针对传统无迹卡尔曼滤波(unscented Kalman filter,UKF)算法无法处理未知噪声的问题,设计了一种基于附加噪声预测器UKF的摄像机标定算法。首先,对传统UKF算法进行改进,引入噪声估计器并用极大后验估计求取次优解,解决未知可变情况下的噪声问题。然后,利用该改进UKF算法对摄像机进行标定,实现了标定精度的有效提高。实验结果显示该算法在有效保证滤波收敛性的同时,显著提高了滤波和摄像机的标定精度,由此推断基于附加噪声预测器UKF的摄像机标定算法是可行高效的。There are variable noises in camera shotting due to environmental factors. According to the problem of traditional unscented Kalman filtering （UKF） algorithm can not deal with unknown noise, an additional noise predictor UKF was designed based camera calibration algorithm. Firstly, the traditional UKF algorithm was im- proved by the noise estimator, and the maximum posteriori estimation was used to obtain suboptimal solutions to solve the problem of noise with unknown variable conditions. Then, the improved UKF algorithm was used to the camera calibration, which can effectively improve the accuracy of calibration. The results show that the algorithm could effectivly guarantee filtering convergence, at the same time, by which the camera calibration precision was significantly improves, thus infer camera noise predictor based UKF algorithm is feasible and efficient.

关 键 词：噪声 预测器 无迹卡尔曼滤波 摄像机标定 次优 

分 类 号：TP391.72[自动化与计算机技术—计算机应用技术]