MAPSAC与EKF结合的单目相机鲁棒定位估计算法与实验验证

Robust Location Estimation Algorithm Design and Experimental Verification for Monocular Camera Combining MAPSAC and EKF

由于RANSAC数据关联存在随机选取样本、主模型先验假设、内点不做区分的问题,导致相机鲁棒定位估计效率不高、环境适应性受约束、准确性受到影响.本文结合EKF滤波器和MAPSAC数据关联方法,提出一种单目相机鲁棒定位估计算法.EKF滤波器预测和更新初步估计可提高定位估计效率,增强环境适应性;MAPSAC数据关联采用对内点加权的最小化损失函数剔除误关联数据,提高了单目相机定位估计的准确性.通过实验,验证了所设计单目相机鲁棒定位估计算法的有效性,研究结果为单目相机定位估计提供参考.

Because there are problems of the random sampling,the priori hypothesis of the main model and the failure to distinguish the interior points in RANSAC data association,the efficiency of robust location estimation is not high,the environment adaptability is restricted and the accuracy is affected in the camera.Based on EKF filter and MAPSAC data association method,we proposes a robust location estimation algorithm for monocular cameras.It is preliminarily estimated that EKF filter prediction and update will improves the efficiency of localization estimation,while enhancing environmental resilience.And MAPSAC data association uses the minimization loss function weighted for the inner point to reject the error association data,which improves the accuracy of monocular camera location estimation.The effectiveness of the proposed monocular camera robust location estimation algorithm is verified by experiments,and the research results provide reference for localization estimation of monocular camera.