基于OPA的时空约束单目视觉位姿估计算法

Temporal Constraint Pose Estimation Algorithm Based on OPA in Monocular Vision Systems

针对单目视觉系统中载体纯旋转运动引起的估计退化问题,结合运动参数约束条件,提出了一种基于OPA的时空约束单目视觉位姿估计方法。该方法利用OPA算法在估计本质矩阵和平移方向之前判断载体的运动形式,检测修正估计退化问题,并结合摄像机在运动过程中的时间相关约束及图像特征点在空间中的相关约束,实现单目视觉系统的位姿估计。两组对比实验结果分析表明,新算法相比于传统的估计算法具有更高的精度和稳定性。

Considering the estimation degradation caused by the pure rotational movement in the monocular vision system and the motion parameter constraints, we proposed an OPA-based spatial-time constraint pose estimation algorithm for monocular vision systems. The OPA algorithm was used to detect and correct the estimation degradation before estimating the essential matrix and translational direction vector. Combined with the time-related constraints of camera in motion process and the space-related constraints of image feature point, the estimation to navigation parameters of the monocular vision system were obtained.Comparative analysis of two sets of experimental results shows that the proposed algorithm has higher accuracy and stability compared with the conventional estimation algorithm.