一种室内无标志动态目标视觉定位方法

An Indoor Unmarked Dynamic Target Visual Positioning Method

提出了一种室内无标志动态目标的视觉定位方式,采用卡尔曼滤波估计动态目标图像坐标,并利用分区定位方式对多相机视场内的动态目标进行定位,根据观测到动态目标的相机数目将多相机视场范围区域划分为单目区域、双目区域和多目区域,对不同的区域采用不同的定位方式。仿真和真实实验表明,本文算法可以达到10 cm左右的定位精度,运行效率可以达到50 ms/帧。提出的无标志动态目标定位方法具有更高的鲁棒性,基本可以满足实时性要求且定位精度较高。

A visual positioning method for indoor unmarked dynamic targets is proposed in this paper.Kalman filter is used to estimate the dynamic target image coordinates,and the dynamic target in the multi-camera field of view is located by using the partition positioning method.According to the number of cameras that observe the dynamic target,the field of view of multi-camera is divided into monocular area,binocular area,and multi-view area,and uses different positioning methods are used in different areas.Simulation and real experiments show that the positioning accuracy of the proposed algorithm can achieve about 10 cm,and the running efficiency of that can reach 50 ms/frame.The unmarked dynamic target localization method proposed in this paper has higher robustness,and can basically meet the real-time requirements with high positioning accuracy.