基于关键帧的视觉惯性定位算法

Keyframe-based Visual-inertial Localization Algorithm

针对视觉里程计定位精度易受环境因素干扰而降低的问题,本文采用视觉惯性定位方法。首先,利用惯性测量单元(IMU)在短时间内提供精确测量信息的特点,将视觉信息与IMU数据进行紧耦合处理,提高系统的定位精度。其次,为提高系统的实时性,将图像帧进行关键帧筛选,并提出一种关键帧选择方法。然后,采用滑动窗口融合优化模型,求解位姿估计得到相机的运动轨迹。最后,通过EuRo C数据集对本系统进行实验,评价了本文算法定位精度与时间效率,实验结果表明,与主流算法OKVIS相比本文算法在定位精度与实时性上均有所提高。

Aiming at the problem that the localization accuracy of visual odometery is easily affected by environmental factors,this paper adopts the method of visual-inertial localization.Firstly,the inertial measurement unit(IMU)is used to provide accurate measurement information in a short time,and the visual information are tightly coupled with the IMU data to improve the localization accuracy of the system.Secondly,in order to improve the real-time performance of the system,keyframe selection is performed on image frames,and a keyframe selection method is proposed.Then,a sliding window fusion optimizer is used to obtain the motion trajectory of the camera.Finally,the system is tested by EuRoC dataset,and the positioning accuracy and time efficiency of the proposed algorithm are evaluated.The experimental results show that compared with the mainstream algorithm OKVIS,the proposed algorithm improves the localization accuracy and real-time performance.