基于scan context描述子的LVI-SAM算法改进研究  

Research on improvement of LVI-SAM algorithm based on scan context descriptor

作　　者：刘顺卫 何元烈[1] LIU Shunwei;HE Yuanlie(Guangdong University of Technology,Guangzhou 510000,China)

机构地区：[1]广东工业大学,广州510000

出　　处：《计算机应用文摘》2022年第8期68-70,共3页Chinese Journal of Computer Application

摘　　要：针对LVI-SAM算法中视觉惯性子系统容易因快速运动或明暗变化激烈导致算法回环检测无效的问题,文章设计了一种结合scan context描述子的回环检测系统的多传感器融合算法。首先,在原有的LVI-SAM算法的激光惯性子系统中增加scan context模块,形成scan context矩阵,用二进制矩阵来形成关键帧提取策略,以减少算法的计算冗余。同时,基于scan context描迷子解决在光照环境不佳情况下的回环检测问题。Aiming at the problem that the visual inertial subsystem in the LVI SAM algorithm is easy to make the loop closure detection of the algorithm invalid due to rapid motion or large illumination changes,a multi-sensor fusion algorithm of the loop closure detection system combined with the scan context descriptor is designed.First,a scan context module is added to the laser inertial subsystem of the original LVI SAM algorithm to form a scan context matrix,and a binary matrix is used to form a key frame extraction strategy to reduce the com putational redundancy of the algorithm.At the same time,based on the scan context descriptor,the loop closure detection prablem in the case of poor lighting environment is solved.

关 键 词：LVI-SAM scan context描述子 关键帧策略 回环检测 

分 类 号：TP31[自动化与计算机技术—计算机软件与理论]