双目并行辅助匹配视觉SLAM系统算法研究  

Research on the Algorithm of Binocular Parallel-Assisted Matching Visual SLAM System

作　　者：符强[1,2,3,4] 梁燚东 纪元法 任风华[1,2] FU Qiang;LIANG Yi-dong;JI Yuan-fa;REN Feng-hua(Guangxi Key Laboratory of Precision Navigation Technology and Application,Guilin University of Electronic Science and Technology,Guilin Guangxi 541004,China;School of Information and Communication,Guilin University of Electronic Science and Technology,Guilin Guangxi 541004,China;National and Local Joint Engineering Research Center for Satellite Navigation Positioning and Service,Guilin Guangxi 541004,China;Nanning Guidian Electronic Technology Research Institute Co.,LTD.,Nanning Guangxi 530031,China)

机构地区：[1]桂林电子科技大学广西精密导航技术与应用重点实验室,广西桂林541004 [2]桂林电子科技大学信息与通信学院,广西桂林541004 [3]卫星导航定位与位置服务国家地方联合工程研究中心,广西桂林541004 [4]南宁桂电电子科技研究院有限公司,广西南宁530031

出　　处：《计算机仿真》2025年第3期386-389,共4页Computer Simulation

基　　金：广西科技厅项目(桂科AA20302022,桂科AB21196041,桂科AB22035074,桂科AD22080061);国家自然科学基金(62061010,62161007);广西八桂学者团队项目(桂林市科技项目)(20210222-1);广西高校中青年教师科研基础能力提升项目(2022KY0181);广西精密导航技术与应用重点实验室(DH202216);桂林电子科技大学研究生教育创新计划资助项目(2022YCXS050)。

摘　　要：介绍了一种并行辅助匹配SLAM方法,解决了SLAM系统中常见的跟踪精度不足以及在复杂环境下丢失跟踪的问题。对图像做光流跟踪估计运动得到当前帧的关键点。同时根据上一帧的关键点样本数量计算迭代参数,利用Prosac迭代计算单应矩阵,去除外点做第一次优化。获得了新的单应矩阵后通过透视变换更新关键点。匹配关键点的描述子以及对应的地图点更新状态变量。做第二次优化图优化去除外点计算当前相机位姿。标记外点并将优化后的关键点标记在图像中。增加了场景中的地图点与关键点以此来达到增强系统鲁棒性的目的,同时对于系统的精度也有较大的提升。An approach of parallel-assisted matching SLAM is introduced.This method addresses the common issues in SLAM systems,such as insufficient tracking accuracy and tracking loss in complex environments.First,optical flow tracking is performed on the images to estimate the motion and obtain the key points of the current frame.At the same time,the iteration parameters are calculated according to the number of key point samples in the previous frame,and the homography matrix is computed iteratively by Prosac,and the first optimization is done by eliminating the exception points.After obtaining a new homography matrix,the key points are updated by perspective transformation.Match the descriptor of the key point and the corresponding map point to update the state variable.Do a second optimization map optimization to calculate the current camera pose by excluding points.Mark the outer points and mark the optimized key points in the image.In this paper,map points and key points in the scene are added to achieve the purpose of enhancing the robustness of the system,and the accuracy of the system is also greatly improved.

关 键 词：同步定位与建图 单应矩阵 图优化 光流法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]