基于改进的LSD线特征提取算法研究  

Research on improved LSD line feature extraction algorithm

作　　者：谭帅奇 孙皓[2] 李健铭 Tan Shuaiqi;Sun Hao;Li Jianming(School of Electronic Information Engineering,Chongqing Technology and Business Institute,Chongqing 401520,China;School of Management,Chongqing University of Technology,Chongqing 400054,China)

机构地区：[1]重庆工商职业学院电子信息工程学院,重庆401520 [2]重庆理工大学管理学院,重庆400054

出　　处：《现代计算机》2024年第6期69-72,共4页Modern Computer

基　　金：重庆开放大学重庆工商职业学院2023年度科研项目(NDYB2023-02)。

摘　　要：基于LSD直线段检测算法可能导致直线提取碎片化的问题,提出了改进的LSD线特征检测算法,根据角度和距离对直线进行筛选,解决了直线提取碎片化问题,采用LBD(line band discriptor)描述相邻帧图像之间的线特征相似度,提高了视觉SLAM位姿优化的计算效率,在公开数据集上对LSD算法与改进后的算法进行对比实验。结果表明,改进后的LSD在匹配精度上提升了27.8%。In view of the problem that the LSD straight line segment detection algorithm may lead to the fragmentation of straight line extraction,proposes an improved LSD line feature detection algorithm,which filters straight lines based on angles and distances,solves the problem of fragmentation of straight line extraction,and uses LBD descriptors to describe The similarity of line features between adjacent frame images improves the computational efficiency of visual SLAM pose optimization.A comparative ex-periment was conducted between the LSD algorithm and improved algorithm on a public data set.The results show that the im-proved LSD improves matching accuracy by 27.8%.

关 键 词：视觉SLAM LSD-SLAM 线特征 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]