基于车载激光扫描点云的交通标志检测研究-计算机科学与技术专业论文.docxVIP

万方数据 万方数据 基于车载激光扫描点云的交通标志检测研究 AbstractAbstract Abstract Abstract As an important transport inf．rastlllcture， tramc signs can proVide guide iIlf．onIlation for嘶vers and pedestrians，stalldardize the behavior of nle血Ver and ensure safb哆at the s锄e time．As pan of intelligent trmsportation，road仃amc sigIl detection has iInportant si卿ficaIlce for impmVing仃amc safb够V．ellicle-bome laser scall血ng system has the abil“y to collect point clouds of road aIld tellraill on botll sides of me ground quickly，including the real geo铲apllic space iIlfomation aIld ech0 renection ene唱y，wllich call renect tlle propeny of material su“犯e．Therefore， Mobile laser scanniIlg(MLS)is a pro商sing tecllIlology for r印id 3D mapping of road． In仳s thesis，a road仃amc si盟detection a11d shape_based classification method based on laser scaIlIling point clouds is presented．The main work is as follows (1)Laser scanIliIlg poiIl_t clouds data pr印rocessiIlg，includiIlg noise filtering， establiShment of me spatial data iIldex锄d grouIld filtering，proVide emcient data s旬nlcture aIld search index for ou茁algorimm (2)An automated memod to detect road仃amc sigll in outdoor enV沛衄ents using mobile light detection and raIlging (LiDAR) aIld inenial naVigation tecImologies is presented．First，仃amc signs are detected based on the intensity at仃ibutes，as tramc si印血ensity has sigIliflcant di丘．erence wim omer objects in point clouds．Second，a仃amc si弘detection memod based on shape feature is印plied， w11ich recognizes pole．1ike objects(PLOs)丘．om raw point clouds and extracted by P血cipal Componem Analysis．Fillally’廿le t、)l，o detection menlods are combined to f．om tlle fmal detection result．The proposed algoritll】：Il caIl mal【e up the defects of detection algoritllm based on reflectance imensitywllich could miss some cases when the仃a伍c intens时c籼ot be distinguished舶m other targets in point clouds，and perfonn mutual aumentication of t11e觚o metllods． (3)Tramc si印s are classified accordmg to their sh印e，wllich is circul码 rectaIlgIllar and t