基于超体素点云路灯杆定位与提取的方法研究-计算机科学与技术专业论文.docxVIP

优秀毕业论文 精品参考文献资料 AbstractS Abstract S仃eet 1ights play an important role in transpoIrtation systems aus tlley proVide 1ight for pedestrians，Vehicles and road sun，eillaIlce canleLaS during nle night。，nley caIl significantly reduce the f．requency of nigMiIne crashes，especially for roadway inte】瞎ections．GeoSpatial infbmlation of s臼．eet light poles have been used in many 印plicationS including urball pl蛐g，enViro姗e删iInpact assessrnent，Intelligent TraIlSpo僦ion SysteIIls(ITS)，dis叙er m猢gement．F0r ITS related applications，me i11f0咖ation of me s仃eet l追ht poles can be used for road i111b随mcture maintailling， road saf．e够anaJySts，adV孤1ced“Ver assistaIlce，semantic mapping，aIld smart ci够 印plicationS．For iIlstance，in a driVer assistance system，me position of stI．eet light poles call be used to improVe tlle s讪ili母of road仃ackiIlg．There如re，regular survey and mainten锄ce of s骶et light poles is imponam．Due to此l鹕e number of street 1ight poles on road，心adition甜maIlual survey memod is e)(仃ernely time-consumiIlg．A r印id a11d rob呶m咖od is lligllly needed to obtain street light pole iIlfomation． Due to me uIlique advalltage of mobile lidar syStem，more a11d more scholars are at乜．acted onⅡle蚰ldy of ex仃acting street light poles in point clouds．HoweVer，due to me huge舢mbers of points and s骶et light poles occluded by trees or omer objects in point clouds，ex仃acting咖eet l迢ht poles疔om po缸clouds becomes challenging． Motivated by above problems，an autornated，e伍cient，觚d robust脒妯od of eXtracting street light poles in point clouds baSed on supervoxels is proposed． Our proposed memod consiSts of flVe steps：pre-processing， localization， segmentation，f．eature extraction，aIld classification．First，the raw point clouds are divided imo segments along tlle t喇ectory’t11e ground points are remoVed alld也e remailling pointS are segmented into supervoxels．Then，a robust local切tion memod is proposed to obtain pole—like objectS．NeXt，a 10c2Llization-guided segmentation method is proposed to obtajn pole·like objects．Subsequent