首页 | 本学科首页   官方微博 | 高级检索  
     

基于自旋图的三维自动目标识别
引用本文:刘瑶,马杰,赵季,田金文,熊凌.基于自旋图的三维自动目标识别[J].红外与激光工程,2012,41(2):543-548.
作者姓名:刘瑶  马杰  赵季  田金文  熊凌
作者单位:1. 华中科技大学图像识别与人工智能研究所多谱信息处理技术国家级重点实验室,湖北武汉,430074
2. 武汉科技大学信息科学与工程学院,湖北武汉,430081
基金项目:国家自然科学基金(61074156,61104191)
摘    要:将三维表面匹配领域的自旋图(spin-images)方法运用于扫描激光雷达自动目标识别中,基于自旋图特征,提出了平均最大相似性度量、候选对应关系数目、几何一致性筛选后对应关系数目、分组筛选后最大组的对应关系数目这4个识别指标,并将其组合构成组合识别准则,从而进行三维目标识别。同时针对实际应用需求,进一步研究了点云空间分辨率、激光雷达测距误差以及遮挡对目标识别率的影响,为激光雷达三维目标识别系统的设计提供了参考。

关 键 词:三维目标识别  自旋图  点云  表面重建

Three dimensional automatic target recognition based on spin-images
Liu Yao , Ma Jie , Zhao Ji , Tian Jinwen , Xiong Ling.Three dimensional automatic target recognition based on spin-images[J].Infrared and Laser Engineering,2012,41(2):543-548.
Authors:Liu Yao  Ma Jie  Zhao Ji  Tian Jinwen  Xiong Ling
Affiliation:1.National Key Laboratory of Science and Technology on Multi-spectral Information Processing,Institute for Pattern Recognition and Artificial Intelligence,Huazhong University of Science and Technology,Wuhan 430074,China; 2.College of Information Science and Engineering,Wuhan University of Science and Technology,Wuhan 430081,China)
Abstract:Scanning LIDAR automatic target recognition based on spin-images method was proposed which was used in three dimensional surface match area.The four criterions were proposed,including the average maximum similarity measure,the number of candidate correspondences,the number of correspondences after geometric consistency filtering,the correspondences number of maximum group after group filtering correspondence.The combined four criterions were constituted a combined recognition criterion for target recognition.In view of the actual application requirements,the influences of point cloud spatial resolution,LIDAR ranging noise and occlusion on target recognition rate were further studied.The study in this paper can provide reference for parameter design of the three dimensional target recognition system.
Keywords:3-D target recognition  spin-images  point clouds  surface reconstruction
本文献已被 CNKI 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号