基于SLAM定位的多位姿点云拼接与分割方法研究  

Research on splicing and segmentation method of multiple position and attitude point cloud based on SLAM localization

作　　者：王立鹏[1] 张佳鹏 张智[1] 孟浩[1] 肖绍桐 苏丽[1] WANG Lipeng;ZHANG Jiapeng;ZHANG Zhi;MENG Hao;XIAO Shaotong;SU Li(College of Intelligent Systems Science and Engineering,Harbin Engineering University,Harbin 150001,China;China Electronics 54th Institute,Shijiazhuang 050080,China)

机构地区：[1]哈尔滨工程大学智能科学与工程学院,黑龙江哈尔滨150001 [2]中国电子科技集团公司第五十四研究所,河北石家庄050080

出　　处：《实验技术与管理》2022年第4期39-44,50,共7页Experimental Technology and Management

基　　金：哈尔滨工程大学研究生教学改革项目“智能船舶产业与教学相结合的研究生实践能力培养模式探索”(JG2021Y050);哈尔滨工程大学本科教学改革项目“依托科研案例的机器人工程专业模块化综合设计实践模式探索”(JG2021B0409);哈尔滨工程大学高教研究课题“基于理论与实践互馈机制的机器人工程专业创新人才培养模式探索”(HEUGJYJ2021011);教育部高等教育司产学合作协同育人项目“产学合作协同育人模式下机器人工程专业实践条件建设研究”(201902155018);黑龙江省教育科学“十三五”规划专项课题“以学生为中心的人才培养模式研究”(GBE1317040)。

摘　　要：该文提出一种基于SLAM定位的多位置和姿态点云的拼接与分割方法,可提供信息量大且准确的点云样本数据。首先,利用RTAB-MAP框架构建SLAM算法,由KinectV2传感器采集获得带有位姿信息的多角度RGBD图像样本,并转化为三维点云场景数据;其次,改进ICP算法并基于SVD算法设计多位姿点云的拼接算法;再次,根据原始点云的RGB信息与凹凸性特征,设计三维点云分割算法,以此形成一系列点云目标;最后,开展点云拼接和点云分割实验,实验结果验证了该文所提算法的正确性和有效性。In this paper,a multiple position and attitude point cloud splicing and segmentation method based on SLAM localization is proposed,which can provides large amount of information and accurate point cloud sample data.First of all,SLAM algorithm is constructed by using RTAB-MAP framework,multi-angle RGBD image samples with position and attitude are collected by Kinect V2 sensor and transformed into three-dimensional point cloud scene data.Secondly,an algorithm of splicing multiple position and attitude point cloud is established,which uses improved ICP method and SVD algorithm.Thirdly,according to the RGB information and convexity characteristics of the original point cloud,3D point cloud segmentation is designed to form a series of point cloud targets.Finally,point cloud splicing and segmentation experiments are executed to verify the correctness and effectiveness of the proposed algorithm.

关 键 词：SLAM 定位 多角度 点云拼接 点云分割 

分 类 号：TP249[自动化与计算机技术—检测技术与自动化装置]