基于HQB-SHOT特征描述符门座起重机特征匹配研究  

Research on feature matching of portal crane based on HQB-SHOT feature descriptor

作　　者：徐承军[1] 李佩泽 吉雨岢 Xu Chengjun;Li Peize;Ji Yuke

机构地区：[1]武汉理工大学交通与物流工程学院,武汉430000 [2]武汉理工大学海南研究院,三亚572000

出　　处：《起重运输机械》2024年第22期57-62,73,共7页Hoisting and Conveying Machinery

摘　　要：为推动港口件杂货码头的自动化发展和减少激光雷达在门座起重机特征匹配过程中的计算开销,文中从SHOT特征描述符的构建过程和实际应用方面进行改进,提出了半球化的区间双编码位SHOT特征描述符:将局部参考坐标系获得的球形空间选取包含点多的一半球形为邻域,然后沿着仰角、方位角和径向线将球形邻域划分,统计每个小块中邻居点的法线和特征点的法线间的角度形成直方图,最后把直方图组合形成特征,在此基础上对所得到的描述子进行QB-SHOT二进制编码,最后得到半球化的区间双编码SHOT特征描述子。在采集到的门座起重机点云数据集进行实验,实验结果表明,HQB-SHOT在算法速度和匹配精度上均有所提高,实验验证了HQB-SHOT在港口门座起重机上应用的可行性和有效性。To promote the automation development of port general cargo terminal and reduce the calculation cost of lidar in the feature matching process of portal crane,the construction process and practical application of SHOT feature descriptor are improved.A hemispherical interval binary coded SHOT feature descriptor(HQB-SHOT) is proposed.The spherical space obtained by local reference coordinate system was selected as the neighborhood,and then the spherical neighborhood was divided along elevation,azimuth and radial lines,and the angles between the normals of neighbor points and the normals of feature points in each block were counted to form a histogram.Finally,the histograms were combined to form features,and on this basis,the obtained descriptors were coded by QB-SHOT binary,and the hemispherical interval double-coded SHOT feature descriptors were obtained.The collected point cloud data set of portal crane was tested,and the results show that the algorithm speed and matching accuracy of HQB-SHOT were improved,which verifies the feasibility and effectiveness of HQB-SHOT in port portal crane application.

关 键 词：SHOT 特征提取 特征匹配 门座起重机 点云 

分 类 号：TH213.4[机械工程—机械制造及自动化]