基于Feature-RNet的三维大场景点云分类框架  

Cloud Classification Framework of 3D Large-scale Scene Based on Feature-RNet

作　　者：雷根华 王蕾[1,2] 张志勇 LEI Gen-hua;WANG Lei;ZHANG Zhi-yong(School of Information Engineering,East China University of Technology,Nanchang 330013,China;Jiangxi Engineering Technology Research Center of Nuclear Geoscience Data Science and System,Nanchang 330013,China)

机构地区：[1]东华理工大学信息工程学院,江西南昌330013 [2]江西省核地学数据科学与系统工程技术研究中心,江西南昌330013

出　　处：《计算机技术与发展》2022年第6期85-91,共7页Computer Technology and Development

基　　金：国家自然科学基金资助项目(61561003,61761003);江西省核地学数据科学与系统工程技术研究中心基金(JETRCNGDSS201902)。

摘　　要：随着大场景三维点云应用在越来越多的领域中,近些年对激光点云大场景下的分类研究不断深入,各种分类模型层出不穷,在大场景点云分类任务中表现优异,但是依然存在训练时间长、计算复杂以及分类精度低等问题。针对分类精度低这一问题,提出一种基于Feature-RNet的三维大场景点云分类框架模型来实现点云分类工作。该框架模型不直接以三维点云数据作为输入,而是以通过KNN方法提取三维点云的二维特征和三维特征构建的特征图像作为输入,避免了网络框架对三维点云数据直接处理的不适应性;该模型设计的RNet框架结构利用了残差模块,并对其进行变型以达到提高分类精度的效果。采用公开的Oakland三维数据集对Feature-RNet框架模型进行训练,与现有的其他深度学习分类框架相比,提出的Feature-RNet框架模型在分类精度上有较大的提升,总体分类准确率能达到97.7%。With the application of large scene 3 D point cloud in more and more fields,the classification research of laser point cloud in large scene has been deepened in recent years,and various classification models emerge in endlessly and perform well in the large-scale scenic spot cloud classification task,but there are still problems such as long training time,complex calculation and low classification accuracy.Aiming at such problem of low classification accuracy,a three-dimensional scenic spot cloud classification framework model based on Feature-RNet is proposed to realize point cloud classification.The framework model does not directly take the three-dimensional point cloud data as the input,but takes the two-dimensional features of the three-dimensional point cloud extracted by KNN method and the feature image constructed by the three-dimensional features as the input,which avoids the inadaptability of the network framework to the direct processing of the three-dimensional point cloud data.The framework structure of RNet designed by the model uses the residual module and modifies it to improve the classification accuracy.The open Oakland 3 D data set is used to train the Feature-RNet framework model.Compared with other existing deep learning classification frameworks,the proposed Feature-RNet framework model has a great improvement in classification accuracy,and the overall classification accuracy can reach 97.7%.

关 键 词：点云特征图像 RNet网络框架 大场景点云分类 Oakland数据集 深度学习 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]