基于高光谱影像和LiDAR点云的城市地物分类实验方案设计  

A Novel Experimental Scheme for Urban Ground Object Classification Based on Hyperspectral Image and LiDAR Point Cloud Data

作　　者：王斌[1] 宋冬梅[1] 戴嵩 王海起[1] WANG Bin;SONG Dongmei;DAI Song;WANG Haiqi(College of Oceanography and Space Informatics,China University of Petroleum(East China),Qingdao 266580,Shandong,China)

机构地区：[1]中国石油大学(华东)海洋与空间信息学院,山东青岛266580

出　　处：《实验室研究与探索》2023年第2期5-10,151,共7页Research and Exploration In Laboratory

基　　金：国家自然科学基金-联合基金项目-重点支持项目(U1906217);山东省重点研发计划项目(2019GGX101033);山东省教学改革面上项目(M2021155);中国石油大学(华东)青年教师教学改革项目(QN201801);研究性教学方法改革项目(YK201802);专业建设与改革类项目(ZY202025)。

摘　　要：为更好地进行数据融合实现地物分类,设计了一种基于卷积神经网络(CNN)的高光谱与激光雷达数据融合分类实验方案。首先通过构建CNN模型实现高光谱及激光雷达数据的特征提取;然后分别在特征级和决策级上实现两种数据的融合,其中特征级融合采取求和与最大化策略,决策级融合采取加权融合策略;最后在休斯顿与特伦托数据集上使用标准训练集与测试集验证该模型的有效性。结果表明,相较于SVM和ELM两种对比方案,此方案总体分类精度分别提升了10.18%和8.75%。证实了新方案能够实现更高精度的城市地物分类。With the purpose of better completing ground object classification,a novel classification scheme has been proposed based on convolutional neural network(CNN)with a fusion of hyperspectral and LiDAR data.Firstly,feature extraction of hyperspectral and LiDAR data is achieved by constructed CNN model.Then,the fusion of the two kinds of data is implemented at the feature level and the decision level,respectively.The summation and maximization strategy are adopted in the feature level fusion,and the weighted fusion strategy is used for decision-level fusion.Finally,the validity of the model is verified by the experiment on the standard training set and test set on Houston and Trento data.The results show that compared with SVM and ELM,the overall classification accuracy of this scheme is improved by 1018%and 875%,respectively,It is confirmed that the new scheme can obtain higher precision urban feature classification.

关 键 词：地物分类 高光谱图像 激光雷达 卷积神经网络 特征融合 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]