| 基于谱聚类算法的三维激光点云数据分类研究 |
| |
| 引用本文： | 吴翔,王凤艳,林楠,王明常.基于谱聚类算法的三维激光点云数据分类研究[J].世界地质,2020,39(2):479-486. |
| |
| 作者姓名： | 吴翔 王凤艳 林楠 王明常 |
| |
| 作者单位： | 吉林大学 地球探测科学与技术学院, 长春130026;吉林建筑大学 测绘与勘查工程学院, 长春130018 |
| |
| 基金项目： | 国家自然科学基金;国家自然科学基金;国家重点实验室开放基金;国家重点实验室开放基金 |
| |
| 摘 要： | 基于Z+F IMAGER 5010C扫描仪采集实验区点云数据,经栅格处理后,结合纹理和形状等信息,采用谱聚类算法对其进行分类,利用混淆矩阵中的Kappa系数对分类结果进行精度评价。通过与传统的K-means算法和高斯混合模型的分类结果进行对比,结果表明:谱聚类算法的分类效果明显,且分类精度较高,且加入纹理和形状信息的分类精度会高于仅含反射强度信息的分类精度,其总体分类精度达到81.36%,Kappa系数达到0.713 8。
|
| 关 键 词： | 三维激光扫描 点云数据 谱聚类算法 Kappa系数 |
Research on classification of 3D laser point cloud data based on spectral clustering algorithm |
| |
| WU Xiang,WANG Feng-yan,LIN Nan,WANG Ming-chang.Research on classification of 3D laser point cloud data based on spectral clustering algorithm[J].World Geology,2020,39(2):479-486. |
| |
| Authors: | WU Xiang WANG Feng-yan LIN Nan WANG Ming-chang |
| |
| Institution: | (College of Geo-exploration Science and Technology,Jilin University,Changchun 130026,China;College of Surveying and Prospecting Engineering,Jilin Jianzhu University,Changchun 130118,China) |
| |
| Abstract: | After applying raster processing to the point cloud data collected by Z+F IMAGER 5010C scanner in the experimental area,combined with texture and shape information,the point cloud data are classified using the spectral clustering algorithm,and subsequently the accuracy of classification is evaluated by the Kappa coefficient in confusion matrix.Compared with the classification results of traditional K-means algorithm and Gaussian mixture model,the spectral clustering algorithm has more distinct classification effect and higher classification accuracy.Moreover,adding texture and shape information leads to higher accuracy than that only uses reflection intensity information.The overall accuracy of the classification is 81.36%,and the Kappa coefficient is 0.7138. |
| |
| Keywords: | 3D laser scanning point cloud data spectral clustering algorithm Kappa coefficient |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
