| 177 | 0 | 155 |
| 下载次数 | 被引频次 | 阅读次数 |
当前深度学习技术极大提升了遥感数据的自动化处理能力，针对DSM到DEM生产过程中降高区域提取环节，采用深度学习语义分割的U-Net模型实现了降高区域的自动提取，构建了面向DEM生产的样本分类系统，形成了规范化样本标注技术方法和优化后的DEM制作技术流程，并在DEM生产实践中检验了该方法的实用性。结果表明，在地表景观层次分明、地物可辨性高的场景下，能得到较好的降高区域提取结果，分类精度可达0.952。相较于传统人工勾绘或逐图幅监督分类的降高区域提取方法，深度学习智能解译辅助下的DEM生产效率可整体提高20%～30%，且能确保区域尺度DEM产品的协调一致性，具有重要的实用价值。
Abstract：At present, deep learning technology has greatly improved the automatic processing ability of remote sensing data. For the extraction of height reduction area in the production process from DSM to DEM, we adopted U-Net model of deep learning semantic segmentation to realize the automatic extraction of height reduction area. At the same time, we constructed a sample classification system for DEM production,formed a standardized sample labeling technology method and DEM production technology flow after optimization, and tested the practicability of this method in the DEM production practice. The result shows that this method can get better classification results in the scene with clear surface landscape and high ground object identifiability, and the classification accuracy can reach 0.952. Compared with the traditional height reduction area extraction method of manual sketching or map by map supervised classification, the overall DEM production efficiency assisted by deep learning intelligent interpretation can be improved by 20%～30%, and ensure the coordination and consistency of regional scale DEM products,which has important practical value.
[1]童善群.数字高程模型的野外采样密度与精度[J].解放军测绘学院学报，1990(1):63-68
[2]胡海燕.资源三号三线阵影像立体匹配与DSM生成研究[D].北京:北京建筑大学，2016
[3]卜丽静，寇可心.基于资源三号数据的三维变化检测可行性研究[J].测绘通报，2013(12):31-35
[4]周国新，唐建波，雷丽珍，等.基于机载点云数据的高精度水域DEM快速生成方法[J].地理信息世界，2019,26(5):119-122
[5]王铁军，朱岩丽，陶源，等.利用机载InSAR生产1∶50 000DEM的试验分析[J].地理信息世界，2012,19(6):77-80
[6]田海波，陈利军，张宏伟，等.全球4D数据产品研制[J].测绘通报，2020(7):1-4
[7]李德仁.我国第一颗民用三线阵立体测图卫星:资源三号测绘卫星[J].测绘学报，2012,41(3):317-322
[8]唐新明，王鸿燕，祝小勇.资源三号卫星测绘技术与应用[J].测绘学报，2017,46(10):1 482-1 491
[9] ZHANG L,ZHANG L,DU B. Deep Learning for Remote Sensing Data:A Technical Tutorial on the State of the Art[J]. IEEE Geoscience and Remote Sensing Magazine,2016,4(2):22-40
[10]吴永静，吴锦超，林超，等.基于深度学习的高分辨率遥感影像光伏用地提取[J].测绘通报，2021(5):96-101
[11] LEI S. How Training Data Affect the Accuracy and Robustness of Neural Networks for Image Classification[C].ICLR Conference,2019
[12] Atharva S,LIU X W,YANG X J. A Patch-based Convolutional Neural Network for Remote Sensing Image Classification[J]. Neural Networks,2017,95:19-28
[13] Long J,Shelhamer E,Darrell T. Fully Convolutional Networks for Semantic Segmentation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2014,39(4):640-651
[14] Ronneberger O,Fischer P,Brox T. U-Net:Convolutional Networks for Biomedical Image Segmentation[C]. International Conference on Medical Image Computing and Computer-assisted Intervention,2015:234-241
[15] Badrinarayanan V,Kendall A,Cipolla R. Segnet:A Deep Convolutional Encoder-decoder Architecture for Image Segmentation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2017,39(12):2 481-2 495
[16] CHEN L C,Papandreou G,Kokkinos I,et al. Semantic Image Segmentation with Deep Convolutional Nets and Fully Connected CRFs[J]. Computer Science,2014(4):357-361
[17]范荣双，陈洋，徐启恒，等.基于深度学习的高分辨率遥感影像建筑物提取方法[J].测绘学报，2019,48(1):34-41
[18]刘建歌，白穆，王馨爽.基于地理国情的深度学习分类容错性研究[J].地理空间信息，2022,20(2):9-14
[19] ZHANG Z,LIU Q,WANG Y. Road Extraction by Deep Residual U-Net[J]. IEEE Geoscience and Remote Sensing Letters,2018,15(5):749-753
[20] WANG Y H,CHEN E X,LI Z Y,et al. Deep U-Net Optimization Method for Forest Type Classification with High Resolution Multispectral Remote Sensing Images[J]. Forest Research,2020,33(1):11-18
[21]吕道双，林娜，冯丽蓉，等.基于改进型U-Net网络的高分辨率遥感影像建筑物提取[J].地理空间信息，2021,19(1):30-34
[22]赵英时.遥感应用分析原理与方法[M].北京:科学出版社，2003:203-208
基本信息:
中图分类号:TP18;P208
引用信息:
[1]王馨爽,刘建歌,李桢,等.深度学习智能解译支持下的DEM生成方法[J].地理空间信息,2023,21(07):32-36.
基金信息:
国家自然科学基金资助项目（41971352）
2023-07-26
2023-07-26