李雪,张力,王庆栋,石壮,牛雨

• 1:中国测绘科学研究院
• 2:山东建筑大学

摘要(Abstract)：

针对图像密集匹配生产的数字表面模型(DSM)进行点云滤波,算法对地形依赖大,参数设置复杂,精度不高,后续人工编辑修饰的工作量大、效率低的问题,该文设计了第一套针对DSM滤波、涵盖多种样本形式(栅格、矢量)的航空图像建筑物数据集。针对航空图像建筑物尺度较大等特点,将膨胀卷积加入U-Net构成Dilated U-Net,并综合运用其进行建筑物语义分割,利用分割结果在相应图像密集匹配得到的DSM上滤除建筑物点,然后采用投票插值策略得到过滤掉建筑物点的DSM。实验证明:利用该文网络DU-Net将DSM中非地面建筑物点滤除,Ⅰ类误差在5.8%以内,Ⅱ类误差在2.4%以内,其可以在30 s内完成超过9 000万个建筑点与非建筑物点位置的预测,效率高、成本低。DU-Net网络建筑物语义分割过程不受地形、高差的限制,对于其他非地面点的滤波具有一定的借鉴意义。

关键词(KeyWords)： 深度学习;卷积神经网络;建筑物语义分割;数字表面模型

Abstract:

Keywords:

基金项目(Foundation): 国家重点研发计划课题项目（2019YFB1405602）

作者(Author): 李雪,张力,王庆栋,石壮,牛雨

DOI: 10.16251/j.cnki.1009-2307.2021.02.013

参考文献(References)：

• [1] 柯长青,冯学智,刘刚.基于ERDAS软件的敦煌壁画正射影像的制作[J].测绘科学,2004,29(1):57-60.(KE Changqing,FENG Xuezhi,LIU Gang.ERDAS-based orthoimage making of mural paintings of Dunhuang[J].Science of Surveying and Mapping,2004,29(1):57-60.)
• [2] 叶玮,雷邦俊,周扬.地理国情普查工作中车载移动测量系统更新DEM方法的探讨[J].测绘通报,2013(8):58-60.(YE Wei,LEI Bangjun,ZHOU Yang.On DEM updating method for census of national geographic conditions using vehicle-borne mobile measurement system[J].Bulletin of Surveying and Mapping,2013(8):58-60.)
• [3] 杨玉冰,秦晓莉.地理国情普查DEM精细化生产技术及应用浅析[J].国土资源导刊,2015,12(4):85-89.(YANG Yubing,QIN Xiaoli.Analysis of DEM fine production technology and its application in geographical conditions census[J].Land & Resources Herald,2015,12(4):85-89.)
• [4] 周亮,万昀昕,孙畅.基于地形智能识别的自适应DSM滤波方法研究[J].测绘与空间地理信息,2016,39(7):18-20.(ZHOU Liang,WAN Yunxin,SUN Chang.An adaptive filtering algorithm of DSM based on smart terrain identification[J].Geomatics & Spatial Information Technology,2016,39 (7):18-20.)
• [5] 曾妮红,岳迎春,魏占营,等.车载 LiDAR 点云滤波的改进不规则三角网加密方法[J].测绘科学,2016,41(9):136-139.(ZENG Nihong,YUE Yingchun,WEI Zhanying,et al.An improved irregular triangular network encryption method for vehicle-borne LiDAR point cloud clouds[J].Science of Surveying and Mapping,2016,41(9):136-139.)
• [6] 杨晓云.一种基于形态学的激光雷达数据滤波算法的改进[J].测绘科学,2011,36(4):166-168.(YANG Xiaoyun.Improved irregular triangular network encryption method for on-board LiDAR point cloud filtering[J].Science of Surveying and Mapping,2011,36(4):166-168.)
• [7] LU W L,MURPHY K P,LITTLE J J,et al.A hybrid conditional random field for estimating the underlying ground surface from airborne LiDAR data[J].IEEE Transactions on Geoscience and Remote Sensing,2009,47(8):2913-2922.
• [8] 刘经南,许晓东,张小红,等.机载激光扫描测高数据分层迭代选权滤波方法及其质量评价[J].武汉大学学报(信息科学版),2008,33(6):551-555.(LIU Jingnan,XU Xiaodong,ZHANG Xiaohong,et al.Adaptive hierarchical and weighted iterative filtering of airborne LiDAR data and its quality assessment[J].Geomatics and Information Science of Wunan University,2008,33(6):551-555.)
• [9] WACK R,WIMMER A.Digital terrain models from airborne laser scanner data:a grid based approach[J].International Archives of Photogrammetry Remote Sensing and Spatial Information Sciences,2002,34(3/B):293-296.
• [10] KAMPFFMEYER M,SALBERG A B,JENSSEN R.Semantic segmentation of small objects and modeling of uncertainty in urban remote sensing images using deep convolutional neural networks[C]//IEEE Computer Society Computer Vision and Pattern Recognition Workshops.Las Vegas:IEEE,2016:1-9.
• [11] LI K,CHENG G,BU S,et al.Rotation-insensitive and context-augmented object detection in remote sensing images[J].IEEE Transactions on Geoscience and Remote Sensing,2017,56(4):2337-2348.
• [12] 赵玲娜,花向红,李万,等.基于深度图像的建筑物点云平面边界线提取算法[J].测绘地理信息,2017,42(3):48-52.(ZHAO Lingna,HUA Xianghong,LI Wan,et al.Plane boundary extraction algorithm of buildings point clouds based on depth images[J].Journal of Geomatics,2017,42(3):48-52.)
• [13] SUN Y,ZHANG X,ZHAO X,et al.Extracting building boundaries from high resolution optical images and LiDAR data by integrating the convolutional neural network and the active contour model[J].Remote Sensing,2018,10(9):1459.
• [14] DUARTE D,NEX F,KERLE N,et al.Multi-resolution feature fusion for image classification of building damages with convolutional neural networks[J].Remote Sensing,2018,10(10):1636.
• [15] BADRINARAYANAN V,KENDALL A,CIPOLLA R.U-Net:a deep convolutional encoder-decoder architecture for image segmentation[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2017,39(12):2481-2495.
• [16] HU J,SHEN L,SUN G.Squeeze-and-excitation networks[C]//IEEE Computer Society Computer Vision and Pattern Recognition Workshops.Salt Lake City:IEEE,2018:7132-7141.
• [17] HE K,GKIOXARI G,DOLLáR P,et al.Mask R-CNN[C]∥IEEE Computer Society International Conference on Computer Vision Workshops.Venice:IEEE,2017:2961-2969.
• [18] KRESO I,SEGVIC S,KRAPAC J.Ladder-style DenseNets for semantic segmentation of large natural images[C]//IEEE Computer Society International Conference on Computer Vision Workshops.Venice:IEEE,2017:238-245.
• [19] 何海清,游琦,陈晓勇.卷积神经网络支持下的低空摄影测量DEM修补[J].遥感信息,2017,32(1):115-119.(HE Haiqing,YOU Qi,CHEN Xiaoyong.Convolutional neural network supported automatic DEM correction in low-altitude photogrammetry[J].Remote Sensing Information,2017,32(1):115-119.)
• [20] LONG J,SHELHAMER E,DARRELL T.Fully convolutional networks for semantic segmentation[C]∥ IEEE Computer Society Computer Vision and Pattern Recognition Workshops.Boston:IEEE,2015:3431-3440.
• [21] 季顺平,魏世清.遥感影像建筑物提取的卷积神经元网络与开源数据集方法[J].测绘学报,2019,48(4):448-459.(JI Shunping,WEI Shiqing.Building extraction via convolutional neural networks from an open remote sensing building dataset[J].Acta Geodaetica et Cartographica Sinica,2019,48(4):448-459.)
• [22] 白少琦.基于Unet无线网络规划的GIS的研究与应用[D].北京:北京邮电大学,2015.(BAI Shaoqi.Research and application on the GIS based on the wireless network planning software Unet[D].Beijing:Beijing University of Posts and Telecommunications,2015.)
• [23] YANG Z,HU Z,SALAKHUTDINOV R,et al.Improved variational autoencoders for text modeling using dilated convolutions[C]//JMLR Computer Society International Conference on Machine Learning Workshops.Sydney:JMLR,2017:3881-3890.
• [24] MAGGIORI E,TARABALKA Y,CHARPIAT G,et al.Can semantic labeling methods generalize to any city?The Inria aerial image labeling benchmark[C]//IEEE Computer Society International Geoscience and Remote Sensing Symposium.Fort Worth:IEEE,2017:3226-3229.