一种基于边界感知的遥感影像建筑物提取方法

doi:10.19665/j.issn1001-2400.2022.01.025

摘要/Abstract

摘要：

遥感影像的复杂性给建筑物提取研究带来了极大的挑战。深度学习的引入提高了遥感影像建筑物提取的准确率,但仍存在边界模糊、目标漏检和提取区域不完整等问题。针对这些问题,提出了一种基于边界感知的遥感影像建筑物提取网络,该网络包括特征融合网络、特征增强网络和特征细化网络三部分。首先,特征融合网络采用编码-解码结构提取不同尺度特征,并设计了交互聚合模块融合不同尺度的特征;然后,特征增强网络采用减法和级联操作对漏检目标进行学习增强,得到更加全面的特征;最后,特征细化网络使用编码-解码结构对特征增强网络的输出进一步细化,得到丰富的建筑物边界特征。此外,为使网络更加稳定有效,将二值交叉熵损失和结构相似性损失相结合,从像素和图像结构两个层次监督模型的训练学习。通过在数据集WHU上的测试,可知本网络较其他经典算法的客观指标交并比和准确率均有提升,分别达到了96.0%和97.9%;同时主观视觉上提取的建筑物边界更加清晰,区域更加完整分明。

关键词: 建筑物提取, 边界感知, 编码解码, 遥感影像, 深度学习

Abstract:

The complexity of remote sensing images brings a great challenge for building extraction research.The introduction of deep learning improves the accuracy of building extraction from remote sensing images,but there are still some problems such as blurred boundaries,missing targets and incomplete extraction areas.To address these issues,this paper proposes a boundary-aware network for building extraction from remote sensing images,including the feature fusion network,feature enhancement network and feature refinement network.First,the feature fusion network uses the encoding-decoding structure to extract different scale features,and designs the interactive aggregation module to fuse different scale features.Then,the feature enhancement network enhances the learning of missed targets through subtraction and cascade operation to obtain more comprehensive features.Finally,the feature refinement network further refines the output of the feature enhancement network by using the encoding-decoding structure to obtain rich building boundary features.In addition,in order to make the network more stable and effective,this paper combines the binary cross-entropy loss and the structure similarity loss,and supervises the training and learning of the model on both pixel and image structure levels.Through the test on the dataset WHU,in terms of objective metrics,the IoU and Precision of this network are improved compared with other classical algorithms,reaching 96.0% and 97.9% respectively.At the same time,in terms of subjective vision,the extracted building boundary is clearer and the region is more complete.

Key words: building extraction, boundary aware, encoding-decoding, remote sensing image, deep learning

中图分类号:

TP391

张艳,王翔宇,张众维,孙叶美,刘树东. 一种基于边界感知的遥感影像建筑物提取方法[J]. 西安电子科技大学学报, 2022, 49(1): 236-244.

ZHANG Yan,WANG Xiangyu,ZHANG Zhongwei,SUN Yemei,LIU Shudong. Boundary-aware network for building extraction from remote sensing images[J]. Journal of Xidian University, 2022, 49(1): 236-244.

图/表 9

图1

图2

图3

图4

表1

图5

表2

图6

图7

参考文献 28

[1]	YANG P, SONG W, ZHAO X B, et al. An Improved Otsu Threshold Segmentation Algorithm[J]. International Journal of Computational Science and Engineering, 2020, 22(1):146-153. doi: 10.1504/IJCSE.2020.107266
[2]	CHEN J S, HUERTAS A, MEDIONI G. Fast Convolution with Laplacian-of-Gaussian Masks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1987, 9 (4):584-590.
[3]	KANOPOULOS N, VASANTHAVADA N, BAKER R L. Design of an Image Edge Detection Filter Using the Sobel Operator[J]. IEEE Journal of Solid-State Circuits, 2002, 23(2):358-367. doi: 10.1109/4.996
[4]	黄亮, 姚丙秀, 陈朋弟, 等. 高分辨率遥感影像超像素的模糊聚类分割法[J]. 测绘学报, 2020, 49(5):589-597.
	HUANG Liang, YAO Bingxiu, CHEN Pengdi, et al. Superpixel Segmentation Method of High-Resolution Remote Sensing Image Based on Fuzzy Clustering[J]. Acta Geodaeticaet Cartographica Sinica, 2020, 49(5):589-597.
[5]	OK A O. Automated Detection of Buildings from Single VHR Multispectral Images Using Shadow Information and Graph Cuts[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2013, 86(12):21-40. doi: 10.1016/j.isprsjprs.2013.09.004
[6]	LI M, ZANG S, ZHANG B, et al. A Review of Remote Sensing Image Classification Techniques:The Role of Spatio-Contextual Information[J]. European Journal of Remote Sensing, 2014, 47(1):389-411. doi: 10.5721/EuJRS20144723
[7]	SIRMACEK B, UNSALAN C. Building Detection from Aerial Images Using Invariant Color Features and Shadow Information[C]// 2008 23rd International Symposium on Computer and Information Sciences.Piscataway:IEEE, 2008:1-5.
[8]	ZHONG S, HUANG J, XIE W. A New Method of Building Detection from a Single Aerial Photograph[C]// 2008 9th International Conference on Signal Processing.Piscataway:IEEE, 2008:1219-1222.
[9]	ZhANG Y. Optimisation of Building Detection inSsatellite Images by Combining Multispectral Classification and Texture Filtering[J]. ISPRS Journal of Photogrammetry and Rremote Sensing, 1999, 54(1):50-60.
[10]	FERRAIOLI G. Multichannel InSAR Building Edge Detection[J]. IEEE Transactions on Geoence & Remote Sensing, 2010, 48(3):1224-1231.
[11]	吕凤华, 舒宁, 龚龑, 等. 利用多特征进行航空影像建筑物提取[J]. 武汉大学学报:信息科学版, 2017, 42(5):656-660.
	LV Fenghua, SHU Ning, GONG Yan, et al. Regular Building Extraction from High Resolution Image Based on Multilevel-Features[J]. Geomatics and Information Science of Wuhan University, 2017, 42(5):656-660.
[12]	DUNAEVA A V, KORNILOV F A. Specific Shape Building Detection from Aerial Imagery in Infrared Range[J]. Vestn.YuUrGU.Ser.Vych.Matem.Inform., 2017, 6(3):84-100.
[13]	吴炜, 骆剑承, 沈占锋, 等. 光谱和形状特征相结合的高分辨率遥感图像的建筑物提取方法[J]. 武汉大学学报:信息科学版, 2012, 37(7):800-805.
	WU Wei, LUO Jiancheng, SHEN Zhanfeng, et al. Building Extraction from High Resolution Remote Sensing Imagery Based on Spatial-Spectral Method[J]. Geomatics and Information Science of Wuhan University, 2012, 37(7):800-805.
[14]	ZHONG C, XU Q, YANG F, et al. Building Change Detection for High-Resolution Remotely Sensed Images Based on a Semantic Dependency[C]// 2015 IEEE International Geoscience and Remote Sensing Symposium. Piscataway: IEEE, 2015:3345-3348.
[15]	崔艳鹏, 王元皓, 胡建伟. 一种改进YOLOv3的动态小目标检测方法[J]. 西安电子科技大学学报, 2020, 47(3):1-7.
	CUI Yanpeng, WANG Yuanhao, HU Jianwei. Detection Method for a Dynamic Small Target Using the Improved YOLOv3[J]. Journal of Xidian University, 2020, 47(3):1-7.
[16]	YANG X, YANG J, YAN J, et al. SCRDet:Towards More Robust Detection for Small, Cluttered and Rotated Objects[C]// 2019 IEEE/CVF International Conference on Computer Vision.Piscataway:IEEE, 2019:8231-8240.
[17]	TIAN Z, SHEN C, CHEN H, et al. FCOS:Fully Convolutional One-Stage Object Detection[C]// 2019 IEEE/CVF International Conference on Computer Vision.Piscataway:IEEE, 2019:9626-9635.
[18]	WANG G, WANG K, LIN L. Adaptively Connected Neural Networks[C]// 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition.Piscataway:IEEE, 2019:1781-1790.
[19]	BERTINETTO L, MUELLER R, TERTIKAS K, et al. Making Better Mistakes:Leveraging Class Hierarchies with Deep Networks[C]// 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition.Piscataway:IEEE, 2020:12503-12512.
[20]	ZORAN D, CHRZANOWSKI M, HUANG P S, et al. Towards Robust Image Classification Using Sequential Attention Models[C]// 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition.Piscataway:IEEE, 2020:9480-9489.
[21]	QIN X, ZHANG Z, HUANG C, et al. BASNet:Boundary-Aware Salient Object Detection[C]// 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition.Piscataway:IEEE, 2019:7471-7481.
[22]	CHEN Z, XU Q, CONG R, et al. Global Context-Aware Progressive Aggregation Network for Salient Object Detection[J]. Proceedings of the AAAI Conference on Artificial Intelligence, 2020, 34(7):10599-10606. doi: 10.1609/aaai.v34i07.6633
[23]	MAGGIORI E, TARABALKA Y, CHARPIAT G, et al. Fully Convolutional Neural Networks for Remote Sensing Image Classification[C]// 2016 IEEE International Geoscience and Remote Sensing Symposium.Piscataway:IEEE, 2016:5071-5074.
[24]	YUAN J. Learning Building Extraction in Aerial Scenes with Convolutional Networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 40(11):2793-2798. doi: 10.1109/TPAMI.2017.2750680
[25]	许慧敏. 基于深度学习U-Net模型的高分辨率遥感影像分类方法研究[D]. 成都: 西南交通大学, 2018.
[26]	WU G, SHAO X, GUO Z, et al. Automatic Building Segmentation of Aerial Imagery Using Multi-Constraint Fully Convolutional Networks[J]. Remote Sensing, 2018, 10(3):407. doi: 10.3390/rs10030407
[27]	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 Net Works from an Open Remote Sensing Building Dataset[J]. Acta Geodaeticaet Cartographica Sinica, 2019, 48(4):448-459.
[28]	张春森, 葛英伟, 蒋萧. 基于稀疏约束 SegNet 的高分辨率遥感影像建筑物提取[J]. 西安科技大学学报, 2020, 40(3):441-448.
	ZHANG Chunsen, GE Yingwei, JIANG Xiao. High-Resolution Remote Sensing Image Building Extraction Based on Sparsely Constrained SegNet[J]. Journal of Xi'an University of Science and Technology, 2020, 40(3):441-448.

模型	网络结构			损失函数		评价指标
模型	Baseline	IAM	FEN	L_BCE	L_SSIM	IoU	准确率
1	√			√	√	0.925	0.958
2	√	√		√	√	0.942	0.968
3	√	√	√	√		0.948	0.970
4	√	√	√	√	√	0.960	0.979

实验	方法	IoU	准确率	召回率
1	U-Net	0.843	0.870	0.964
2	SegNet	0.840	0.865	0.895
3	MC-FCN	0.854	0.880	0.967
4	SU-Net	0.888	0.952	0.930
5	LSPNet	0.864	0.882	0.971
6	BANet(文中方法)	0.960	0.979	0.945