结合植被光谱特征与Sep-...的城市植被信息智能提取方法_林娜.pdf

引用格式：林娜,何静,王斌,等.结合植被光谱特征与Sep-UNet的城市植被信息智能提取方法J.地球信息科学学报,2023,25(8):1717-1729.Lin N,He J,Wang B,et al.Intelligent extraction of urban vegetation information based on vegetation spectral signature and Sep-UNetJ.Journal of Geo-information Science,2023,25(8).1717-1729.DOI:10.12082/dqxxkx.2023.220866结合植被光谱特征与Sep-UNet的城市植被信息智能提取方法林娜1，何静1*，王斌2，唐菲菲1，周俊宇1，郭江11.重庆交通大学智慧城市学院，重庆 400074；2.重庆市地理信息与遥感应用中心，重庆 401147Intelligent Extraction of Urban Vegetation Information based on Vegetation SpectralSignature and Sep-UNetLIN Na1,HE Jing1*,WANG Bin2,TANG Feifei1,ZHOU Junyu1,GUO Jiang11.School of Smart City,Chongqing Jiaotong University,Chongqing 400074,China;2.Chongqing Geomatics and Remote Sensing Center,Chongqing 401147,ChinaAbstract:Urban vegetation is an important component of urban ecosystems and plays a vital role in humansettlements,urban ecology,urban planning,and sustainable development.It is urgent to develop an efficient andaccurate method to achieve the intelligent extraction of urban vegetation.In view of the problems of lowefficiency and strong human intervention in the extraction of urban vegetation by traditional methods,andinsufficient utilization of spectral information in deep learning methods,this study focused on intelligentextraction of urban vegetation from GF-1D high resolution remote sensing images by combining the mostimportant spectral reflection characteristics of vegetation in the Near Infrared(NIR)band and the Sep U-Net,anoptimization of U-Net.The main work of this research includes:(1)we created three sample sets considering thehigh reflectance of vegetation in the NIR band:the true-color green space sample set(true sample set),whichserved as the control group,the standard false-color green space sample set(the fake sample set),and the false-color green space sample set synthesized by NDVI(NDVI sample set);(2)the Sep-UNet was optimized basedon U-Net.On the basis of U-Net,Sep-UNet expanded the network receptive field and increased the networkdepth by increasing the number of concatenated convolutions to achieve the purpose of enhancing the networksinformation processing of vegetation details.The separable convolution was used to reduce the networkparameters while maintaining the underlying characteristics and depth of the network,avoiding thecomputational burden caused by network deepening.The batch normalization layer and Tanh activation function收稿日期：2022-11-07；修回日期：2023-03-28.基金项目：国家重点研发计划项目（2021YFB2600600、21YFB2600603）；教育部产学合作协同育人项目（220702313111054）；重庆市技术创新与应用发展专项重点项目（CSTB2022TIAD-CUX0016）；宁夏自治区重点研发计划（2022CMG02014）。Foundation items:National Key Research and Development Program of China,No.2021YFB2600600,2021YFB2600603;Industry-UniversityCooperationEducationProgramofMinistryofEducation,No.220702313111054;ChongqingTechnologyInnovationandApplicationDevelopment(KeyIndustryResearchandDe-velopment),No.CSTB2022TIAD-CUX0016;KeyR&DProgramofNingxiaAutonomousRegion,No.2022CMG02014.作者简介：林娜（1981），女，湖北襄阳人，博士，副教授，主要从事遥感影像智能处理等方面的研究。E-mail:*通讯作者：何静（1997），女，重庆人，硕士生，研究方向为遥感影像智能处理。E-mail:Vol.25,No.8Aug.,2023第25卷 第8期2023年8月地 球 信 息 科 学 学 报2023年were also used to enhance the operability and robustness of the network;(3)we extracted urban vegetation frommultiple scene images.Four typical urban scenes were selected from remote sensing images outside the sampleset,and models were applied to extract urban vegetation from scene images with typical urban characteristics toverify the applicability and transferability of the models.The results show that:(1)The inclusion of NIR bandsignificantly enhanced the accuracy of urban vegetation extraction.The extraction results using standard falsecolor sample set were the best,followed by NDVI sample set,and the extraction results using true color sampleset were the worst;(2)Based on the validation sample set,the urban vegetation extraction accuracy using Sep-UNet(ACC:0.9576,IOU:0.8938,Recall:0.9549)was better than that using U-Net(ACC:0.9389,IOU:0.7593,Recall:0.9405),and much better than that using SegNet(ACC:0.8897,IOU:0.8019,Recall:0.8867),whichdemonstrated that the proposed model in this paper obtained best results for extraction of urban vegetation;(3)The extraction results from multiple scene images based on Sep-UNet showed that the model can also extracturban vegetation well from images beyond the sampling area,and the inclusion of NIR band can eliminate theinterference of thin clouds,mist,and buildings,making the model have good generalizability.Key words:high resolution remote sensing image;deep learning;the separable convolution;Sep-UNet;urbanvegetation information;near infrared;extraction*Corresponding author:HE Jing,E-mail:摘要：城市植被在城市环境中起着举足轻重的作用，高效、准确的城市被信息提取是目前亟需解决的任务之一。针对研究中U-Net网络存在的深度较浅使得植被细节特征遗漏，小数量植被样本造成网络拟合表现不佳以及大量参数导致的网络运算负担过重等问题，本文在U-Net网络的基础上，利用可分离卷积、批量标准化层、Tanh激活函数，设计提出了一种Sep-UNet网络。基于GF-1D高分遥感影像，结合Sep-UNet与植被在近红外波段上最具标志性的光谱反射特性，进行了高分遥感影像城市植被信息提取的研究。结果表明：标准假彩色样本集的植被提取模型精度最佳，NDVI样本集次之，真彩色波段的最差；Sep-UNet在验证样本集中提取城市植被信息的各项精度指标(ACC：0.9576，IOU：0.8938，Recall：0.9549)