基于多尺度特征感知网络的城市植被无人机遥感分类

doi:10.12082/dqxxkx.2022.210572

地球信息科学学报 ›› 2022, Vol. 24 ›› Issue (5): 962-980.doi: 10.12082/dqxxkx.2022.210572

基于多尺度特征感知网络的城市植被无人机遥感分类

蒯宇¹(), 王彪¹^,^*(), 吴艳兰¹^,², 陈搏涛¹, 陈兴迪¹, 薛维宝¹

1.安徽大学资源与环境工程学院,合肥 230601
2.安徽省地理信息智能技术工程研究中心,合肥 230000

收稿日期:2021-09-23 修回日期:2021-11-09 出版日期:2022-05-25 发布日期:2022-07-25
通讯作者: * 王彪（1987—）,男,山东曲阜人,副教授,主要从事摄影测量与遥感技术研究。E-mail: wangbiao-rs@ahu.edu.cn
作者简介:蒯宇（1995—）,男,安徽合肥人,硕士生,主要从事深度学习遥感影像信息提取。E-mail: kuaiyu1020@163.com
基金资助:
国家自然科学基金项目(41971311);国家自然科学基金项目(41902182);安徽省自然科学基金(2008085QD188)

Urban Vegetation Classification based on Multi-scale Feature Perception Network for UAV Images

KUAI Yu¹(), WANG Biao¹^,^*(), WU Yanglan¹^,², CHEN Botao¹, CHEN Xingdi¹, XUE Weibao¹

1. School of Resources and Environmental Engineering, Anhui University, Hefei 230601, China
2. Anhui Province Geographic Information Intelligent Technology Engineering Research Center, Hefei 230000, China

Received:2021-09-23 Revised:2021-11-09 Online:2022-05-25 Published:2022-07-25
Contact: WANG Biao
Supported by:
National Natural Science Foundation of China(41971311);National Natural Science Foundation of China(41902182);Natural Science Foundation of Anhui Province(2008085QD188)

摘要/Abstract

摘要：

目前城市植被分类受特征相近、光谱相似影响导致植被漏分、错分。因此,设计了一种多尺度特征感知网络（MFDN）结合高分辨率无人机可见光影像对城市植被分类。该网络针对漏分、错分问题,通过在输入层引入坐标卷积减少空间信息的丢失;构建并行网络增强多尺度特征信息并在网络之间引入重复多尺度融合模块使整个过程保持高分辨率表示,减少细节特征的丢失;同时添加分离特征模块扩大感受野,获取多尺度特征,从而有效缓解了城市植被错分、漏分现象。结果表明,MFDN方法在仅使用无人机可见光影像条件下主要是通过空间模式而不是光谱信息促进了城市植被分类,平均总体精度为89.54%,平均F1得分为75.85%,平均IOU为65.45%,分割结果准确完整。因此,所提方法与易于操作的低成本无人机系统相匹配,适用于城市植被快速调查,可以为城市空间利用和生态资源调查提供技术支持和科学依据。

关键词: 城市植被分类, 深度学习, 无人机遥感, 可见光影像, 多尺度特征融合, 语义分割, 多特征感知, 城市生态系统

Abstract:

At present, the classification of urban vegetation is affected by similar characteristics and similar spectra, resulting in misclassification of vegetation. Therefore, a Multi-scale Feature Perception Network (MFDN) combined with high-resolution UAV visible light images is designed to classify urban vegetation. This network addresses the problem of misclassification and reduces the loss of spatial information by introducing coordinate convolution in the input layer. It constructs parallel networks to enhance multi-scale feature information and introduces repeated multi-scale fusion modules between networks to maintain high-resolution representation in the entire process and reduce the loss of detailed features. In addition, the separation feature module is added to expand the receptive field and obtain multi-scale features, thereby effectively alleviating the phenomenon of misclassification and omission of urban vegetation. The results show that the MFDN method improves the classification of urban vegetation mainly through spatial patterns rather than spectral information from UAV visible light images. The average overall accuracy is 89.54%, the average F1 score is 75.85%, and the average IOU is 65.45%. The segmentation results are accurate and complete. Therefore, the proposed method is compatible with the easy-to-operate low-cost UAV system, is suitable for rapid survey of urban vegetation, and can provide technical support and scientific basis for urban space utilization and ecological resource survey.

Key words: urban vegetation classification, deep learning, UAV remote sensing, visible light image, multi-scale feature fusion, semantic segmentation, multi-feature perception, urban ecosystem

蒯宇, 王彪, 吴艳兰, 陈搏涛, 陈兴迪, 薛维宝. 基于多尺度特征感知网络的城市植被无人机遥感分类[J]. 地球信息科学学报, 2022, 24(5): 962-980.DOI:10.12082/dqxxkx.2022.210572

KUAI Yu, WANG Biao, WU Yanglan, CHEN Botao, CHEN Xingdi, XUE Weibao. Urban Vegetation Classification based on Multi-scale Feature Perception Network for UAV Images[J]. Journal of Geo-information Science, 2022, 24(5): 962-980.DOI:10.12082/dqxxkx.2022.210572

图/表 28

图1

图2

图3

图4

图5

图6

图7

表1

图8

图9

图10

图11

表2

表3

图12

表4

表5

表6

图13

图14

图15

表7

表8

图16

图17

图18

表9

表10

参考文献 47

[1]	李莹, 于海洋, 王燕, 等. 基于无人机重建点云与影像的城市植被分类[J]. 国土资源遥感, 2019, 31(1):152-158.
	[ Li Y, Yu H Y, Wang Y, et al. Classification of urban vegetation based on unmanned aerial vehicle reconstruction point cloud and image[J]. Remote Sensing for Land and Resources, 2019, 31(1):152-158. ] DOI: 10.6046/gtzyyg.2019.01.20 doi: 10.6046/gtzyyg.2019.01.20
[2]	刘保生. 基于高分辨率IKONOS影像的城市植被信息提取方法浅析[J]. 测绘通报, 2016(S1):182-184,187.
	[ Liu B S. Analysis based on high resolution IKONOS images urban vegetation information extraction method[J]. Bulletin of Surveying and Mapping, 2016(S1):182-184,187. ] DOI: 10.13474/j.cnki.11-2246.2016.0653 doi: 10.13474/j.cnki.11-2246.2016.0653
[3]	汪雪淼, 孟庆岩, 赵少华, 等. GF-2在城市绿地分类及景观格局度量中的应用[J]. 地球信息科学学报, 2020, 22(10):1971-1982. doi: 10.12082/dqxxkx.2020.200122
	[ Wang X M, Meng Q Y, Zhao S H, et al. Urban green space classification and landscape pattern measurement based on GF-2 image[J]. Journal of Geo-information Science, 2020, 22(10):1971-1982. ] DOI: 10.12082/dqxxkx.2020.200122 doi: 10.12082/dqxxkx.2020.200122
[4]	浮媛媛, 赵云升, 赵文利, 等. 基于多源亮度温度的城市典型植被分类研究[J]. 激光与光电子学进展, 2015, 52(7):267-272.
	[ Fu Y Y, Zhao Y S, Zhao W L, et al. Studies of typical urban vegetation classification based on brightness temperature from multiple sources[J]. Laser and Optoelectronics Progress, 2015, 52(7):267-272. ] DOI: 10.3788/LOP52.072801 doi: 10.3788/LOP52.072801
[5]	Feng Q, Liu J, Gong J. UAV remote sensing for urban vegetation mapping using random forest and texture analysis[J]. Remote Sensing, 2015, 7(1):1074-1094. DOI: 10.3390/rs70101074 doi: 10.3390/rs70101074
[6]	皮新宇, 曾永年, 贺城墙. 融合多源遥感数据的高分辨率城市植被覆盖度估算[J]. 遥感学报, 2021, 25(6):1216-1226.
	[ Pi X Y, Zeng Y N, He C Q, et al. High-resolution urban vegetation coverage estimation based on multi-source remote sensing data fusion[J]. National Remote Sensing Bulletin, 2021, 25(6):1216-1226. ] DOI: 10.11834/jrs.20219178 doi: 10.11834/jrs.20219178
[7]	王美雅, 徐涵秋. 中国大城市的城市组成对城市热岛强度的影响研究[J]. 地球信息科学学报, 2018, 20(12):1787-1798. doi: 10.12082/dqxxkx.2018.180257
	[ Wang M Y, Xu H Q. Analyzing the influence of urban forms on surface urban heat islands intensity in Chinese mega cities[J]. Journal of Geo-information Science, 2018, 20(12):1787-1798. ] DOI: 10.12082/dqxxkx.2018.180257 doi: 10.12082/dqxxkx.2018.180257
[8]	姚方方, 骆剑承, 沈占锋, 等. 高分辨率影像城市植被自动提取算法[J]. 地球信息科学学报, 2016, 18(2):248-254. doi: 10.3724/SP.J.1047.2016.00248
	[ Yao F F, Luo J C, Shen Z F, et al. Automatic urban vegetation extraction method using high resolution imagery[J]. Journal of Geo-information Science, 2016, 18(2):248-254. ] DOI: 10.3724/SP.J.1047.2016.00248 doi: 10.3724/SP.J.1047.2016.00248
[9]	Wang X, Wang Y, Zhou C, et al. Urban forest monitoring based on multiple features at the single tree scale by UAV[J]. Urban Forestry and Urban Greening, 2020, 58:126958. DOI: 10.1016/j.ufug.2020.126958 doi: 10.1016/j.ufug.2020.126958
[10]	Hassaan O, Nasir A K, Roth H, et al. Precision Forestry: Trees counting in urban areas using visible imagery based on an Unmanned Aerial Vehicle[J]. IFAC PapersOnLine, 2016, 49(16):16-21. DOI: 10.1016/j.ifacol.2016.10.004 doi: 10.1016/j.ifacol.2016.10.004
[11]	赵云景, 龚绪才, 杜文俊, 等. 基于无人机图像颜色指数的植被识别[J]. 国土资源遥感, 2016, 28(1):78-86.
	[ Zhao Y J, Gong X C, Du W J, et al. Vegetation extraction method based on color indices from UAV images[J]. Remote Sensing for Land and Resources, 2016, 28(1):78-86. ] DOI: 10.6046/gtzyyg.2016.01.12 doi: 10.6046/gtzyyg.2016.01.12
[12]	林志玮, 涂伟豪, 黄嘉航, 等. 基于FC-DenseNet的低空航拍光学图像树种识别[J]. 国土资源遥感, 2019, 31(3):225-233.
	[ Lin Z W, Tu W H, Huang J H, et al. Tree species recognition of UAV aerial images based on FC-DenseNet[J]. Remote Sensing for Land and Resources, 2019, 31(3):225-233. ] DOI: 10.6046/gtzyyg.2019.03.28 doi: 10.6046/gtzyyg.2019.03.28
[13]	Gaparovi M, Di Dobrini. Comparative assessment of machine learning methods for urban vegetation mapping using multitemporal Sentinel-1 imagery[J]. Remote Sensing, 2020, 12(12):1952. DOI: 10.3390/rs12121952 doi: 10.3390/rs12121952
[14]	Rumora L, Miler M, Medak D. Impact of various atmospheric corrections on Sentinel-2 land cover classification accuracy using machine learning classifiers[J]. International Journal of Geo-Information, 2020, 9(4):277. DOI: 10.3390/ijgi9040277 doi: 10.3390/ijgi9040277
[15]	戴鹏钦, 丁丽霞, 刘丽娟, 等. 基于FCN的无人机可见光影像树种分类[J]. 激光与光电子学进展, 2020, 57(10):101001.
	[ Dai P Q, Ding L X, Liu L J, et al. Tree species identification based on FCN using the visible images obtained from an Unmanned Aerial Vehicle[J]. Laser and Optoelectronics Progress, 2020, 57(10):101001. ] DOI: 10.3788/lop57.101001 doi: 10.3788/lop57.101001
[16]	Chen S, Mcdermid G, Castilla G, et al. Measuring vegetation height in linear disturbances in the boreal forest with UAV photogrammetry[J]. Remote Sensing, 2017, 9(12):1257. DOI: 10.3390/rs9121257 doi: 10.3390/rs9121257
[17]	Tian J, Le W, Li X, et al. Comparison of UAV and WorldView-2 imagery for mapping leaf area index of mangrove forest[J]. International Journal of Applied Earth Observation and Geoinformation, 2017, 61:22-31. DOI: 10.1016/j.jag.2017.05.002 doi: 10.1016/j.jag.2017.05.002
[18]	Van L W, Straatsma M, Addink E, et al. Monitoring height and greenness of non-woody floodplain vegetation with UAV time series[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2018, 141(jul.):112-123. DOI: 10.1016/j.isprsjprs.2018.04.011 doi: 10.1016/j.isprsjprs.2018.04.011
[19]	耿仁方, 付波霖, 蔡江涛, 等. 基于无人机影像和面向对象随机森林算法的岩溶湿地植被识别方法研究[J]. 地球信息科学学报, 2019, 21(8):1295-1306. doi: 10.12082/dqxxkx.2019.180631
	[ Geng R F, Fu B L, Cai J T, et al. Object-based karst wetland vegetation classification method using unmanned aerial vehicle images and random forest algorithm[J]. Journal of Geo-information Science, 2019, 21(8):1295-1306. ] DOI: 10.12082/dqxxkx.2019.180631 doi: 10.12082/dqxxkx.2019.180631
[20]	周欣昕, 吴艳兰, 李梦雅, 等. 基于特征分离机制的深度学习植被自动提取方法[J]. 地球信息科学学报, 2021, 23(9):1675-1689. doi: 10.12082/dqxxkx.2021.200641
	[ Zhou X X, Wu Y L, Li M Y, et al. Automatic vegetation extraction method based on feature separation mechanism with deep learning[J]. Journal of Geo-information Science, 2021, 23(9):1675-1689. ] DOI: 10.12082/dqxxkx.2021.200641 doi: 10.12082/dqxxkx.2021.200641
[21]	Safonova A, Guirado E, Maglinets Y, et al. Olive tree biovolume from UAV multi-resolution image segmentation with Mask R-CNN[J]. Sensors, 2021, 21(5):1617. DOI: 10.3390/s21051617 doi: 10.3390/s21051617
[22]	Rsnen A, Virtanen T. Data and resolution requirements in mapping vegetation in spatially heterogeneous landscapes[J]. Remote Sensing of Environment, 2019, 230:111207. DOI: 10.1016/j.rse.2019.05.026 doi: 10.1016/j.rse.2019.05.026
[23]	Sekertekin A, Marangoz A M, Akcin H. Pixel-based classification analysis of land use land cover using Sentinel-2 and Landsat-8 data. Int. Arch. Photogramm. Remote Sens. Spat. Inf. Sci. 2017, 42,91-93. DOI: 10.5194/isprs-archives-XLII-4-W6-91-2017 doi: 10.5194/isprs-archives-XLII-4-W6-91-2017
[24]	张军国, 韩欢庆, 胡春鹤, 等. 基于无人机多光谱图像的云南松虫害区域识别方法[J]. 农业机械学报, 2018, 49(5):249-255.
	[ Zhang J G, Han H Q, Hu C H, et al. Identification method of pinus yunnanensis pest area based on UAV multispectral images[J]. Transactions of the Chinese Society of Agricultural Machinery, 2018, 49(5):249-255. ] DOI: 10.6041/j.issn.1000-1298.2018.05.029 doi: 10.6041/j.issn.1000-1298.2018.05.029
[25]	Wei W, Polap D, Li X, et al. Study on remote sensing image vegetation classification method based on decision tree classifier [C]// 2018 IEEE Symposium Series on Computational Intelligence (SSCI). IEEE, 2018.,18-21.
[26]	肖武, 任河, 吕雪娇, 等. 基于无人机遥感的高潜水位采煤沉陷湿地植被分类[J]. 农业机械学报, 2019, 50(2):184-193.
	[ Xiao W, Ren H, Lu X J, et al. Vegetation classification by using UAV remote sensing in coal mining subsidence wetland with high ground-water level[J]. Transactions of the Chinese Society of Agricultural Machinery, 2019, 50(2):177-186. ] DOI: 10.6041/j.issn.1000-1298.2019.02.020 doi: 10.6041/j.issn.1000-1298.2019.02.020
[27]	Deng Y, Liu, et al. Application of UAV-based multi-angle hyperspectral remote sensing in fine vegetation classification[J]. Remote Sensing, 2019, 11(23):2753. DOI: 10.3390/rs11232753 doi: 10.3390/rs11232753
[28]	井然, 邓磊, 赵文吉, 等. 基于可见光植被指数的面向对象湿地水生植被提取方法[J]. 应用生态学报, 2016, 27(5):1427-1436.
	[ Jing R, Deng L, Zhao W J, et al. Object-oriented aquatic vegetation extracting approach based on visible vegetation indices[J]. Chinese Journal of Applied Ecology, 2016:1427-1436. ] DOI: 10.13287/j.1001-9332.201605.002 doi: 10.13287/j.1001-9332.201605.002
[29]	Furuya D, Aguiar J, Estrabis N V, et al. A machine learning approach for mapping forest vegetation in riparian zones in an atlantic biome environment using Sentinel-2 imagery[J]. Remote Sensing, 2020, 12(4086):1. DOI: 10.3390/rs12244086 doi: 10.3390/rs12244086
[30]	Spiering D J, Larsen C, Potts D L. Modelling vegetation succession in post-industrial ecosystems using vegetation classification in aerial photographs, Buffalo, New York[J]. Landscape and Urban Planning, 2020, 198:103792-. DOI: 10.1016/j.landurbplan.2020.103792 doi: 10.1016/j.landurbplan.2020.103792
[31]	Geng R, Jin S, Fu B, et al. Object-based wetland classification using multi-feature combination of ultra-high spatial resolution multispectral images[J]. Canadian Journal of Remote Sensing, 2020. 46(6):784-802. DOI: 10.1080/07038992.2021.1872374 doi: 10.1080/07038992.2021.1872374
[32]	Liu M, Fu B, Xie S, et al. Comparison of multi-source satellite images for classifying marsh vegetation using DeepLabV3 Plus deep learning algorithm[J]. Ecological Indicators, 2021, 125(11):107562. DOI: 10.1016/j.ecolind.2021.107562 doi: 10.1016/j.ecolind.2021.107562
[33]	Liu, Yu, Gu, et al. The impact of spatial resolution on the classification of vegetation types in highly fragmented planting areas based on Unmanned Aerial Vehicle hyperspectral images[J]. Remote Sensing, 2020,12(1):146-. DOI: 10.3390/rs12010146 doi: 10.3390/rs12010146
[34]	Zhang L P, Zhang L F, 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. DOI: 10.1109/MGRS.2016.2540798 doi: 10.1109/MGRS.2016.2540798
[35]	Krizhevsky A, Sutskever I, Hinton G. ImageNet classification with deep convolutional neural networks[J]. Advances in neural information processing systems, 2012, 25(2). DOI: 10.1145/3065386 doi: 10.1145/3065386
[36]	Long J, Shelhamer E, Darrell T. Fully convolutional networks for semantic segmentation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2015, 39(4):640-651. DOI: 10.1109/CVPR.2015.7298965 doi: 10.1109/CVPR.2015.7298965
[37]	Weinstein B G, Marconi S, Bohlman S A, et al. Individual tree-crown detection in RGB imagery using semi-supervised deep learning neural networks. Remote Sensing, 2019, 11(11),1309. DOI: 10.1101/532952 doi: 10.1101/532952
[38]	林志玮, 涂伟豪, 黄嘉航, 等. 深度语义分割的无人机图像植被识别[J]. 山地学报, 2018, 36(6):135-145.
	[ Lin Z W, Tu W H, Huang J H, et al. Unmanned Aerial Vehicle vegetation image recognition using deep semantic segmentation[J]. Mountain Research, 2018, 36(6):135-145. ] DOI: 10.16089/j.cnki.1008-2786.000390 doi: 10.16089/j.cnki.1008-2786.000390
[39]	Sun K, Xiao B, Liu D, et al. Deep high-resolution representation learning for human pose estimation [C]// 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). arXiv, 2019. DOI: 10.1109/CVPR.2019.00584 doi: 10.1109/CVPR.2019.00584
[40]	Huang G, Liu Z, Laurens V, et al. Densely connected convolutional networks [C]//2017 IEEE Conference on Computer Vision and Pattern Recognition(CVPR). IEEE, 2017:2261-2269. DOI: 10.1109/CVPR.2017.243 doi: 10.1109/CVPR.2017.243
[41]	Liu R, Lehman J, Molino P, et al. An intriguing failing of convolutional neural networks and the CoordConv solution[J]. 2018,arXiv:1807.03247v1
[42]	Yao X D, Yang H, Wu Y L, et al. Land use classification of the deep convolutional neural network method reducing the loss of spatial features[J]. Sensors, 2019,19(12):2792-. DOI: 10.3390/s19122792 doi: 10.3390/s19122792
[43]	Yu F, Koltun V. Multi-scale context aggregation by dilated convolutions[J]. arXiv preprint arXiv:1511.07122, 2015.
[44]	Chen L C, Papandreou G, Kokkinos I, et al. Deeplab: Semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected crfs[J]. IEEE transactions on pattern analysis and machine intelligence, 2017, 40(4):834-848. DOI: 10.1109/TPAMI.2017.2699184 doi: 10.1109/TPAMI.2017.2699184
[45]	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):2481-2495. DOI: 10.1109/TPAMI.2016.2644615 doi: 10.1109/TPAMI.2016.2644615 pmid: 28060704
[46]	Yu C, Wang J, Peng C, et al. BiSeNet: Bilateral segmentation network for real-time semantic segmentation [C]// European Conference on Computer Vision. Springer, Cham, 2018. DOI: 10.1007/978-3-030-01261-8_20 doi: 10.1007/978-3-030-01261-8_20
[47]	Chen L C, Zhu Y, Papandreou G, et al. Encoder-decoder with atrous separable convolution for semantic image segmentation [C]// European Conference on Computer Vision. Springer, Cham, 2018:833-851. DOI: 10.1007/978-3-030-01234-2_49 doi: 10.1007/978-3-030-01234-2_49

影像	研究区A	研究区B	研究区C	研究区D
位置	安徽省	江苏省	山东省	江苏省
高度/m	260	150	450	450
分辨率/m	0.03	0.02	0.04	0.04
时间	2019-05-16	2020-11-07	2019-05-28	2021-05-02
飞行次数/次	1	1	1	1
航向重叠度/%	70	70	70	70
旁向重叠度/%	80	80	80	80
图像尺寸/像素	57 285×86 854	32 066×21 832	15 850×15 061	34 505×23 399

区域	草地	灌丛	乔木	OA	Mean F1
研究区B	76.77	77.95	80.94	89.87	78.55
研究区C	66.64	58.96	80.02	86.22	68.54
研究区D	78.32	74.23	88.80	92.52	80.45
平均精度	73.91	70.38	83.25	89.54	75.85

		IOU		Mean IOU
	草地	灌丛	乔木	Mean IOU
研究区B	66.95	66.77	71.30	68.34
研究区C	54.02	47.93	70.26	57.40
研究区D	67.22	64.01	80.57	70.60
平均精度	63.73	59.57	74.04	65.45

	草地	灌丛	乔木	UA
草地	87.92	4.93	2.34	88.40
灌丛	5.04	86.85	5.24	88.37
乔木	4.32	5.98	89.94	93.47
PA	97.92	86.85	89.94
OA 89.87			Kappa 81.72

	草地	灌丛	乔木	UA
草地	81.75	1.85	2.92	81.40
灌丛	2.14	83.55	5.84	81.77
乔木	14.23	13.17	89.92	89.47
PA%	81.75	83.55	89.92
OA 86.22			Kappa 78.57

基于多尺度特征感知网络的城市植被无人机遥感分类

Urban Vegetation Classification based on Multi-scale Feature Perception Network for UAV Images

RichHTML

PDF (PC)

赞

可视化

摘要/Abstract

引用本文

使用本文

图/表 28

参考文献 47

相关文章 15

Metrics

本文评价

推荐阅读 0

	草地	灌丛	乔木	UA
草地	88.02	2.33	1.11	86.53
灌丛	2.54	88.72	2.98	88.98
乔木	4.86	3.75	93.42	94.05
PA	88.02	88.72	93.42
OA 92.52			Kappa 84.74

区域	方法	草地	灌丛	乔木	OA	Mean F1
研究区B	MFDN	76.77	77.95	80.94	89.87	78.55
	DeeplabV3+	72.21	69.45	75.66	87.64	72.11
	Segnet	53.70	58.05	64.08	83.45	58.61
	Bisenet	50.01	45.66	63.98	79.01	53.19
研究区C	MFDN	66.64	58.96	80.02	86.22	68.54
	DeeplabV3+	63.23	58.24	74.55	83.18	65.34
	Segnet	48.70	50.01	71.11	80.91	56.61
	Bisenet	31.62	42.14	68.77	78.17	47.51
研究区D	MFDN	78.32	74.23	88.80	92.52	80.45
	DeeplabV3+	70.21	69.55	83.92	90.83	74.56
	Segnet	72.21	60.06	71.66	82.31	67.97
	Bisenet	39.01	43.74	72.06	76.66	51.60

方法	浮点运算数(Flops)/G	参数数量(Params)/百万	测试时间/s	空间占用/M
MFDN	246.54	50.28	6.21	191
DeeplabV3+	333.77	59.35	7.18	679
Segnet	150.18	8.91	4.08	102
Bisenet	84.38	6.33	3.48	72

[1]	刘潇, 刘智, 林雨准, 王淑香, 左溪冰. 面向遥感影像场景分类的类中心知识蒸馏方法[J]. 地球信息科学学报, 2023, 25(5): 1050-1063.
[2]	侯慧太, 蓝朝桢, 徐青. 基于卫星影像全局和局部深度学习特征检索的无人机绝对定位方法[J]. 地球信息科学学报, 2023, 25(5): 1064-1074.
[3]	张金雷, 陈奕洁, Panchamy Krishnakumari, 金广垠, 王骋程, 杨立兴. 基于注意力机制的城市轨道交通网络级多步短时客流时空综合预测模型[J]. 地球信息科学学报, 2023, 25(4): 698-713.
[4]	李心雨, 闫浩文, 王卓, 王炳瑄. 街景图像与机器学习相结合的道路环境安全感知评价与影响因素分析[J]. 地球信息科学学报, 2023, 25(4): 852-865.
[5]	衡雪彪, 许捍卫, 唐璐, 汤恒, 许怡蕾. 基于改进全卷积神经网络模型的土地覆盖分类方法研究[J]. 地球信息科学学报, 2023, 25(3): 495-509.
[6]	于方圆, 曹家玮, 李发源, 李思进. 顾及对象特征的地面式光伏电站提取及减碳效益评估[J]. 地球信息科学学报, 2023, 25(3): 529-545.
[7]	高建文, 管海燕, 彭代锋, 许正森, 康健, 季雅婷, 翟若雪. 基于局部-全局语义特征增强的遥感影像变化检测网络模型[J]. 地球信息科学学报, 2023, 25(3): 625-637.
[8]	江宝得, 许少芬, 王进, 王淼. 基于条件生成对抗网络的电子地图可视水印去除方法[J]. 地球信息科学学报, 2023, 25(2): 288-297.
[9]	谢谦, 陆明, 谢春山. 基于LBS和深度学习的旅游景区客流量的高时频预测[J]. 地球信息科学学报, 2023, 25(2): 298-310.
[10]	黄昕, 毛政元. 基于时空多图卷积网络的网约车乘客需求预测[J]. 地球信息科学学报, 2023, 25(2): 311-323.
[11]	刘洋, 康健, 管海燕, 汪汉云. 基于双注意力残差网络的高分遥感影像道路提取模型[J]. 地球信息科学学报, 2023, 25(2): 396-408.
[12]	于明洋, 陈肖娴, 张文焯, 刘耀辉. 融合网格注意力阀门和特征金字塔结构的高分辨率遥感影像建筑物提取[J]. 地球信息科学学报, 2022, 24(9): 1785-1802.
[13]	赵祥, 王涛, 张艳, 郑迎辉, 张昆, 王龙辉. 基于改进DeepLabv3+孪生网络的遥感影像变化检测方法[J]. 地球信息科学学报, 2022, 24(8): 1604-1616.
[14]	闵杰, 张永生, 于英, 吕可枫, 王自全, 张磊. 增强型遥感影像SRGAN算法及其在三维重建精度提升中的应用[J]. 地球信息科学学报, 2022, 24(8): 1631-1644.
[15]	彭瑞, 赵文智, 张立强, 陈学泓. 基于多尺度对比学习的弱监督遥感场景分类[J]. 地球信息科学学报, 2022, 24(7): 1375-1390.

			IOU		Mean IOU
		草地	灌丛	乔木	Mean IOU
研究区B	MFDN	66.95	66.77	71.30	68.34
	DeeplabV3+	61.71	60.80	65.75	62.42
	Segnet	46.65	43.42	52.96	48.34
	Bisenet	40.15	35.98	43.61	39.91
研究区C	MFDN	54.02	47.93	70.26	57.40
	DeeplabV3+	50.11	46.33	65.01	56.75
	Segnet	40.07	41.22	60.27	47.18
	Bisenet	20.97	31.07	57.66	36.56
研究区D	MFDN	67.22	64.01	80.57	70.60
	DeeplabV3+	61.44	60.62	74.35	65.47
	Segnet	62.36	50.97	56.98	56.77
	Bisenet	28.96	32.77	61.43	41.05

	研究区B			研究区C			研究区D
	OA	F1	IOU	OA	F1	IOU	OA	F1	IOU
MFDN	89.87	78.55	68.34	86.22	68.54	57.40	92.52	80.45	70.60
MFDN-noCoord	88.29	75.51	65.52	84.69	66.24	57.29	90.86	78.45	64.80
MFDN-noRMF	87.15	71.62	59.55	83.55	64.33	54.87	89.94	74.06	62.04
MFDN-noDSC	88.03	71.76	60.19	83.72	65.21	55.56	90.77	77.17	67.26
MFDN-Baseline	86.76	68.56	58.66	82.23	60.94	51.44	89.02	72.52	61.74