| 454 | 9 | 243 |
| 下载次数 | 被引频次 | 阅读次数 |
高分辨率遥感影像场景分类近年来已经成为遥感科学领域的一个研究热点,是跨越底层特征与高层语义信息之间语义鸿沟的有效途径。由于高分辨率遥感影像中地物种类繁多、分布复杂且同一地物的不同空间组合也可能代表不同的场景类别,直接基于底层特征进行场景分类不能很好地表达场景语义信息。因此,目前场景分类的代表性方法主要包括基于中层特征的场景分类和基于深度学习的场景分类。针对现有遥感影像场景分类方法存在的局限性,进行了耦合多源地理数据的多分辨率遥感影像场景分类方法研究:(1)在模型层面,针对单一分辨率遥感影像特征表达能力有限的问题,将超分辨率重建思想引入遥感影像场景分类领域,提出了一个新的SRGAN-CNN框架。通过从低分辨率重建到高分辨率,在提升遥感影像分辨率的同时,融入不同分辨率遥感影像的特征,提升了场景影像的特征表达能力。(2)在应用层面,针对仅根据遥感影像的特征进行场景分类往往不足以投入实际应用的问题,在遥感影像场景分类中耦合了多源地理数据。使用Open Street Map路网分割原始遥感影像,并融入兴趣点数据(POI)、人口时序数据(RTUD)和夜间灯光等多源地理数据,提升了场景分类模型的实际应用价值。(3)在数据层面,自制了一个综合性的遥感影像场景分类数据集。所提出的耦合多源地理数据的多分辨率遥感影像场景分类方法在SIRI-WHU公开场景数据集和自制的武汉市多分辨率场景数据集上进行了实验,验证了方法的有效性。
Abstract：High-resolution remote sensing image scene classification has become a research hotspot in the field of remote sensing in recent years, and it is an effective way to bridge the semantic gap between low-level features and high-level semantic information.Due to the wide variety and complex distribution of features in high-resolution remote sensing images, and different spatial combinations of the same feature may also represent different scene categories, scene classification based on low-level features cannot well express the semantic information of the scene.Therefore, the current representative methods of scene classification mainly include scene classification based on middle-level features and scene classification based on deep learning.In view of the limitations of the existing remote sensing image scene classification methods, the study on multi-resolution remote sensing image scene classification methods coupled with multi-source geographic data is conducted.The main research work is as follows:(1) At the model level, aiming at the problem that the feature expression ability of single resolution remote sensing image is limited, the idea of super-resolution reconstruction is introduced into the field of remote sensing image scene classification, and a new SRGAN-CNN framework is proposed.By reconstructing from low-resolution to high-resolution, while improving the resolution of remote sensing images, it incorporates the characteristics of remote sensing images of different resolutions, thereby enhancing the feature expression ability of scene images.(2) At the application level, in view of the problem that scene classification based only on the characteristics of remote sensing images is often insufficient for practical applications, the multi-source geographic data is coupled in the remote sensing image scene classification.Use Open Street Map road network to segment the original remote sensing image and incorporate multi-source geographic data such as point of interest data(POI),population time series data(RTUD) and nighttime lights, which enhances the practical application value of the scene classification model.(3) At the data level, a comprehensive remote sensing image scene classification data set is made.The proposed multi-resolution remote sensing image scene classification method coupled with multi-source geographic data is tested on the SIRI-WHU public scene dataset and the self-made Wuhan multi-resolution scene dataset to verify the effectiveness of the method.
[1] 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.
[2] 宫鹏，黎夏，徐冰.高分辨率影像解译理论与应用方法中的一些研究问题[J].遥感学报，2006,10(1):1-5.
[3] BRATASANU D,NEDELCU I,DATCU M.Bridging the Semantic Gap for Satellite Image Annotation and Automatic Mapping Applications[J].IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing,2011,4(1):193-204.
[4] 李德仁，童庆禧，李荣兴，等.高分辨率对地观测的若干前沿科学问题[J].中国科学：地球科学，2012,42(6):805-813.
[5] SERRANO N,SAVAKIS A E,LUO J.Improved Scene Classification Using Efficient Low-level Features and Semantic Cues[J].Pattern Recognition,2004,37(9):1773-1784.
[6] HOFMANN T.Unsupervised Learning by Probabilistic Latent Semantic Analysis[J].Machine Learning,2001,42(1-2):177-196.
[7] BLEI D M,NG A Y,JORDAN M I.Latent Dirichlet Allocation[J].Journal of Machine Learning Research,2003,3(1):993-1022.
[8] ZHU Q,ZHONG Y,ZHAO B,et al.Bag-of-visual-words Scene Classifier with Local and Global Features for High Spatial Resolution Remote Sensing Imagery[J].IEEE Geoscience and Remote Sensing Letters,2016,13(6):747-751.
[9] YUAN Q Q,SHEN H F,LI T W,et al.Deep Learning in Environmental Remote Sensing:Achievements and Challenges[J].Remote Sensing of Environment,2019,241:111716.
[10] ZHONG Y,FEI F,LIU Y,et al.SatCNN:Satellite Image Dataset Classification Using Agile Convolutional Neural Networks[J].Remote Sensing Letters,2017,8(2):136-145.
[11] HU F,XIA G,HU J,et al.Transferring Deep Convolutional Neural Networks for the Scene Classification of High-Resolution Remote Sensing Imagery[J].Remote Sensing,2015,7(11):14680-14707.
[12] MARMANIS D,DATCU M,ESCH T,et al.Deep Learning Earth Observation Classification Using ImageNet Pretrained Networks[J].IEEE Geoscience and Remote Sensing Letters,2016,13(1):105-109.
[13] ZHU Q,ZHONG Y,LIU Y,et al.A Deep-local-global Feature Fusion Framework for High Spatial Resolution Imagery Scene Classification[J].Remote Sensing,2018,10(4):568.
[14] 谭琨，王雪，杜培军.结合深度学习和半监督学习的遥感影像分类进展[J].中国图象图形学报，2019,24(11):1823-1841.
[15] 乔婷婷，李鲁群.结合数据增广和迁移学习的高分辨率遥感影像场景分类[J].测绘通报，2020(2):37-42.
[16] 许夙晖，慕晓冬，赵鹏，等.利用多尺度特征与深度网络对遥感影像进行场景分类[J].测绘学报，2016,45(7):834-840.
[17] KRIZHEVSKY A,SUTSKEVER I,HINTON G E.Imagenet Classification with Deep Convolutional Neural Networks:Advances in Neural Information Processing Systems[C]∥Advances in Neural Information Processing Systems.London:MIT Press,2012:1097-1105.
[18] IANDOLA F N,HAN S,MOSKEWICZ M W,et al.SqueezeNet:AlexNet-level Accuracy with 50x Fewer Parameters and <0.5 MB Model Size[J].arXiv preprint arXiv:1602.07360,2016.
[19] HUANG G,LIU Z,VAN DER MAATEN L,et al.Densely Connected Convolutional Networks[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW).Honolulu:IEEE,2017:1-10.
[20] SIMONYAN K,ZISSERMAN A.Very Deep Convolutional Networks for Large-scale Image Recognition[J].arXiv preprint arXiv:1409.1556,2014.
基本信息:
中图分类号:TP751
引用信息:
[1]范鑫,胡昌苗,霍连志.耦合多源地理数据的多分辨率遥感影像场景分类方法研究[J].无线电工程,2021,51(12):1449-1460.
基金信息:
国家自然科学基金资助项目(41971396)~~
2021-11-22
2021-11-22
2021-11-22