计算机科学 ›› 2019, Vol. 46 ›› Issue (12): 286-291.doi: 10.11896/jsjkx.181202257

• 图形图像与模式识别 • 上一篇    下一篇

基于词向量融合的遥感场景零样本分类算法

吴晨1,3, 袁昱纬2, 王宏伟3, 刘宇3, 刘思彤4, 全吉成3   

  1. (中国人民解放军海军航空大学 山东 烟台264001)1;
    (91977部队 北京102200)2;
    (中国人民解放军空军航空大学 长春130022)3 ;
    (西安飞行学院 西安710000)4
  • 收稿日期:2018-12-04 出版日期:2019-12-15 发布日期:2019-12-17
  • 通讯作者: 全吉成(1960-),男,博士,教授,主要研究方向为三维虚拟现实,E-mail:jicheng_quan@126.com。
  • 作者简介:吴晨(1991-),男,博士生,主要研究方向为遥感图像场景分类,E-mail:wuchen_research@aliyun.com;袁昱纬(1988-),男,博士,工程师,主要研究方向为遥感图像智能识别;王宏伟(1984-),男,博士生,讲师,主要研究方向为遥感图像智能识别;刘宇(1968-),男,博士,副教授,主要研究方向为遥感图像解译;刘思彤(1990-),女,硕士,讲师,主要研究方向为遥感图像处理。
  • 基金资助:
    本文受国家青年自然科学基金(61301233)资助。

Word Vectors Fusion Based Remote Sensing Scenes Zero-shot Classification Algorithm

WU Chen1,3, YUAN Yu-wei2, WANG Hong-wei3, LIU Yu3, LIU Si-tong4, QUAN Ji-cheng3   

  1. (Naval Aeronautical University,Yantai,Shandong 264001,China)1;
    (The 91977 of PLA,Beijing 102200,China)2;
    (Aviation University Air Force,Changchun 130022,China)3;
    (Xi’an Flight Academy of PLA,Xi’an 710000,China)4
  • Received:2018-12-04 Online:2019-12-15 Published:2019-12-17

摘要: 零样本分类算法无须标注要识别的类别样本,因而能大幅度降低实际应用成本,近年来受到广泛关注。遥感场景类别的语义词向量与图像特征空间原型的结构不一致问题,严重影响了遥感场景零样本的分类效果。利用不同词向量间的互补性,文中提出一种基于语义词向量融合的遥感场景零样本分类算法,即耦合式解析字典学习(Coupled Analysis Dictionary Learning,CADL)方法。首先,采用稀疏编码效率较高的解析字典学习方法获取各语义词向量的稀疏系数,以减少冗余信息;然后,将对应的稀疏编码系数串接后作为融合语义词向量表示,并将融合语义词向量线性映射到图像特征空间,与图像特征空间场景类别原型表示进行结构对齐,以降低结构差异性;最后,计算得到要识别的场景类别的图像特征原型,并采用最近邻分类器在图像特征空间完成分类。在UCM和AID数据集上对多种语义词向量的融合进行定量实验,同时将RSSCN7数据集作为已知场景类别的数据集来对两幅实际遥感图像进行定性实验。在UCM和AID上的定量实验分别获得了最高总体分类准确度48.40%和60.23%,相比于典型零样本分类方法的总体分类准确度分别提升了4.80%和6.98%。对两幅实际遥感图像的定性实验,同样获得了最佳零样本的分类效果。实验结果表明,多种语义词向量融合,可以获得与图像特征空间原型结构更一致的语义词向量,且显著提升了遥感场景零样本分类的准确度。

关键词: 词向量融合, 结构对齐, 解析字典学习, 零样本分类, 遥感场景分类

Abstract: Zero-shot classification algorithm does not need to label the sample of unseen classes to be recognized,so it can greatly reduce the cost of practical application,which has attracted wide attention in recent years.The problem of structure difference between word vectors and image feature prototypesseriously affects the zero-shot classification performance of remote sensing scenes.Based on the complementarity among different kinds of word vectors,the remote sensing scenes zero-shot classification algorithm based on word vectors fusion,named coupled analysis dictionary lear-ning method,was proposed.Firstly,the sparse coefficients of different kinds of word vectors are obtained by the more efficient analysis dictionary learning to reduce the redundant information.Then,the sparse coefficients are concatenated and denoted as the fused word vectors,and a structure alignment operation is performed based on the image feature prototypes to reduce structural differences by embedding the fused word vectors into image feature space.Finally,the image feature prototypes of the scene classes unseen are calculated,and the nearest neighbor classifier is employed to complete the classification in the image feature space.Quantitative experiments of the fusion of multiple semantic word vectors were carried out on UCM and AID datasets.At the same time,two real remote sensing images were qualitatively tested with RSSCN7 datasets as the seen dataset.Auantitative experiments obtaines the highest overall classification accuracies of 48.40% and 60.23% on UCM and AID,which respectively exceeds the typical comparative methods by 4.80% and 6.98%.In qualitative experiments on two real remote sensing images ,the algorithm also obtaines the best zero-shot classification performance.The experimental results show that the fused word vectors are more consistent with the prototypes in image feature space,and the zero-shot classification accuracies of remote sensing scenes can be significantly improved.

Key words: Analysis dictionary learning, Remote sensing scenes classification, Structure alignment, Word vectors fusion, Zero-shot classification

中图分类号: 

  • TP181
[1]CHEN S,TIAN Y L.Pyramid of spatial relations for scene-level land use classification[J].IEEE Transactions on Geoscience & Remote Sensing,2014,53(4):1947-1957.
[2]LI A,LU Z,WANG L,et al.Zero-shot scene classification for high spatial resolution remote sensing images[J].IEEE Tran-sactions on Geoscience & Remote Sensing,2017,55(7):4157-4167.
[3]XIAN Y Q,AKATA Z,SHARMA G,et al.Latent embeddings for zero-shot classification[C]//Proceedings of Computer Vision and Pattern Recognition.LA USA:IEEE,2016:69-77.
[4]WANG D,LI Y,LIN Y,et al.Relational Knowledge Transfer for Zero-Shot Learning[C]//Proceedings of AAAI.CA USA:2016,2-7.
[5]ZHANG Z,SALIGRAMA V.Zero-shot learning via joint latent similarity embedding[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.LA USA:IEEE,2016:6034-6042.
[6]ZHANG Z,SALIGRAMA V.Zero-shot learning via semantic similarity embedding[C]//Proceedings of the IEEE InternationalConference on Computer Vision.Boston USA:IEEE,2015:4166-4174.
[7]WANG Q,CHEN K.Zero-shot visual recognition via bidirec- tional latent embedding[J].International Journal of Computer Vision,2017,124(3):356-383.
[8]LI Y,WANG D,HU H,et al.Zero-shot recognition using dual visual-semantic mapping paths[J].arXiv:1703.05002,2017.
[9]ZHAO B,WU B,WU T,et al.Zero-shot learning posed as a missing data problem[J].arXiv:1612.00560,2016.
[10]LAMPERT C H,NICKISCH H,HARMELING S.Attribute- based classification for zero-shot visual object categorization[J].IEEE Transactions on Pattern Analysis & Machine Intelligence,2014,36(3):453-465.
[11]SOCHER R,GANJOO M,MANNING C D,et al.Zero-shot learning through cross-modal transfer[C]//Proceedings of Advances in Neural Information Processing Systems.Lake Tahoe USA,ACM 2013:935-943.
[12]MIKOLOV T,CHEN K,CORRADO G,et al.Efficient estimation of word representations in vector space[J].arXiv:1301.3781,2013.
[13]SOCHER R,MANNING C D.Glove:Global vectors for word representation[C]//Empirical Methods in Natural Language Processing (EMNLP).2014:1532-1543 [14]YANG M,CHANG H Y,LUO W X.Discriminative analysis-synthesis dictionary learning for image classification[J].Neurocomputing,2017,219:404-411.
[15]WANG J J,GUO Y Q,GUO J,et al.Synthesis linear classifier-based analysis dictionary learning for pattern classification [J].Neurocomputing,2017,238:103-113.
[16]RAVISHANKAR S,BRESLER Y.L0 sparsifying transform learning with efficient optimal updates and convergence guarantees[J].IEEE Transactions on Signal Processing,2015,63(9):2389-2404.
[17]YANG Y,NEWSAM S.Bag-of-visual-words and spatial extensions for land-use classification[C]//Proceedings of the 18th SIGSPATIAL International Conference on Advances In Geographic Information Systems.ACM,2010:270-279.
[18]XIA G S,HU J,HU F,et al.AID:A benchmark dataset for performance evaluation of aerial scene classification[J].IEEE Transactions on Geoscience and Remote Sensing,2017,55(7):3965-3981.
[19]ZOU Q,NI L,ZHANG T,et al.Deep learning based feature selection for remote sensing scene classification[J].IEEEGeo-science & Remote Sensing Letters,2015,12(11):2321-2325.
[20]SIMONYAN K,ZISSERMAN A.Very deep convolutional networks for large-scale image recognition[J].arXiv:1409.1556,2014.
[1] 程章桃, 钟婷, 张晟铭, 周帆.
基于图学习的推荐系统研究综述
Survey of Recommender Systems Based on Graph Learning
计算机科学, 2022, 49(9): 1-13. https://doi.org/10.11896/jsjkx.210900072
[2] 熊丽琴, 曹雷, 赖俊, 陈希亮.
基于值分解的多智能体深度强化学习综述
Overview of Multi-agent Deep Reinforcement Learning Based on Value Factorization
计算机科学, 2022, 49(9): 172-182. https://doi.org/10.11896/jsjkx.210800112
[3] 齐秀秀, 王佳昊, 李文雄, 周帆.
基于概率元学习的矩阵补全预测融合算法
Fusion Algorithm for Matrix Completion Prediction Based on Probabilistic Meta-learning
计算机科学, 2022, 49(7): 18-24. https://doi.org/10.11896/jsjkx.210600126
[4] 高振卓, 王志海, 刘海洋.
嵌入典型时间序列特征的随机Shapelet森林算法
Random Shapelet Forest Algorithm Embedded with Canonical Time Series Features
计算机科学, 2022, 49(7): 40-49. https://doi.org/10.11896/jsjkx.210700226
[5] 孙晓寒, 张莉.
基于评分区域子空间的协同过滤推荐算法
Collaborative Filtering Recommendation Algorithm Based on Rating Region Subspace
计算机科学, 2022, 49(7): 50-56. https://doi.org/10.11896/jsjkx.210600062
[6] 刘卫明, 安冉, 毛伊敏.
基于聚类和WOA的并行支持向量机算法
Parallel Support Vector Machine Algorithm Based on Clustering and WOA
计算机科学, 2022, 49(7): 64-72. https://doi.org/10.11896/jsjkx.210500040
[7] 周慧, 施皓晨, 屠要峰, 黄圣君.
基于主动采样的深度鲁棒神经网络学习
Robust Deep Neural Network Learning Based on Active Sampling
计算机科学, 2022, 49(7): 164-169. https://doi.org/10.11896/jsjkx.210600044
[8] 苏丹宁, 曹桂涛, 王燕楠, 王宏, 任赫.
小样本雷达辐射源识别的深度学习方法综述
Survey of Deep Learning for Radar Emitter Identification Based on Small Sample
计算机科学, 2022, 49(7): 226-235. https://doi.org/10.11896/jsjkx.210600138
[9] 于滨, 李学华, 潘春雨, 李娜.
基于深度强化学习的边云协同资源分配算法
Edge-Cloud Collaborative Resource Allocation Algorithm Based on Deep Reinforcement Learning
计算机科学, 2022, 49(7): 248-253. https://doi.org/10.11896/jsjkx.210400219
[10] 王宇飞, 陈文.
基于DECORATE集成学习与置信度评估的Tri-training算法
Tri-training Algorithm Based on DECORATE Ensemble Learning and Credibility Assessment
计算机科学, 2022, 49(6): 127-133. https://doi.org/10.11896/jsjkx.211100043
[11] 洪志理, 赖俊, 曹雷, 陈希亮, 徐志雄.
基于遗憾探索的竞争网络强化学习智能推荐方法研究
Study on Intelligent Recommendation Method of Dueling Network Reinforcement Learning Based on Regret Exploration
计算机科学, 2022, 49(6): 149-157. https://doi.org/10.11896/jsjkx.210600226
[12] 陈章辉, 熊贇.
基于解耦-检索-生成的图像风格化描述生成模型
Stylized Image Captioning Model Based on Disentangle-Retrieve-Generate
计算机科学, 2022, 49(6): 180-186. https://doi.org/10.11896/jsjkx.211100129
[13] 徐辉, 康金梦, 张加万.
基于特征感知的数字壁画复原方法
Digital Mural Inpainting Method Based on Feature Perception
计算机科学, 2022, 49(6): 217-223. https://doi.org/10.11896/jsjkx.210500105
[14] 许杰, 祝玉坤, 邢春晓.
机器学习在金融资产定价中的应用研究综述
Application of Machine Learning in Financial Asset Pricing:A Review
计算机科学, 2022, 49(6): 276-286. https://doi.org/10.11896/jsjkx.210900127
[15] 罗俊仁, 张万鹏, 陆丽娜, 陈璟.
即时策略博弈在线对抗规划方法综述
Survey on Online Adversarial Planning for Real-time Strategy Game
计算机科学, 2022, 49(6): 287-296. https://doi.org/10.11896/jsjkx.210600168
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!