High-Resolution SAR Image Classification via Deep Convolutional Autoencoders

被引:249
作者
Geng, Jie [1 ]
Fan, Jianchao [2 ,3 ]
Wang, Hongyu [1 ]
Ma, Xiaorui [1 ]
Li, Baoming [4 ]
Chen, Fuliang [4 ]
机构
[1] Dalian Univ Technol, Sch Informat & Commun Engn, Dalian 116024, Peoples R China
[2] Dalian Univ Technol, Sch Control Sci & Engn, Dalian 116024, Peoples R China
[3] Natl Marine Environm Monitoring Ctr, Dalian 116023, Peoples R China
[4] Space Star Technol Co Ltd, Beijing 100086, Peoples R China
基金
中国博士后科学基金;
关键词
Deep convolutional autoencoders (DCAEs); supervised classification; synthetic aperture radar (SAR) image; FACE RECOGNITION; NEURAL-NETWORKS; REPRESENTATION; FEATURES; FUSION; ROBUST; FILTER; GABOR;
D O I
10.1109/LGRS.2015.2478256
中图分类号
P3 [地球物理学]; P59 [地球化学];
学科分类号
0708 ; 070902 ;
摘要
Synthetic aperture radar (SAR) image classification is a hot topic in the interpretation of SAR images. However, the absence of effective feature representation and the presence of speckle noise in SAR images make classification difficult to handle. In order to overcome these problems, a deep convolutional autoencoder (DCAE) is proposed to extract features and conduct classification automatically. The deep network is composed of eight layers: a convolutional layer to extract texture features, a scale transformation layer to aggregate neighbor information, four layers based on sparse autoencoders to optimize features and classify, and last two layers for postprocessing. Compared with hand-crafted features, the DCAE network provides an automatic method to learn discriminative features from the image. A series of filters is designed as convolutional units to comprise the gray-level cooccurrence matrix and Gabor features together. Scale transformation is conducted to reduce the influence of the noise, which integrates the correlated neighbor pixels. Sparse autoencoders seek better representation of features to match the classifier, since training labels are added to fine-tune the parameters of the networks. Morphological smoothing removes the isolated points of the classification map. The whole network is designed ingeniously, and each part has a contribution to the classification accuracy. The experiments of TerraSAR-X image demonstrate that the DCAE network can extract efficient features and perform better classification result compared with some related algorithms.
引用
收藏
页码:2351 / 2355
页数:5
相关论文
共 18 条
[11]   A gender classification system robust to occlusion using Gabor features based (2D)2PCA [J].
Rai, Preeti ;
Khanna, Pritee .
JOURNAL OF VISUAL COMMUNICATION AND IMAGE REPRESENTATION, 2014, 25 (05) :1118-1129
[12]   Active extreme learning machines for quad-polarimetric SAR imagery classification [J].
Samat, Alim ;
Gamba, Paolo ;
Du, Peijun ;
Luo, Jieqiong .
INTERNATIONAL JOURNAL OF APPLIED EARTH OBSERVATION AND GEOINFORMATION, 2015, 35 :305-319
[13]  
Srivastava N, 2014, J MACH LEARN RES, V15, P1929
[14]   Deep Learning Face Representation from Predicting 10,000 Classes [J].
Sun, Yi ;
Wang, Xiaogang ;
Tang, Xiaoou .
2014 IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR), 2014, :1891-1898
[15]  
Vincent P, 2010, J MACH LEARN RES, V11, P3371
[16]   Robust Face Recognition via Sparse Representation [J].
Wright, John ;
Yang, Allen Y. ;
Ganesh, Arvind ;
Sastry, S. Shankar ;
Ma, Yi .
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2009, 31 (02) :210-227
[17]   MULTILAYER FEATURE LEARNING FOR POLARIMETRIC SYNTHETIC RADAR DATA CLASSIFICATION [J].
Xie, Huiming ;
Wang, Shuang ;
Lin, Kun ;
Lin, Shaopeng ;
Hou, Biao .
2014 IEEE INTERNATIONAL GEOSCIENCE AND REMOTE SENSING SYMPOSIUM (IGARSS), 2014,
[18]   Saliency-Guided Unsupervised Feature Learning for Scene Classification [J].
Zhang, Fan ;
Du, Bo ;
Zhang, Liangpei .
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 2015, 53 (04) :2175-2184