Transfer Change Rules from Recurrent Fully Convolutional Networks for Hyperspectral Unmanned Aerial Vehicle Images without Ground Truth Data

被引:6
作者
Song, Ahram [1 ]
Kim, Yongil [1 ]
机构
[1] Seoul Natl Univ, Dept Civil & Environm Engn, 1 Gwanak Ro, Seoul 08826, South Korea
基金
新加坡国家研究基金会;
关键词
change detection; hyperspectral unmanned aerial vehicle; spectral similarity measures; UNSUPERVISED CHANGE DETECTION; SPECTRAL SIMILARITY; DISCRIMINATION;
D O I
10.3390/rs12071099
中图分类号
X [环境科学、安全科学];
学科分类号
08 ; 0830 ;
摘要
Change detection (CD) networks based on supervised learning have been used in diverse CD tasks. However, such supervised CD networks require a large amount of data and only use information from current images. In addition, it is time consuming to manually acquire the ground truth data for newly obtained images. Here, we proposed a novel method for CD in case of a lack of training data in an area near by another one with the available ground truth data. The proposed method automatically entails generating training data and fine-tuning the CD network. To detect changes in target images without ground truth data, the difference images were generated using spectral similarity measure, and the training data were selected via fuzzy c-means clustering. Recurrent fully convolutional networks with multiscale three-dimensional filters were used to extract objects of various sizes from unmanned aerial vehicle (UAV) images. The CD network was pre-trained on labeled source domain data; then, the network was fine-tuned on target images using generated training data. Two further CD networks were trained with a combined weighted loss function. The training data in the target domain were iteratively updated using he prediction map of the CD network. Experiments on two hyperspectral UAV datasets confirmed that the proposed method is capable of transferring change rules and improving CD results based on training data extracted in an unsupervised way.
引用
收藏
页数:20
相关论文
共 41 条
[1]   Spectral-Spatial Classification of Hyperspectral Images: Three Tricks and a New Learning Setting [J].
Acquarelli, Jacopo ;
Marchiori, Elena ;
Buydens, Lutgarde M. C. ;
Thanh Tran ;
van Laarhoven, Twan .
REMOTE SENSING, 2018, 10 (07)
[2]   Hyperspectral Imaging: A Review on UAV-Based Sensors, Data Processing and Applications for Agriculture and Forestry [J].
Adao, Telmo ;
Hruska, Jonas ;
Padua, Luis ;
Bessa, Jose ;
Peres, Emanuel ;
Morais, Raul ;
Sousa, Joaquim Joao .
REMOTE SENSING, 2017, 9 (11)
[3]  
Adep R.N., 2016, PERSPECTIVESCI, V8, P722, DOI DOI 10.1016/J.PISC.2016.06.070
[4]  
[Anonymous], 2008, Int. Arch. Photogrammetry, Remote Sens. Spatial Inf. Sci.
[5]   A Framework for Automatic and Unsupervised Detection of Multiple Changes in Multitemporal Images [J].
Bovolo, Francesca ;
Marchesi, Silvia ;
Bruzzone, Lorenzo .
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 2012, 50 (06) :2196-2212
[6]   An analysis of spectral variability in hyperspectral imagery: a case study of stressed oil palm detection in Colombia [J].
Camacho, Ariolfo ;
Correa, Claudia V. ;
Arguello, Henry .
INTERNATIONAL JOURNAL OF REMOTE SENSING, 2019, 40 (19) :7603-7623
[7]   A New Approach to Change Vector Analysis Using Distance and Similarity Measures [J].
Carvalho Junior, Osmar A. ;
Guimaraes, Renato F. ;
Gillespie, Alan R. ;
Silva, Nilton C. ;
Gomes, Roberto A. T. .
REMOTE SENSING, 2011, 3 (11) :2473-2493
[8]   An information-theoretic approach to spectral variability, similarity, and discrimination for hyperspectral image analysis [J].
Chang, CI .
IEEE TRANSACTIONS ON INFORMATION THEORY, 2000, 46 (05) :1927-1932
[9]   A Novel Adaptive Scheme for Evaluating Spectral Similarity in High-Resolution Urban Scenes [J].
Chen, Bin ;
Vodacek, Anthony ;
Cahill, Nathan D. .
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING, 2013, 6 (03) :1376-1385
[10]   Change Detection in Multisource VHR Images via Deep Siamese Convolutional Multiple-Layers Recurrent Neural Network [J].
Chen, Hongruixuan ;
Wu, Chen ;
Du, Bo ;
Zhang, Liangpei ;
Wang, Le .
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 2020, 58 (04) :2848-2864