Active extreme learning machines for quad-polarimetric SAR imagery classification

被引:40
作者
Samat, Alim [1 ,2 ]
Gamba, Paolo [3 ]
Du, Peijun [1 ,2 ]
Luo, Jieqiong [1 ,2 ]
机构
[1] Nanjing Univ, Mapping & Geoinformat China, State Adm Surveying, Key Lab Satellite Mapping Technol & Applicat, Nanjing, Jiangsu, Peoples R China
[2] Nanjing Univ, Jiangsu Prov Key Lab Geog Informat Sci & Technol, Nanjing, Jiangsu, Peoples R China
[3] Univ Pavia, Dept Elect Comp & Biomed Engn, I-27100 Pavia, Italy
基金
中国国家自然科学基金;
关键词
PolSAR; Extreme learning machine; Ensemble learning; Active learning; Active extreme learning machines; HIGH-RESOLUTION; HYPERSPECTRAL DATA; DECOMPOSITION; SEGMENTATION; RECOGNITION; MODEL;
D O I
10.1016/j.jag.2014.09.019
中图分类号
TP7 [遥感技术];
学科分类号
081102 ; 0816 ; 081602 ; 083002 ; 1404 ;
摘要
Supervised classification of quad-polarimetric SAR images is often constrained by the availability of reliable training samples. Active learning (AL) provides a unique capability at selecting samples with high representation quality and low redundancy. The most important part of AL is the criterion for selecting the most informative candidates (pixels) by ranking. In this paper, class supports based on the posterior probability function are approximated by ensemble learning and majority voting. This approximation is statistically meaningful when a large enough classifier ensemble is exploited. In this work, we propose to use extreme learning machines and apply AL to quad-polarimetric SAR image classification. Extreme learning machines are ideal because of their fast operation, straightforward solution and strong generalization. As inputs to the so-called active extreme learning machines, both polarimetric and spatial features (morphological profiles) are considered. In order to validate the proposed method, results and performance are compared with random sampling and state-of-the-art AL methods, such as margin sampling, normalized entropy query-by-bagging and multiclass level uncertainty. Experimental results for four quad-polarimetric SAR images collected by RADARSAT-2, AirSAR and EMISAR indicate that the proposed method achieves promising results in different scenarios. Moreover, the proposed method is faster than existing techniques in both the learning and the classification phases. (C) 2014 Elsevier B.V. All rights reserved.
引用
收藏
页码:305 / 319
页数:15
相关论文
共 52 条
[1]   Classification comparisons between dual-pol, compact polarimetric and quad-pol SAR imagery [J].
Ainsworth, T. L. ;
Kelly, J. P. ;
Lee, J. -S. .
ISPRS JOURNAL OF PHOTOGRAMMETRY AND REMOTE SENSING, 2009, 64 (05) :464-471
[2]  
[Anonymous], J MACH LEARN RES
[3]  
[Anonymous], OPTIMAL ACTIVE LEARN
[4]  
[Anonymous], IEEE P 17 INT C PATT
[5]  
[Anonymous], P SPIE
[6]  
[Anonymous], COMBINING PATTERN CL
[7]   Classification of hyperspectral data from urban areas based on extended morphological profiles [J].
Benediktsson, JA ;
Palmason, JA ;
Sveinsson, JR .
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 2005, 43 (03) :480-491
[8]   Bagging predictors [J].
Breiman, L .
MACHINE LEARNING, 1996, 24 (02) :123-140
[9]   A review of target decomposition theorems in radar polarimetry [J].
Cloude, SR ;
Pottier, E .
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 1996, 34 (02) :498-518
[10]   A test statistic in the complex Wishart distribution and its application to change detection in polarimetric SAR data [J].
Conradsen, K ;
Nielsen, AA ;
Sehou, J ;
Skriver, H .
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 2003, 41 (01) :4-19