Dynamic Block-Based Parameter Estimation for MRF Classification of High-Resolution Images

被引:12
作者
Aghighi, Hossein [1 ,2 ]
Trinder, John [1 ]
Tarabalka, Yuliya [3 ]
Lim, Samsung [1 ]
机构
[1] Univ New S Wales, Sch Civil & Environm Engn, Sydney, NSW 2052, Australia
[2] Shahid Beheshti Univ, Fac Earth Sci, Dept Remote Sensing & GIS, Tehran 1983963113, Iran
[3] Inria Sophia Antipolis Mediterranee, AYIN Res Team, F-06902 Sophia Antipolis, France
关键词
Classification; Markov random field (MRF); smoothing parameter; support vector machine (SVM); ALGORITHM;
D O I
10.1109/LGRS.2014.2305913
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
A Markov random field is a graphical model that is commonly used to combine spectral information and spatial context into image classification problems. The contributions of the spatial versus spectral energies are typically defined by using a smoothing parameter, which is often set empirically. We propose a new framework to estimate the smoothing parameter. For this purpose, we introduce the new concepts of dynamic blocks and class label cooccurrence matrices. The estimation is then based on the analysis of the balance of spatial and spectral energies computed using the spatial class co-occurrence distribution and dynamic blocks. Moreover, we construct a new spatially weighted parameter to preserve the edges, based on the Canny edge detector. We evaluate the performance of the proposed method on three data sets: a multispectral DigitalGlobe WorldView-2 and two hyperspectral images, recorded by the AVIRIS and the ROSIS sensors, respectively. The experimental results show that the proposed method succeeds in estimating the optimal smoothing parameter and yields higher classification accuracy values when compared with state-of-the-art methods.
引用
收藏
页码:1687 / 1691
页数:5
相关论文
共 17 条
[1]  
Aghighi H., 2013, P SPIE REM SENS
[2]  
Candemir S, 2010, LECT NOTES COMPUT SC, V6111, P117, DOI 10.1007/978-3-642-13772-3_13
[3]   MODELING AND SEGMENTATION OF NOISY AND TEXTURED IMAGES USING GIBBS RANDOM-FIELDS [J].
DERIN, H ;
ELLIOTT, H .
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 1987, 9 (01) :39-55
[4]   ASSUMPTIONS IMPLICIT IN REMOTE-SENSING DATA ACQUISITION AND ANALYSIS [J].
DUGGIN, MJ ;
ROBINOVE, CJ .
INTERNATIONAL JOURNAL OF REMOTE SENSING, 1990, 11 (10) :1669-1694
[5]   Thematic map comparison: Evaluating the statistical significance of differences in classification accuracy [J].
Foody, GM .
PHOTOGRAMMETRIC ENGINEERING AND REMOTE SENSING, 2004, 70 (05) :627-633
[6]   Optimal edge detection using multiple operators for image understanding [J].
Giannarou, Stamatia ;
Stathaki, Tania .
EURASIP JOURNAL ON ADVANCES IN SIGNAL PROCESSING, 2011,
[7]  
Hedley M., 1992, Journal of Electronic Imaging, V1, P374, DOI 10.1117/12.61158
[8]   Managing the spectral-spatial mix in context classification using Markov random fields [J].
Jia, X. ;
Richards, J. A. .
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS, 2008, 5 (02) :311-314
[9]   A MAXIMUM A POSTERIORI PROBABILITY EXPECTATION MAXIMIZATION ALGORITHM FOR IMAGE-RECONSTRUCTION IN EMISSION TOMOGRAPHY [J].
LEVITAN, E ;
HERMAN, GT .
IEEE TRANSACTIONS ON MEDICAL IMAGING, 1987, 6 (03) :185-192
[10]   Spatially adaptive smoothing parameter selection for Markov random field based sub-pixel mapping of remotely sensed images [J].
Li, Xiaodong ;
Du, Yun ;
Ling, Feng .
INTERNATIONAL JOURNAL OF REMOTE SENSING, 2012, 33 (24) :7886-7901