Robust Classification Technique for Hyperspectral Images Based on 3D-Discrete Wavelet Transform

被引:32
作者
Anand, R. [1 ]
Veni, S. [1 ]
Aravinth, J. [1 ]
机构
[1] Amrita Vishwa Vidyapeetham, Dept Elect & Commun Engn, Amrita Sch Engn, Coimbatore 641112, Tamil Nadu, India
关键词
discrete wavelet transform; support vector machine; machine learning; K-nearest neighbor; random forest; classification; hyperspectral image;
D O I
10.3390/rs13071255
中图分类号
X [环境科学、安全科学];
学科分类号
08 ; 0830 ;
摘要
Hyperspectral image classification is an emerging and interesting research area that has attracted several researchers to contribute to this field. Hyperspectral images have multiple narrow bands for a single image that enable the development of algorithms to extract diverse features. Three-dimensional discrete wavelet transform (3D-DWT) has the advantage of extracting the spatial and spectral information simultaneously. Decomposing an image into a set of spatial-spectral components is an important characteristic of 3D-DWT. It has motivated us to perform the proposed research work. The novelty of this work is to bring out the features of 3D-DWT applicable to hyperspectral images classification using Haar, Fejer-Korovkin and Coiflet filters. Three-dimensional-DWT is implemented with the help of three stages of 1D-DWT. The first two stages of 3D-DWT are extracting spatial resolution, and the third stage is extracting the spectral content. In this work, the 3D-DWT features are extracted and fed to the following classifiers (i) random forest (ii) K-nearest neighbor (KNN) and (iii) support vector machine (SVM). Exploiting both spectral and spatial features help the classifiers to provide a better classification accuracy. A comparison of results was performed with the same classifiers without DWT features. The experiments were performed using Salinas Scene and Indian Pines hyperspectral datasets. From the experiments, it has been observed that the SVM with 3D-DWT features performs better in terms of the performance metrics such as overall accuracy, average accuracy and kappa coefficient. It has shown significant improvement compared to the state of art techniques. The overall accuracy of 3D-DWT+SVM is 88.3%, which is 14.5% larger than that of traditional SVM (77.1%) for the Indian Pines dataset. The classification map of 3D-DWT + SVM is more closely related to the ground truth map.
引用
收藏
页数:19
相关论文
共 41 条
[1]   Performance evaluation of empirical mode decomposition, discrete wavelet transform, and wavelet packed decomposition for automated epileptic seizure detection and prediction [J].
Alickovic, Emina ;
Kevric, Jasmin ;
Subasi, Abdulhamit .
BIOMEDICAL SIGNAL PROCESSING AND CONTROL, 2018, 39 :94-102
[2]  
Anand R, 2017, 2017 INTERNATIONAL CONFERENCE ON ADVANCES IN COMPUTING, COMMUNICATIONS AND INFORMATICS (ICACCI), P1808, DOI 10.1109/ICACCI.2017.8126107
[3]  
Bhushan D. B., 2011, Proceedings of the 2011 International Symposium on Ocean Electronics (SYMPOL 2011), P34, DOI 10.1109/SYMPOL.2011.6170495
[4]   Analysis of Feature Extraction Algorithm Using Two Dimensional Discrete Wavelet Transforms in Mammograms to Detect Microcalcifications [J].
Bose, Subash Chandra ;
Veerasamy, Murugesh ;
Mubarakali, Azath ;
Marina, Ninoslav ;
Hadzieva, Elena .
COMPUTATIONAL VISION AND BIO-INSPIRED COMPUTING, 2020, 1108 :26-39
[5]  
Bowker D.E., 1985, Spectral reflectances of natural targets for use in remote sensing studies, V1139
[6]   Integration of 3-dimensional discrete wavelet transform and Markov random field for hyperspectral image classification [J].
Cao, Xiangyong ;
Xu, Lin ;
Meng, Deyu ;
Zhao, Qian ;
Xu, Zongben .
NEUROCOMPUTING, 2017, 226 :90-100
[7]  
Chang C.-I., 2003, HYPERSPECTRAL IMAGIN
[8]   Mechanical fault diagnosis using Convolutional Neural Networks and Extreme Learning Machine [J].
Chen, Zhuyun ;
Gryllias, Konstantinos ;
Li, Weihua .
MECHANICAL SYSTEMS AND SIGNAL PROCESSING, 2019, 133
[9]  
ElMasry G., 2010, Hyperspectral Imaging for Food Quality Analysis and Control, P3
[10]  
Engel JM, 2015, IEEE ENG MED BIO, P2030, DOI 10.1109/EMBC.2015.7318785