An Ultralightweight Hybrid CNN Based on Redundancy Removal for Hyperspectral Image Classification

被引:0
作者
Ma, Xiaohu [1 ]
Wang, Wuli [1 ]
Li, Wei [2 ]
Wang, Jianbu [3 ]
Ren, Guangbo [3 ]
Ren, Peng [1 ]
Liu, Baodi [4 ]
机构
[1] China Univ Petr East China, Coll Oceanog & Space Informat, Qingdao 266580, Peoples R China
[2] Beijing Inst Technol, Sch Informat & Elect, Beijing 100081, Peoples R China
[3] Minist Nat Resources, Inst Oceanog 1, Lab Marine Phys & Remote Sensing, Qingdao 266061, Peoples R China
[4] China Univ Petr East China, Coll Control Sci & Engn, Qingdao 266580, Peoples R China
来源
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING | 2024年 / 62卷
基金
中国国家自然科学基金;
关键词
Feature extraction; Convolution; Computational modeling; Kernel; Training; Three-dimensional displays; Convolutional neural networks; A few training samples; hyperspectral image classification; redundancy removal; ultralightweight;
D O I
10.1109/TGRS.2024.3356524
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
Convolutional neural network (CNN)-based hyperspectral image (HSI) classification models often exhibit high volume and complexity. This not only poses challenges in deploying them on mobile and embedded devices due to storage and power constraints, but also introduces a dilemma between the growing demand for labeled samples and the high cost associated with manual labeling. To address these challenges, we propose an ultralightweight hybrid CNN based on redundancy removal (ULite-R2HCN), specifically designed for HSI classification in scenarios with limited samples. To reduce computational costs and enhance feature extraction effectiveness, we focus on optimizing the widely used depthwise convolution (DW-Conv) and pointwise convolution (PW-Conv) in the lightweight HSI classification model. For DW-Conv, we design a spatial convolution with redundancy removal (R2Spatial-Conv). This involves the design of multiscale 3-D convolution kernels with specific structures instead of 2-D convolution kernels, aiming to reduce redundant convolution kernels and extract multiscale spatial features. Simultaneously, for PW-Conv, we design a spectral convolution with redundancy removal (R2Spectral-Conv). This utilizes a "copy-splicing-grouping" structure to extract spectral features within arbitrary range intervals, effectively reducing redundant spectral extractions and capturing long-range spectral relationships. Numerous experiments have shown that the proposed ULite-R2HCN achieves higher classification accuracy with an ultralight volume for a few training samples. In addition, sufficient ablation experiments also verified the advanced performance of the designed R2Spatial-Conv and R2Spectral-Conv.
引用
收藏
页码:1 / 12
页数:12
相关论文
共 34 条
[1]   A Fast and Compact 3-D CNN for Hyperspectral Image Classification [J].
Ahmad, Muhammad ;
Khan, Adil Mehmood ;
Mazzara, Manuel ;
Distefano, Salvatore ;
Ali, Mohsin ;
Sarfraz, Muhammad Shahzad .
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS, 2022, 19
[2]   Kernel-based methods for hyperspectral image classification [J].
Camps-Valls, G ;
Bruzzone, L .
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 2005, 43 (06) :1351-1362
[3]   A Lightweight Spectral-Spatial Feature Extraction and Fusion Network for Hyperspectral Image Classification [J].
Chen, Linlin ;
Wei, Zhihui ;
Xu, Yang .
REMOTE SENSING, 2020, 12 (09)
[4]   Spectral-Spatial Classification of Hyperspectral Data Based on Deep Belief Network [J].
Chen, Yushi ;
Zhao, Xing ;
Jia, Xiuping .
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING, 2015, 8 (06) :2381-2392
[5]   LiteDepthwiseNet: A Lightweight Network for Hyperspectral Image Classification [J].
Cui, Benlei ;
Dong, Xue-Mei ;
Zhan, Qiaoqiao ;
Peng, Jiangtao ;
Sun, Weiwei .
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 2022, 60
[6]   Lightweight Spectral-Spatial Attention Network for Hyperspectral Image Classification [J].
Cui, Ying ;
Xia, Jinbiao ;
Wang, Zhiteng ;
Gao, Shan ;
Wang, Liguo .
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 2022, 60
[7]   Modified Fisher's linear discriminant analysis for hyperspectral imagery [J].
Du, Qian .
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS, 2007, 4 (04) :503-507
[8]   A New Spatial-Spectral Feature Extraction Method for Hyperspectral Images Using Local Covariance Matrix Representation [J].
Fang, Leyuan ;
He, Nanjun ;
Li, Shutao ;
Plaza, Antonio J. ;
Plaza, Javier .
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 2018, 56 (06) :3534-3546
[9]  
Howard AG, 2017, Arxiv, DOI [arXiv:1704.04861, DOI 10.48550/ARXIV.1704.04861]
[10]   Hyperspectral Image Classification Using a Hybrid 3D-2D Convolutional Neural Networks [J].
Ghaderizadeh, Saeed ;
Abbasi-Moghadam, Dariush ;
Sharifi, Alireza ;
Zhao, Na ;
Tariq, Aqil .
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING, 2021, 14 (14) :7570-7588