A Hybrid Capsule Network for Hyperspectral Image Classification
Open Access
- 9 November 2021
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 14 (19391404) , 11824-11839
Abstract
Limited training data, high dimensionality, image complexity, and similarity between classes are challenges confronting hyperspectral image (HSI) classification often resulting in suboptimal classification performance. The capsule network (CapsNet) preserves the hierarchy between different parts of the entity in an image by replacing scalar representations with vectors and can address these aforementioned issues. Motivated by CapsNet, this article presents a novel end-to-end deep learning (DL) architecture, the hybrid capsule network (HCapsNet), for HSI classification. HCapsNet employs 2-D and 3-D convolutional neural networks (CNNs) to extract higher level spatial and spectral features. In order to establish a route between capsules in the lower layers to the most-related capsule in the higher layer, dynamic routing (DR) is used to identify several overlapped objects during training sessions. Hyperparameter optimization is performed using nested cross-validation (nested-CV) to ensure thorough generalization evaluation. The proposed HCapsNet significantly outperformed the state-of-the-art methods in terms of overall classification accuracy on three widely used hyperspectral datasets, Indian Pines dataset achieving ( $>3\%$ , $p< {1}\times 1\times 10^{-11}$ ), the University of Pavia dataset ( $>4\%$ , $p< {1}\times 1\times 10^{-9}$ ), the Salinas Valley dataset ( $>3\%$ , $p< {1}\times 1\times 10^{-10}$ ) when using only 1% of the data for training. The performance of all CNN-based approaches degraded significantly with smaller training sample sizes. HCapsNet, therefore, is demonstrated to offer significant advantages in HSI classification problems with low sample sizes.Keywords
Funding Information
- UKRIResearch Council projects
This publication has 60 references indexed in Scilit:
- Unsupervised Joint Feature Learning and Encoding for RGB-D Scene LabelingIEEE Transactions on Image Processing, 2015
- Deep learningNature, 2015
- Reducing Leaf-Level Hyperspectral Data to 22 Components of Biochemical and Biophysical Bands Optimizes Tree Species DiscriminationIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2015
- Hyperspectral Image Classification Based on Three-Dimensional Scattering Wavelet TransformIEEE Transactions on Geoscience and Remote Sensing, 2014
- Spectral–Spatial Classification of Hyperspectral Data Using Local and Global Probabilities for Mixed Pixel CharacterizationIEEE Transactions on Geoscience and Remote Sensing, 2014
- Band Selection for Hyperspectral Image Classification Using Mutual InformationIEEE Geoscience and Remote Sensing Letters, 2006
- Reducing the Dimensionality of Data with Neural NetworksScience, 2006
- A Fast Learning Algorithm for Deep Belief NetsNeural Computation, 2006
- On the Impact of PCA Dimension Reduction for Hyperspectral Detection of Difficult TargetsIEEE Geoscience and Remote Sensing Letters, 2005
- Gradient-based learning applied to document recognitionProceedings of the IEEE, 1998