Hyperspectral Image Classification Using CapsNet With Well-Initialized Shallow Layers

In this letter, an alternative data-driven HSI classification model based on CapsNet is proposed rather than recently predominant convolutional neural network (CNN)-based models. To adjust the CapsNet to HSI classification, we tune a new CapsNet architecture with three convolutional layers. The added shallow layer provides higher level features to the primary capsules, which indirectly speeds up the following routing procedure. To guarantee a good convergence of the whole CapsNet, the three shallow layers are initialized by transferring convolutional parameters from a pretrained CNN model. The improved CapsNet-based models with and without vote strategy both achieve significantly superior performance in HSI classification to the state-of-the-art CNN-based methods on real hyperspectral data sets.

[1]  Lorenzo Bruzzone,et al.  Classification of Hyperspectral Images With Regularized Linear Discriminant Analysis , 2009, IEEE Transactions on Geoscience and Remote Sensing.

[2]  Geoffrey E. Hinton,et al.  Dynamic Routing Between Capsules , 2017, NIPS.

[3]  Xiuping Jia,et al.  Deep Fusion of Remote Sensing Data for Accurate Classification , 2017, IEEE Geoscience and Remote Sensing Letters.

[4]  Sen Jia,et al.  Gabor Feature-Based Collaborative Representation for Hyperspectral Imagery Classification , 2015, IEEE Transactions on Geoscience and Remote Sensing.

[5]  Qian Du,et al.  Classification of Hyperspectral Imagery Using a New Fully Convolutional Neural Network , 2018, IEEE Geoscience and Remote Sensing Letters.

[6]  Joydeep Ghosh,et al.  Investigation of the random forest framework for classification of hyperspectral data , 2005, IEEE Transactions on Geoscience and Remote Sensing.

[7]  Dumitru Erhan,et al.  Going deeper with convolutions , 2014, 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[8]  Lorenzo Bruzzone,et al.  Classification of hyperspectral remote sensing images with support vector machines , 2004, IEEE Transactions on Geoscience and Remote Sensing.

[9]  Xiuping Jia,et al.  Hyperspectral Image Classification Using Joint Sparse Model and Discontinuity Preserving Relaxation , 2018, IEEE Geoscience and Remote Sensing Letters.

[10]  Xiuping Jia,et al.  Deep Feature Extraction and Classification of Hyperspectral Images Based on Convolutional Neural Networks , 2016, IEEE Transactions on Geoscience and Remote Sensing.

[11]  Qian Du,et al.  Hyperspectral Image Classification Using Deep Pixel-Pair Features , 2017, IEEE Transactions on Geoscience and Remote Sensing.

[12]  Jian Sun,et al.  Deep Residual Learning for Image Recognition , 2015, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[13]  Yang Jin,et al.  Capsule Network Performance on Complex Data , 2017, ArXiv.

[14]  Xiuping Jia,et al.  Segment-Oriented Depiction and Analysis for Hyperspectral Image Data , 2017, IEEE Transactions on Geoscience and Remote Sensing.

[15]  Jiang Li,et al.  Dimensionality reduction of hyperspectral data using discrete wavelet transform feature extraction , 2002, IEEE Trans. Geosci. Remote. Sens..

[16]  Jonathan Cheung-Wai Chan,et al.  Learning and Transferring Deep Joint Spectral–Spatial Features for Hyperspectral Classification , 2017, IEEE Transactions on Geoscience and Remote Sensing.

[17]  Trac D. Tran,et al.  Hyperspectral Image Classification Using Dictionary-Based Sparse Representation , 2011, IEEE Transactions on Geoscience and Remote Sensing.

[18]  Ben Somers,et al.  Heathland conservation status mapping through integration of hyperspectral mixture analysis and decision tree classifiers , 2012 .

[19]  Heesung Kwon,et al.  Going Deeper With Contextual CNN for Hyperspectral Image Classification , 2016, IEEE Transactions on Image Processing.