Convolutional neural network (CNN) has been successfully introduced to hyperspectral image (HSI) classification and achieved effective performance. With the depth of the CNN increases, it may cause the gradient to become zero, and the structure lacks the utilization of the correlated spatial feature information between different convolutional layers. At the same time, this single-scale convolution kernel is insufficient in expressing the complex spatial structure information of HSI. Additionally, the CNN-based methods treat the HSIs spectral band data as a disordered vector in the process of feature extraction, which abandons the exploitation of its internal spectral correlations. To address these issues, we propose a novel spectral-spatial network classification framework based on multiscale dense connected convolutional network (DenseNet) and bi-direction recurrent neural network (Bi-RNN) with attention mechanism network (MDRN). For the proposed MDRN, in terms of spatial feature extraction, a multi-scale DenseNet is exploited to combine shallow and deep convolution features to extract the multi-scale and complex spatial structure features at each layer. In the aspects of spectral feature extraction, Bi-RNN with attention mechanism is used to capture the inner spectral correlations within a continuous spectrum. Three standard real hyperspectral datasets were used to verify the effectiveness of the proposed MDRN approach. Experimental results indicate that the proposed MDRN method can make full use of the spectral and spatial information of the image, and it has better performance than some advanced algorithms in HSI classification. Finally, in the application of hyperspectral data captured by Gaofen-5 (GF-5) satellite, the practicability of the proposed MDRN method is also superior to other methods.

No abstract

With the depth of the Convolutional neural network(CNN) increases, CNN may lead to the problem of gradient disappearance. Simultaneously, single scale convolutional kernel may not reflect the complex spatial structural information in hyperspectral image(HSI). In addition, the CNN based approach regards the spectral band data on a single pixel of the HSI as a disordered high dimensional vector for processing, which does not meet the characteristics of the spectral band data. To tackle these aforementioned issues, a novel classification approach based on multi-scale densely connected convolutional network(Densenet) and bi-direction recurrent neural network(Bi-RNN) with attention framework is introduced in this study. Specifically, multi-scale Densenet is exploited to fully extract the multiple scales complex spatial structural information and utilize the strong complementary yet correlated spatial feature information between convolution layers, and Bi-RNN with attention is designed to obtain inner spectral correlations within a continuous spectrum. For comparison and verifying the effectiveness of our proposed method, we test the proposed method with nine other recently proposed methods on Salinas dataset, and the experimental results demonstrate that the proposed method can sufficiently exploit spectral and spatial information and outperforms other competitive methods.

Traditional convolutional neural networks (CNNs) can be applied to obtain the spectral-spatial feature information from hyperspectral images (HSIs). However, they often introduce significant redundant spatial feature information. The octave convolution network is frequently utilized instead of traditional CNN to decrease spatial redundant information of the network and extend its receptive field. However, the 3D octave convolution-based approaches may introduce extensive parameters and complicate the network. To solve these issues, we propose a new HSI classification approach with a multi-scale spectral-spatial network-based framework that combines 2D octave and 3D CNNs. Our method, called MOCNN, first utilizes 2D octave convolution and 3D DenseNet branch networks with various convolutional kernel sizes to obtain complex spatial contextual feature information and spectral characteristics, separately. Moreover, the channel and the spectral attention mechanisms are, respectively, applied to these two branch networks to emphasize significant feature regions and certain important spectral bands that comprise discriminative information for the categorization. Furthermore, a sample balancing strategy is applied to address the sample imbalance problem. Expansive experiments are undertaken on four HSI datasets, demonstrating that our MOCNN approach outperforms several other methods for HSI classification, especially in scenarios dominated by limited and imbalanced sample data.