A Novel Cubic Convolutional Neural Network for Hyperspectral Image Classification

Wang, Jinwei; Song, Xiao; Sun, Le; Huang, Wei; Wang, Jin

doi:10.1109/jstars.2020.3008949

Cited by 32 publications

(15 citation statements)

References 49 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Furthermore, Roy et al [13] combined convolution kernels with generative adversarial minority oversampling to enhance the model performance by addressing the imbalanced data challenge imposed by HSI classification. Wang et al [14] proposed an end-to-end cubic CNN, which applies convolutions in different directions of the feature volume to fully exploit spatial and spatial-spectral features. Driven by the goal of extracting and exploiting the best possible features, Alipour et al [15] and Roy et al [16] explored new architectural designs to make the convolutional kernel more flexible.…”

Section: Introductionmentioning

confidence: 99%

Morphological Convolutional Neural Networks for Hyperspectral Image Classification

Roy¹,

Mondal

Paoletti

et al. 2021

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

View full text Add to dashboard Cite

Convolutional neural networks (CNNs) have become quite popular for solving many different tasks in remote sensing data processing. The convolution is a linear operation which extracts features from the input data. However, nonlinear operations are able to better characterize the internal relationships and hidden patterns within complex remote sensing data, such as hyperspectral images (HSIs). Morphological operations are powerful nonlinear transformations for feature extraction that preserve the essential characteristics of the image, such as borders, shape and structural information. In this paper, a new end-to-end morphological deep learning framework (called MorphConvHyperNet) is introduced. The proposed approach efficiently models nonlinear information during the training process of HSI classification. Specifically our method includes spectral and spatial morphological blocks to extract relevant features from the HSI input data. These morphological blocks consist of two basic 2D morphological operators (erosion and dilation) in the respective layers, followed by a weighted combination of the feature maps. Both layers can successfully encode the nonlinear information related to shape and size, playing an important role in classification performance. Our experimental results, obtained on five widely used HSIs, reveal that our newly proposed MorphConvHyperNet offers comparable (and even superior) performance when compared to traditional 2D and 3D

show abstract

Section: Introductionmentioning

confidence: 99%

Morphological Convolutional Neural Networks for Hyperspectral Image Classification

Roy¹,

Mondal

Paoletti

et al. 2021

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

View full text Add to dashboard Cite

show abstract

“…However, in actual scenarios, due to the complex nonlinear mixture, the ideal value of parameter p is difficult to guarantee. Therefore, in our experiments, the value of p is fixed to 4. we suggest choosing the optimal value of p in the range of [3][4][5][6][7][8].…”

Section: The Dimension Of Subspace Pmentioning

confidence: 99%

“…However, the existing mixed noise generated during the process of digital imaging, e.g., Gaussian noise, salt and pepper, stripes and deadlines, seriously diminishes the accuracy of the above applications [2]. In addition, a series of HSI processing applications (e.g., unmixing [3,4], classification [5][6][7], super-resolution [8], target detection [9,10]) are greatly dependent on the imaging quality of HSI. Therefore, as a key preprocessing step of the above-mentioned applications, the studies on HSI mixed denoising methods have become a hot research topic and aroused widespread interest in the field.…”

Section: Introductionmentioning

confidence: 99%

Spectral-Smoothness and Non-Local Self-Similarity Regularized Subspace Low-Rank Learning Method for Hyperspectral Mixed Denoising

Liu

Sun

2021

Remote Sensing

Self Cite

View full text Add to dashboard Cite

During the imaging process, hyperspectral image (HSI) is inevitably affected by various noises, such as Gaussian noise, impulse noise, stripes or deadlines. As one of the pre-processing steps, the removal of mixed noise for HSI has a vital impact on subsequent applications, and it is also one of the most challenging tasks. In this paper, a novel spectral-smoothness and non-local self-similarity regularized subspace low-rank learning (termed SNSSLrL) method was proposed for the mixed noise removal of HSI. First, under the subspace decomposition framework, the original HSI is decomposed into the linear representation of two low-dimensional matrices, namely the subspace basis matrix and the coefficient matrix. To further exploit the essential characteristics of HSI, on the one hand, the basis matrix is modeled as spectral smoothing, which constrains each column vector of the basis matrix to be a locally continuous spectrum, so that the subspace formed by its column vectors has continuous properties. On the other hand, the coefficient matrix is divided into several non-local block matrices according to the pixel coordinates of the original HSI data, and block-matching and 4D filtering (BM4D) is employed to reconstruct these self-similar non-local block matrices. Finally, the formulated model with all convex items is solved efficiently by the alternating direction method of multipliers (ADMM). Extensive experiments on two simulated datasets and one real dataset verify that the proposed SNSSLrL method has greater advantages than the latest state-of-the-art methods.

show abstract

“…Moreover, the reason why we employ hybrid CNN-RNN sequence structure as the basic structure depends on these factors. Primarily, 1-D CNNs have been proved to achieve superior performance in terms of dealing with sequence data sets, which are capable of extracting middle-level, locally invariant and discriminative features from the input spectrum [56]. However, 1-D CNNs might be insensitive about the location information of spectral features.…”

Section: B Iops Estimation Networkmentioning

confidence: 99%

Hybrid Sequence Networks for Unsupervised Water Properties Estimation From Hyperspectral Imagery

Xue

Gong³

et al. 2021

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

View full text Add to dashboard Cite

Generating accurate estimation of water inherent optical properties (IOPs) from hyperspectral images plays a significant role in marine exploration. Traditional methods mainly adopt bathymetric models and numerical optimization algorithms to deal with this problem. However, these methods usually tend to simplify the bathymetric models and lack the capability of describing the discrepancy between reference spectrum and estimation spectrum, resulting in a limited estimation performance. To get a more precise result, in this work we propose a novel network based on deep learning to retrieve the IOPs. The proposed network, named as IOPs estimation network (IOPE-Net), explores a hybrid sequence structure to establish IOPs estimation module that acquires high-dimensional nonlinear features of water body spectrums for water IOPs estimation. Moreover, considering the insufficiency of labelled training samples, we design a spectrum reconstruction module combined with classical bathymetric model to train the proposed network in an unsupervised manner. Then, aiming at further promoting the estimation performance, a multi-criterion loss is developed as the objective function of IOPE-Net. In particular, we construct a hierarchical multiscale sequence loss as the key component to retain the details of original spectral information. Thus, the discrepancy between different spectrums can be better described by the obtained learning model. Experimental results on both simulated and real data sets demonstrate the effectiveness and efficiency of our method in comparison with the state-of-theart water IOPs retrieving methods.

show abstract

A Novel Cubic Convolutional Neural Network for Hyperspectral Image Classification

Cited by 32 publications

References 49 publications

Morphological Convolutional Neural Networks for Hyperspectral Image Classification

Morphological Convolutional Neural Networks for Hyperspectral Image Classification

Spectral-Smoothness and Non-Local Self-Similarity Regularized Subspace Low-Rank Learning Method for Hyperspectral Mixed Denoising

Hybrid Sequence Networks for Unsupervised Water Properties Estimation From Hyperspectral Imagery

Contact Info

Product

Resources

About