Learning Sensor-Specific Spatial-Spectral Features of Hyperspectral Images via Convolutional Neural Networks

Abstract. Hyperspectral images (HSIs) contain hundreds of spectral bands, providing high-resolution spectral information pertaining to the Earth’s surface. Additionally, abundant spatial contextual information can also be obtained simultaneously from a HSI. To characterize the properties of ground objects, classification is the most widely-used technology in the field of remote sensing, where each pixel in a HSI is assigned to a pre-defined class. Over the past decade, deep learning has attracted increasing attention in the machine-learning and computer-vision domains, due to its favourable performances for various types of tasks, and it has been successfully introduced to the remote-sensing community. Instead of utilizing the shallow features within in a given image, which is the approach that is generally adopted in other conventional classification methods, deep-learning algorithms can extract hierarchical features from raw HSI data. Within the deep-learning framework, recurrent neural networks (RNNs), which are able to encode sequential features, have exhibited promising capabilities and have achieved encouraging performances, especially for the natural-language processing and speech-recognition communities. As multi-temporal remote-sensing images can be readily obtained from increasing numbers of satellite and unmanned aircraft systems, and since analysis of such multi-temporal data comprises a critical issue within numerous research subfields, including land-cover and land-change analyses, and land-resource management, RNNs have been applied in recent studies in order to extract temporal sequential features from multi-temporal remote-sensing images for the purpose of image classification. Apart from using multi-temporal image datasets, RNNs can also be utilized on a single image, where the spectral feature/band of each individual pixel can be taken as a sequential feature for the input layer of RNNs. However, the application of such sequential feature extraction that relies on a single image still needs to be further investigated since applying RNNs to spectral bands will directly introduce more parameters that need to be optimized, consequently increasing the total training time.In this study, we propose a novel RNN-based HSI classification framework. In this framework, unlabelled pixels obtained from a single image are considered when constructing sequential features. Two spatial similarity measurements, referred to as pixel-matching and block-matching, respectively, are employed to extract pixels that are “similar” to the target pixel. Then, the sequential feature of the target pixel is constructed by exploiting several of the most “similar” pixels and ordering them based on their similarities to the target pixel. The aforementioned two schemes are advantageous, as unlabelled pixels within the given HSI are taken into consideration for similarity measurement and sequential feature construction for the RNN model. Moreover, the block-matching scheme also takes advantage of spatial contextual information, which has been widely utilized in spatial-spectral-based HSI classification methods. To evaluate the proposed methods, two benchmark HSIs are used, including a HSI collected over Pavia University, Italy by the airborne Reflective Optics System Imaging Spectrometer (ROSIS) sensor, and an image acquired over the Salinas Valley, California, USA via the Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) sensor. Spatio-temporally coincident ground-reference data accompanies each of these respective HSIs. In addition, the proposed methods are compared with three state-of-the-art algorithms, including support vector machine (SVM), the 1-dimensional convolutional neural network (1DCNN), and the 1-dimensional RNN (1DRNN).Experimental results indicate that our proposed methods achieve markedly better classification performance compared with the baseline algorithms on both datasets. For example, for the Pavia University image, the block-matching based RNN achieves the highest overall classification accuracy, with 94.32% accuracy, which is 9.87% higher than the next most accurate algorithm of the aforementioned three baseline methods, which in this case is the 1DCNN, with 84.45% overall accuracy. More specifically, the block-matching method performs better than the pixel-matching method in terms of both quantitative and qualitative assessments. Based on visual assessment/interpretation of the classification maps, it is apparent that “salt-and-pepper” noise is markedly alleviated; with block-matching, smoother classified images are generated compared with pixel-matching-based methods and the three baseline algorithms. Such results demonstrate the effectiveness of utilizing spatial contextual information in the similarity measurement.

ABSTRACTGraph-based methods are developed to efficiently extract data information. In particular, these methods are adopted for high-dimensional data classification by exploiting information residing on weighted graphs. In this paper, we propose a new hyperspectral texture classifier based on graph-based wavelet transform. This recent graph transform allows extracting textural features from a constructed weighted graph using sparse representative pixels of hyperspectral image. Different measurements of spectral similarity between representative pixels are tested to decorrelate close pixels and improve the classification precision. To achieve the hyperspectral texture classification, Support Vector Machine is applied on spectral graph wavelet coefficients. Experimental results obtained by applying the proposed approach on Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) and Reflective Optics System Imaging Spectrometer (ROSIS) datasets provide good accuracy which could exceed 98.7%. Compared to other famous classification methods as conventional deep learning-based methods, the proposed method achieves better classification performance. Results have shown the effectiveness of the method in terms of robustness and accuracy.

Hyperspectral remote sensing offers a powerful tool in many different application contexts. The imbalance between the high dimensionality of the data and the limited availability of training samples calls for the need to perform dimensionality reduction in practice. Among traditional dimensionality reduction techniques, feature extraction is one of the most widely used approaches due to its flexibility to transform the original spectral information into a subspace. In turn, band selection is important when the application requires preserving the original spectral information (especially the physically meaningful information) for the interpretation of the hyperspectral scene. In the case of hyperspectral image classification, both techniques need to discard most of the original features/bands in order to perform the classification using a feature set with much lower dimensionality. However, the discriminative information that allows a classifier to provide good performance is usually classdependent and the relevant information may live in weak features/bands that are usually discarded or lost through subspace transformation or band selection. As a result, in practice, it is challenging to use either feature extraction or band selection for classification purposes. Relevant lines of attack to address this problem have focused on multiple feature selection aiming at a suitable fusion of diverse features in order to provide relevant information to the classifier. In this paper, we present a new dimensionality reduction technique, called multiple criteria-based spectral partitioning, which is embedded in an ensemble learning framework to perform advanced hyperspectral image classification. Driven by the use of a multiple band priority criteria that is derived from classic band selection techniques, we obtain multiple spectral partitions from the original hyperspectral data that correspond to several band subgroups with much lower spectral dimensionality as compared with the original band set. An ensemble learning technique is then used to fuse the information from multiple features, taking advantage of the relevant information provided by each classifier. Our experimental results with two real hyperspectral images, collected by the reflective optics system imaging spectrometer (ROSIS) over the University of Pavia in Italy and the Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) over the Salinas scene, reveal that our presented method, driven by multiple band priority criteria, is able to obtain better classification results compared with classic band selection techniques. This paper also discusses several possibilities for computationally efficient implementation of the proposed technique using various high-performance computing architectures.

On the use of small training sets for neural network-based characterization of mixed pixels in remotely sensed hyperspectral images

Due to the physical limitations of the imaging devices, hyperspectral images (HSIs) are commonly distorted by a mixture of Gaussian noise, impulse noise, stripes, and dead lines, leading to the decline in the performance of unmixing, classification, and other subsequent applications. In this paper, we propose a novel end-to-end low-rank spatial-spectral network (LR-Net) for the removal of the hybrid noise in HSIs. By integrating the low-rank physical property into a deep convolutional neural network (DCNN), the proposed LR-Net simultaneously enjoys the strong feature representation ability from DCNN and the implicit physical constraint of clean HSIs. Firstly, spatial-spectral atrous blocks (SSABs) are built to exploit spatial-spectral features of HSIs. Secondly, these spatial-spectral features are forwarded to a multi-atrous block (MAB) to aggregate the context in different receptive fields. Thirdly, the contextual features and spatial-spectral features from different levels are concatenated before being fed into a plug-and-play low-rank module (LRM) for feature reconstruction. With the help of the LRM, the workflow of low-rank matrix reconstruction can be streamlined in a differentiable manner. Finally, the low-rank features are utilized to capture the latent semantic relationships of the HSIs to recover clean HSIs. Extensive experiments on both simulated and real-world datasets were conducted. The experimental results show that the LR-Net outperforms other state-of-the-art denoising methods in terms of evaluation metrics and visual assessments. Particularly, through the collaborative integration of DCNNs and the low-rank property, the LR-Net shows strong stability and capacity for generalization.

This study is focused on time optimization for the classification problem presenting a comparison of five Artificial Neural Network Multilayer Perceptron (ANN-MLP) architectures. We use the Artificial Neural Network (ANN) because it allows to recognize patterns in data in a lower time rate. Time and classification accuracy are taken into account together for the comparison. According to time comparison, two paradigms in the computational field for each ANN-MLP architecture are analysed with three schemes. Firstly, sequential programming is applied by using a single CPU core. Secondly, parallel programming is employed over a multi-core CPU architecture. Finally, a programming model running on GPU architecture is implemented. Furthermore, the classification accuracy is compared between the proposed five ANN-MLP architectures and a state-of.the-art Support Vector Machine (SVM) with three classification frames: 50%,60% and 70% of the data set's observations are randomly selected to train the classifiers. Also, a visual comparison of the classified results is presented. The Peak Signal to Noise Ratio (PSNR) and Structural Similarity Index Measure (SSIM) criteria are also calculated to characterise visual perception. The images employed were acquired by the Airborne Visible/Infrared Imaging Spectrometer (AVIRIS), the Reflective Optics System Imaging Spectrometer (ROSIS) and the Hyperion sensor.

Hyperspectral (HS) image is a three dimensional data image where the 3rd dimension carries the wealth of spectrum information. HS image compression is one of the areas that has attracted increasing attention for big data processing and analysis. HS data has its own distinguishing feature which differs with video because without motion, also different with a still image because of redundancy along the wavelength axis. The prediction based method is playing an important role in the compression and research area. Reflectance distribution of HS based on our analysis indicates that there is some nonlinear relationship in intra-band. The Multilayer Propagation Neural Networks (MLPNN) with backpropagation training are particularly well suited for addressing the approximation function. In this paper, an MLPNN based predictive image compression method is presented. We propose a hybrid Adaptive Prediction Mechanism (APM) with MLPNN model (APM-MLPNN). MLPNN is trained to predict the succeeding bands by using current band information. The purpose is to explore whether MLPNN can provide better image compression results in HS images. Besides, it uses less computation cost than a deep learning model so we can easily validate the model. We encoded the weights vector and the bias vector of MLPNN as well as the residuals. That is the only few bytes it then sends to the decoder side. The decoder will reconstruct a band by using the same structure of the network. We call it an MLPNN decoder. The MLPNN decoder does not need to be trained as the weights and biases have already been transmitted. We can easily reconstruct the succeeding bands by the MLPNN decoder. APM constrained the correction offset between the succeeding band and the current spectral band in order to prevent HS image being affected by large predictive biases. The performance of the proposed algorithm is verified by several HS images from Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) reflectance dataset. MLPNN simulation results can improve prediction accuracy; reduce residual of intra-band with high compression ratio and relatively lower bitrates.

In recent studies, superpixel segmentation has been integrated into hyperspectral (HS) image classification methods. However, the existing superpixel-based classification methods usually suffer from two serious problems. First, the accuracy and efficiency of current superpixel segmentation approaches cannot meet the demands of practical applications for HS images; second, conventional superpixel-based classification methods generally consider each generated superpixel as a unit for the image classification, which may help to reduce the computing time but result in a significant decrease of the classification accuracy. To solve the problems, we propose a fast region growing based superpixel segmentation (FRGSS) algorithm and a novel texture-adaptive superpixel integration strategy (TASIS) for the HS image classification. Experimental results on real Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) HS images demonstrate that the proposed FRGSS outperforms the state-of-the-art superpixel algorithm. In addition, the superiority of the TASIS is verified compared to the pixel-wise and the conventional superpixel-based classification methods.

Hyperspectral image (HSI) classification remains a challenging problem due to unique characteristics of HSI data (such as numerous bands and strong correlations in the spectral and spatial domains) and small sample size. To address such concerns, we propose a novel spectral–spatial feature extraction method for HSI classification by employing graph embedding and deep learning (DL) models. Since the conventional graph cannot capture the complex manifold relationship of HSI data, and there exist the observations of within-class variation as well as the similarity between different classes in the spectral domain, we construct the supervised within-class/between-class hypergraph (SWBH) to extract the spectral features of HSI. Since it is difficult for DL models to learn representative features for HSI data when the labeled training samples are limited, we propose the random zero settings to newly generate a large amount of labeled HSI samples for the training of convolutional neural network (CNN). The designed sample expanded CNN (SECNN) is used to extract the HSI spatial features. Thus, the spectral–spatial features of HSI can be learned by integrating the features extracted from SWBH and SECNN, respectively. Experiments on three real HSI datasets demonstrate higher classification accuracy of the proposed SWBH–SECNN method.

The application of graph convolutional networks (GCN) in hyperspectral image (HSI) classification has become a promising method, thanks to its flexible convolution operation in any irregular image region. For the classification of HSI, GCN can extract more superpixel-level features with a topological structure, in comparison to the traditional convolutional neural networks (CNNs) using fixed square kernels distilling pixel-level features. To fully leverage the different levels of features, this study proposes a novel deep network referred to as a CNN-combined graph residual network ( C 2 GRN), which integrates the multilevel graph residual module and spectral-spatial features continuous learning module. During the extraction of topology information using the former module, HSI pixels are divided into superpixels and served as input nodes of the module to reduce the computational complexity and obtain the multilevel spatial relevance between adjacent superpixels. Besides, for the latter module, the spectral-spatial features are learnt continuously, which could obtain the finer pixel-level features. Finally, the captured spectral-spatial features of different levels are concatenated. This strategy could not only adequately utilize the correlation and difference of adjacent spatial but also obtain the finer and more valuable spectral-spatial information, which makes a significant boost in the HSI classification. Additionally, the experiment results demonstrate the superiority and availability of the C 2 GRN on three benchmark datasets of HSI, compared with the state-of-the-art methods for the classification of HSI.

The purpose of this study is to develop hyperspectral imaging (HSI) for automatic detection of head and neck cancer cells on histologic slides. A compact hyperspectral microscopic system is developed in this study. Histologic slides from 15 patients with squamous cell carcinoma (SCC) of the larynx and hypopharynx are imaged with the system. The proposed nuclei segmentation method based on principle component analysis (PCA) can extract most nuclei in the hyperspectral image without extracting other sub-cellular components. Both spectra-based support vector machine (SVM) and patch-based convolutional neural network (CNN) are used for nuclei classification. CNNs were trained with both hyperspectral images and pseudo RGB images of extracted nuclei, in order to evaluate the usefulness of extra information provided by hyperspectral imaging. The average accuracy of spectra-based SVM classification is 68%. The average AUC and average accuracy of the HSI patch-based CNN classification is 0.94 and 82.4%, respectively. The hyperspectral microscopic imaging and classification methods provide an automatic tool to aid pathologists in detecting SCC on histologic slides.

Recently, we have proposed the pairing of Deep Neural Networks and evolutionary algorithms as a versatile route for the optimal design of optical devices and systems.1 Here, we extend that work and investigate the use of deep Convolutional Neural Networks (CNN) as opposed to fully connected feedforward architectures. We show that networks built with convolutional layers, batch normalization (BN), parametric REctified Linear Unit (ReLU) and residual blocks achieve drastic reduction in parameter weights and training epochs in comparison to dense, fully-connected networks for comparable accuracy. The proposed lightweight CNNs when used for approximate objective evaluation in concert with DE global optimization resulted in nearly 8x speedup in convergence time. The choice of a network architecture and optimization of its hyperparameters is a tedious task and it is hoped that the systematic hyperparameter search reported here would assist others in faster model selection.

Data acquired with Advanced Land Imager (ALI), Hyperspectral Imager (Hyperion) and Airborne Visible/Infrared Imaging Spectrometer (AVIRIS), were used to estimate and map forest LAI. Analysis methods include 1) simulating the total at-sensor radiances using MODTRAN4, 2) modifying retrieved surface reflectance with ground spectroradiometric measurements, 3) constructing 6-term LAI prediction models to predict pixel-based LAI, and 4) mapping LAI. The experimental results indicate that the retrieval of surface reflectance is the most successful with AVIRIS, followed by Hyperion and ALI. AVIRIS data can produce more reasonable LAI map than the other two sensors. The results also indicate that Hyperion data have potentially extensive application values in bio-parameter extraction at varied scales.

Fusion of hyperspectral images and lidar-based dems for coastal mapping

Hyperspectral images (HSI) offer detailed spectral reflectance information about sensed objects through provision of information on hundreds of narrow spectral bands. HSI have a leading role in a broad range of applications, such as in forestry, agriculture, geology, and environmental sciences. The monitoring and management of agricultural lands is of great importance for meeting the nutritional and other needs of a rapidly and continuously increasing world population. In relation to this, classification of HSI is an effective way for creating land use and land cover maps quickly and accurately. In recent years, classification of HSI using convolutional neural networks (CNN), which is a sub-field of deep learning, has become a very popular research topic and several CNN architectures have been developed by researchers. The aim of this study was to investigate the classification performance of CNN model on agricultural HSI scenes. For this purpose, a 3D-2D CNN framework and a well-known support vector machine (SVM) model were compared using the Indian Pines and Salinas Scene datasets that contain crop and mixed vegetation classes. As a result of this study, it was confirmed that use of 3D-2D CNN offers superior performance for classifying agricultural HSI datasets.