Noisy Bands Research Articles

Hyperspectral imagery quality assessment and band reconstruction using the prophet model

AbstractIn Hyperspectral Imaging (HSI), the detrimental influence of noise and distortions on data quality is profound, which has severely affected the following‐on analytics and decision‐making such as land mapping. This study presents an innovative framework for assessing HSI band quality and reconstructing the low‐quality bands, based on the Prophet model. By introducing a comprehensive quality metric to start, the authors approach factors in both spatial and spectral characteristics across local and global scales. This metric effectively captures the intricate noise and distortions inherent in the HSI data. Subsequently, the authors employ the Prophet model to forecast information within the low‐quality bands, leveraging insights from neighbouring high‐quality bands. To validate the effectiveness of the authors’ proposed model, extensive experiments on three publicly available uncorrected datasets are conducted. In a head‐to‐head comparison, the framework against six state‐of‐the‐art band reconstruction algorithms including three spectral methods, two spatial‐spectral methods and one deep learning method is benchmarked. The authors’ experiments also delve into strategies for band selection based on quality metrics and the quality evaluation of the reconstructed bands. In addition, the authors assess the classification accuracy utilising these reconstructed bands. In various experiments, the results consistently affirm the efficacy of the authors’ method in HSI quality assessment and band reconstruction. Notably, the authors’ approach obviates the need for manually prefiltering of noisy bands. This comprehensive framework holds promise in addressing HSI data quality concerns whilst enhancing the overall utility of HSI.

Deep hierarchical spectral-spatial feature fusion for hyperspectral image classification based on convolutional neural network

Joint spectral-spatial feature extraction has been proven to be the most effective part of hyperspectral image (HSI) classification. But, due to the mixing of informative and noisy bands in HSI, joint spectral-spatial feature extraction using convolutional neural network (CNN) may lead to information loss and high computational cost. More specifically, joint spectral-spatial feature extraction from excessive bands may cause loss of spectral information due to the involvement of convolution operation on non-informative spectral bands. Therefore, we propose a simple yet effective deep learning model, named deep hierarchical spectral-spatial feature fusion (DHSSFF), where spectral-spatial features are exploited separately to reduce the information loss and fuse the deep features to learn the semantic information. It makes use of abundant spectral bands and few informative bands of HSI for spectral and spatial feature extraction, respectively. The spectral and spatial features are extracted through 1D CNN and 3D CNN, respectively. To validate the effectiveness of our model, the experiments have been performed on five well-known HSI datasets. Experimental results demonstrate that the proposed method outperforms other state-of-the-art methods and achieved 99.17%, 98.84%, 98.70%, 99.18%, and 99.24% overall accuracy on Kennedy Space Center, Botswana, Indian Pines, University of Pavia, and Salinas datasets, respectively.

Minimum Noise Fraction and Long Short-Term Memory Model for Hyperspectral Imaging

In recent years, deep learning techniques have presented a major role in hyperspectral image (HSI) classification. Most commonly Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) has greatly advanced the accuracy of hyperspectral image classification, making it powerful tool for remote sensing applications. Deep structure learning, which involves multiple layers of neural network, has shown promising results in effectively addressing nonlinear problems and improving classification accuracy and reduce execution time. The exact categorization of ground topographies from hyperspectral data is a crucial and current research topic that has gotten a lot of attention. This research work focuses on hyperspectral image categorization utilizing several machine learning approaches such as support vector machine (SVM), K-Nearest Neighbour (KNN), CNN and LSTM. To reduce the number of superfluous and noisy bands in the dataset, Principal Component Analysis (PCA) and Minimum Noise Fraction (MNF) were utilized. Different performance evaluation measures like time taken for testing, classification accuracy, kappa accuracy, precision, recall, specificity, F1_score, and Gmean have been taken to prove the efficacy of the models. Based on the simulation results, it is observed that the LSTM model outperforms the other models in terms of accuracy percentage and time consumption, making it the most effective model for classifying hyperspectral imaging datasets.

Noisy band selection based on the integration of the Stacked-Autoencoder and Convolutional Neural Network approaches for hyperspectral data.

The presence of noise on hyperspectral images causes degradation and hinders efficiency of processing for land cover classification. In this sense, removing noise or detecting noisy bands automatically on hyperspectral images becomes a challenge for research in remote sensing. To cope this problem, an integrated model (SAE-1DCNN) is presented in this study, based on Stacked-Autoencoders (SAE) and Convolutional Neural Networks (CNN) algorithms for the selection and exclusion of noisy bands. The proposed model employs convolutional layers to improve the performance of autoencoders focused on discriminating the training data by analyzing the hyperspectral signature of the pixel. Thus, in the SAE-1DCNN model, information can be compressed, and then redundant information can be detected and extracted by taking advantage of the efficiency of the deep architecture based on the convolutional and pooling layers. Hyperspectral data from the AVIRIS (Airborne Visible/Infrared Imaging Spectrometer) sensor were used to evaluate the performance of the proposed automatic method based on feature selection. The results showed effectiveness to identify noisy bands automatically, suggesting that the proposed methodology was found to be promising and can be an alternative to identify noisy bands within the scope of hyperspectral data pre-processing.
 Keywords: noisy bands; feature selection; convolutional neural network; stacked-autoencoders; hyperspectral data

DSSFN: A Dual-Stream Self-Attention Fusion Network for Effective Hyperspectral Image Classification

Hyperspectral images possess a continuous and analogous spectral nature, enabling the classification of distinctive information by analyzing the subtle variations between adjacent spectra. Meanwhile, a hyperspectral dataset includes redundant and noisy information in addition to larger dimensions, which is the primary barrier preventing its use for land cover categorization. Despite the excellent feature extraction capability exhibited by convolutional neural networks, its efficacy is restricted by the constrained receptive field and the inability to acquire long-range features due to the limited size of the convolutional kernels. We construct a dual-stream self-attention fusion network (DSSFN) that combines spectral and spatial information in order to achieve the deep mining of global information via a self-attention mechanism. In addition, dimensionality reduction is required to reduce redundant data and eliminate noisy bands, hence enhancing the performance of hyperspectral classification. A unique band selection algorithm is proposed in this study. This algorithm, which is based on a sliding window grouped normalized matching filter for nearby bands (SWGMF), can minimize the dimensionality of the data while preserving the corresponding spectral information. Comprehensive experiments are carried out on four well-known hyperspectral datasets, where the proposed DSSFN achieves higher classification results in terms of overall accuracy (OA), average accuracy (AA), and kappa than previous approaches. A variety of trials verify the superiority and huge potential of DSSFN.

Potential Assessment of PRISMA Hyperspectral Imagery for Remote Sensing Applications

Hyperspectral imagery plays a vital role in precision agriculture, forestry, environment, and geological applications. Over the past decade, extensive research has been carried out in the field of hyperspectral remote sensing. First introduced by the Italian Space Agency ASI in 2019, space-borne PRISMA hyperspectral imagery (PHSI) is taking the hyperspectral remote sensing research community into the next era due to its unprecedented spectral resolution of ≤12 nm. Given these abundant free data and high spatial resolution, it is crucial to provide remote sensing researchers with information about the critical attributes of PRISMA imagery, making it the most viable solution for various land and water applications. Hence, in the present study, a SWOT analysis was performed for PHSI using recent case studies to exploit the potential of PHSI for different remote sensing applications, such as snow, soil, water, natural gas, and vegetation. From this analysis, it was found that the higher reflectance spectra of PHSI, which have comprehensive coverage, have greater potential to extract vegetation biophysical parameters compared to other applications. Though the possible use of these data was demonstrated in a few other applications, such as the identification of methane gases and soil mineral mapping, the data may not be suitable for continuous monitoring due to their limited acquisition, long revisiting times, noisy bands, atmospheric interferences, and computationally heavy processing, particularly when executing machine learning models. The potential applications of PHSI include large-scale and efficient mapping, transferring technology, and fusion with other remote sensing data, whereas the lifetime of satellites and the need for interdisciplinary personnel pose challenges. Furthermore, some strategies to overcome the aforementioned weaknesses and threats are described in our conclusions.

MR-Selection: A Meta-Reinforcement Learning Approach for Zero-Shot Hyperspectral Band Selection

Band selection is an effective method to deal with the difficulties in image transmission, storage, and processing caused by redundant and noisy bands in hyperspectral images (HSIs). Existing band selection methods usually need to learn a specific model for each HSI dataset, which ignores the inherent correlation and common knowledge among different band selection tasks. Meanwhile, these methods lead to a huge waste of computation. In this article, a novel zero-shot band selection method, called MR-Selection, is proposed for HSI classification. It formalizes zero-shot band selection as a metalearning problem, where advantage actor–critic algorithm-based reinforcement learning (A2C-RL) is designed to extract the metaknowledge in the band selection tasks of various seen hyperspectral datasets through a shared agent. To learn a consistent representation among different tasks, a dynamic structure-aware graph convolutional network is constructed to build a shared agent in A2C-RL. In A2C-RL, the state is tailored in a feasible way and easy to adapt to various tasks. Meanwhile, the reward is defined according to an efficient evaluation network, which can evaluate each state effectively without any fine-tuning. Furthermore, a two-stage optimization strategy is designed to coordinate optimization directions of a shared agent from different tasks effectively. Once the shared agent is optimized, it can be directly applied to unseen HSI band selection tasks without any available samples. Experimental results demonstrate the effectiveness and efficiency of the MR-Selection on the band selection of unseen HSI datasets.

Multiscale Superpixelwise Prophet Model for Noise-Robust Feature Extraction in Hyperspectral Images

Despite of various approaches proposed to smooth the hyperspectral images (HSIs) before feature extraction, the efficacy is still affected by the noise, even using the corrected dataset with the noisy and water absorption bands discarded. In this study, a novel spectral-spatial feature mining framework, Multiscale Superpixelwise Prophet Model (MSPM), is proposed for noise-robust feature extraction and effective classification of the HSI. The prophet model is highly noise-robust for deeply digging into the complex structured features thus enlarging interclass diversity and improving intraclass similarity. First, the superpixelwise segmentation is produced from the first three principal components of an HSI to group pixels into regions with adaptively determined sizes and shapes. A multiscale prophet model is utilized to extract the multiscale informative trend components from the average spectrum of each superpixel. Taking the multiscale trend signal as the input feature, the HSI data are classified superpixelwisely, which is further refined by a majority vote based decision fusion. Comprehensive experiments on three publicly available datasets have fully validated the efficacy and robustness of our MSPM model when benchmarked with eleven state-of-the-art algorithms, including six spectral-spatial methods and five deep learning ones. Besides, MSPM also shows superiority under limited training samples, due to the combined strategies of superpixelwise fusion and multiscale fusion. Our model has provided a useful solution for noise-robust feature extraction as it achieves superior HSI classification even from the uncorrected dataset without prefiltering the water absorption and noisy bands.

Feature extraction and classification of hyperspectral imaging using minimum noise fraction and deep convolutional neural network

Remote sensing technology is constantly developing, which has greatly expanded the use of hyperspectral imaging (HSI). In the arena of remote sensing, the classification of HSI has become a challenging topic. The unique properties of hyperspectral data make accurate categorization difficult. In recent years, deep structured learning has emerged as an effective feature extraction technique for effectively addressing nonlinear problems. It is now extensively used to solve a variety of image processing problems. Deep learning is used to classify images and has shown good performance in recent years. The exact categorization of ground topographies from hyperspectral data is a crucial and current research topic that has gotten a lot of attention. This research work focuses on HSI categorization utilizing several machine learning approaches, such as support vector machine, K-nearest neighbor, and convolutional neural network (CNN). To reduce the number of superfluous and noisy bands in the dataset, principal component analysis and minimum noise fraction (MNF) were utilized. Different performance evaluation measures, such as time taken for testing, classification accuracy, kappa accuracy, precision, recall, specificity, F1_score, and G-mean, were taken to prove the efficacy of the models. The simulation results show that the combination of MNF and CNN produces better classification accuracy compared with the other considered models.

Detection of wheat Fusarium head blight using UAV-based spectral and image feature fusion.

Infection caused by Fusarium head blight (FHB) has severely damaged the quality and yield of wheat in China and threatened the health of humans and livestock. Inaccurate disease detection increases the use cost of pesticide and pollutes farmland, highlighting the need for FHB detection in wheat fields. The combination of spectral and spatial information provided by image analysis facilitates the detection of infection-related damage in crops. In this study, an effective detection method for wheat FHB based on unmanned aerial vehicle (UAV) hyperspectral images was explored by fusing spectral features and image features. Spectral features mainly refer to band features, and image features mainly include texture and color features. Our aim was to explain all aspects of wheat infection through multi-class feature fusion and to find the best FHB detection method for field wheat combining current advanced algorithms. We first evaluated the quality of the two acquired UAV images and eliminated the excessively noisy bands in the images. Then, the spectral features, texture features, and color features in the images were extracted. The random forest (RF) algorithm was used to optimize features, and the importance value of the features determined whether the features were retained. Feature combinations included spectral features, spectral and texture features fusion, and the fusion of spectral, texture, and color features to combine support vector machine, RF, and back propagation neural network in constructing wheat FHB detection models. The results showed that the model based on the fusion of spectral, texture, and color features using the RF algorithm achieved the best performance, with a prediction accuracy of 85%. The method proposed in this study may provide an effective way of FHB detection in field wheat.

Non-Negative Matrix Factorization Based on Smoothing and Sparse Constraints for Hyperspectral Unmixing.

Hyperspectral unmixing (HU) is a technique for estimating a set of pure source signals (end members) and their proportions (abundances) from each pixel of the hyperspectral image. Non-negative matrix factorization (NMF) can decompose the observation matrix into the product of two non-negative matrices simultaneously and can be used in HU. Unfortunately, a limitation of many traditional NMF-based methods, i.e., the non-convexity of the objective function, may lead to a sub-optimal solution. Thus, we put forward a new unmixing method based on NMF under smoothing and sparse constraints to obtain a better solution. First, considering the sparseness of the abundance matrix, a weight sparse regularization is introduced into the NMF model to ensure the sparseness of the abundance matrix. Second, according to the similarity prior of the same feature in the adjacent pixels, a Total Variation regularization is further added to the NMF model to improve the smoothness of the abundance map. Finally, the signatures of each end member are modified smoothly in spectral space. Moreover, it is noticed that discontinuities may emerge due to the removal of noisy bands. Therefore, the spectral data are piecewise smooth in spectral space. Then, in this paper, a piecewise smoothness constraint is further applied to each column of the end-member matrix. Experiments are conducted to evaluate the effectiveness of the proposed method based on two different datasets, including a synthetic dataset and the real-life Cuprite dataset, respectively. Experimental results show that the proposed method outperforms several state-of-the-art HU methods. In the Cuprite hyperspectral dataset, the proposed method’s Spectral Angle Distance is 0.1694, compared to the TV-RSNMF method’s 0.1703, L1/2NMF method’s 0.1925, and VCA-FCLS method’s 0.1872.

Graph regularized spatial–spectral subspace clustering for hyperspectral band selection

Hyperspectral band selection, which aims to select a small number of bands to reduce data redundancy and noisy bands, has attracted widespread attention in recent years. Many effective clustering-based band selection methods have been proposed to accomplish the band selection task and have achieved satisfying performance. However, most of the previous methods reshape the original hyperspectral images (HSIs) into a set of stretched band vectors, which ignore the spatial information of HSIs and the difference between diverse regions. To address these issues, a graph regularized spatial–spectral subspace clustering method for hyperspectral band selection is proposed in this paper, referred to as GRSC. Specifically, the proposed method adopts superpixel segmentation to preserve the spatial information of HSIs by segmenting their first principal component into diverse homogeneous regions. Then the discriminative latent features are generated from each segmented region to represent the whole band, which can mitigate the effect of noise on the band selection. Finally, a self-representation subspace clustering model and an l2,1-norm regularization are utilized to explore the spectral correlation among all bands. In addition, a similarity graph between region-aware latent features is adaptively learned to preserve the spatial structure of HSIs in the latent representation space. Extensive classification experimental results on three public datasets verify the effectiveness of GRSC over several state-of-the-art methods. The demo code of this work is publicly available at https://github.com/WangJun2023/GRSC.

Spatial-Spectral Extraction for Hyperspectral Anomaly Detection

A novel hyperspectral anomaly detection method is proposed in this letter. This method is motivated by two important observations. First, there are hundreds of bands in a hyperspectral image (HSI), among which redundant bands and some noisy bands could be eliminated to increase the discriminability from the spectral domain. Meanwhile, considering the small amount of the anomalies, they may be neglected in the spatial clustering process, and the anomalies can be highlighted by eliminating the clustered result from the original HSI. In this way, the spatial-spectral extraction is proposed to detect the anomalies in the HSI. First, an iterative optimal neighborhood reconstruction (IONR) method is utilized to select the bands while preserving their divergence. Second, the spatial clustering is achieved by minimizing the intradistance in any given category. Finally, the spatial clustered HSI is eliminated from the band selected HSI and acts as the direct input for the detection process. Experimental results and data analysis have demonstrated the effectiveness of the proposed method.

A Band Selection Method With Masked Convolutional Autoencoder for Hyperspectral Image

Band selection (BS) is an effective means to solve the problems of spectral redundancy and Hughes phenomenon in hyperspectral images (HSIs). However, existing BS methods fail to take into account the representativeness, redundancy, and information content of the selected bands simultaneously, and most of them lack consideration of the inherent nonlinear relationship between bands. To address these problems, we propose a novel unsupervised BS framework that can comprehensively consider band representativeness, redundancy, and information content (RRI) in this letter. The band representativeness is estimated by a three-dimensional convolutional autoencoder, which can capture the inherent nonlinear relationship between the bands and leverage the spatial information of the HSI. The redundancy and the information content of a band subset are restricted and enhanced by the correlation coefficient and the information divergence, respectively. Subsequently, RRI combines these three indicators as the subset evaluation criterion and utilizes immune clone selection algorithm to search for the desired band subset. Experimental results verify that the proposed RRI method can provide higher classification accuracy than the competitors and is robust to noisy bands.

Location Soft-Aggregation-Based Band Weighting for Hyperspectral Image Classification

Hyperspectral images (HSIs) comprise hundreds of continuous spectral bands. How to effectively exploit the abundant spectral features of HSI to improve its classification accuracy is the focus of the research. Band weighting (BW) is extensively used due to its ability to emphasize usefully and suppress noisy bands adaptively. Most proposed works aggregate global information to construct band representation vectors in simple ways such as global averaging pooling. Those ways are not capable of retaining a more discriminating feature. Furthermore, modeling for interpixel positional relationships is something they have not considered. To address these problems, we propose a position embedding and importance aggregation BW module. The position embedding section encodes the position information by two 1-D features so that remote dependencies in one spatial direction can be obtained while retaining accurate position information in the other spatial direction. The importance aggregation section aggregates the global information. Finally, a group of weights is learned to recalibrate the raw input. Experiments on three public datasets of HSI demonstrate that our methods obtain competitive results compared to other methods.

Spectral denoising in hyperspectral imaging using the discrete wavelet transform

The use of hyperspectral sensors has gained relevance in agriculture due to its potential in the phytosanitary management of crops. However, these sensors are sensitive to spectral noise, which makes their real application difficult. Therefore, this work focused on the analysis of the spectral noise present in a bank of 180 hyperspectral images of mango leaves acquired in the laboratory, and the implementation of a denoising technique based on the discrete wavelet transform. The noise analysis consisted in the identification of the highest noisy bands, while the performance of the technique was based on the PSNR and SNR metrics. As a result, it was determined that the spectral noise was present at the ends of the spectrum (417-421nm and 969-994nm) and that the Neigh-Shrink method achieved a SNR of the order of 1011 with respect to the order of 102 of the original spectrum.

Identify Informative Bands for Hyperspectral Target Detection Using the Third-Order Statistic

Constrained energy minimization (CEM) has been proposed and widely researched in the field of hyperspectral target detection. Generally, it selects one of the target spectra as the representative and then keeps its output constant while minimizing the average filter output energy of the data. However, it has been proven that as the number of bands (L) increases, CEM will gradually lower the average filter output energy when keeping the representative’s output constant. Unavoidably, due to the inherent spatial and temporal variation of the spectra, this will lead to an unreasonable phenomenon, i.e., if L is particularly large, when adding more bands, CEM will suppress more and more pixels, even including the target pixels. This means that the optimal solution of CEM may not correspond to the target detection result that we desire. To deal with this, in this paper, we introduce the third-order statistic (skewness) of the CEM model, served as an auxiliary index to determine whether a band is beneficial to target detection or not. Theoretically, we prove that the skewness index can always exclude the noisy bands with Gaussian distribution. In addition, experiments on several widely used remote sensing data indicate that the index can also efficiently identify informative bands for a better target detection performance.

A Self-Supervised Denoising Network for Satellite-Airborne-Ground Hyperspectral Imagery

Hyperspectral images (HSIs) are inevitably corrupted with various types of noise, which seriously degrades the data quality and usability. Denoising is an essential preprocessing task of HSI processing. Recently, benefiting from the great learning ability of deep learning, convolutional neural network (CNN) denoisers have obtained state-of-the-art performances for Gaussian noise removal. However, one central problem remains largely unsolved: how to deal with the complicated noise in the real-world HSIs, especially when a paired training data set is unavailable. In this article, a self-supervised hyperspectral image denoising network (SHDN) is proposed, which consists of a noise estimator and a CNN denoiser. Rather than defining a complex noise model to generate training pairs on the clean HSIs, a self-supervised training scheme is first proposed by considering the noisy HSI itself as the training data. Through the noise estimator, the realistic noise samples can be extracted and combined with the clean bands to make up the training pairs. In addition, to jointly restore the target noisy band and to maintain the spectral consistency, a flexible multi-to-single band convolutional network is designed, where the noisy band and the neighboring bands are jointly aggregated via multiscale contextualized dilated blocks and the spectral–spatial convolutional unit. Experiments on HSIs from spaceborne, airborne, unmanned aerial vehicle (UAV)-borne, and ground-based data sets demonstrate the applicability and the generalization of SHDN in the real scenarios. Additionally, the usability of the noisy bands and the suitability of the SHDN framework in the subsequent applications are verified in the land-cover mapping experiments.

Representativeness and Redundancy-Based Band Selection for Hyperspectral Image Classification`

ABSTRACT Hyperspectral band selection (BS) aims to select a subset of bands from the original image cube for subsequent tasks, such as pixel classification. In this paper, we propose a novel unsupervised BS method, termed the representativeness and redundancy-based BS (RRBS) method, by measuring the representativeness and redundancy of the selected bands. The intuitive motivation is to find a subset of bands, which represents the dataset and has low redundancy. The desired bands are obtained sequentially. In each round of lookup, two novel selection criteria based on orthogonal subspace projection are designed for searching the bands that not only well represent the unselected bands but also lowly correlate with the selected bands. Additionally, kernel tricks are used to develop a nonlinear version of the linear selection criteria. Both the linear and nonlinear selection criteria can explicitly evaluate the representativeness and redundancy simultaneously, and they are also robust to noisy bands. The experimental results verify that the proposed method yields excellent classification performance even selecting very a limited number of bands.

Multiscale 2-D Singular Spectrum Analysis and Principal Component Analysis for Spatial–Spectral Noise-Robust Feature Extraction and Classification of Hyperspectral Images

In hyperspectral images (HSI), most feature extraction and data classification methods rely on corrected dataset, in which the noisy and water absorption bands are removed. This can result in not only extra working burden but also information loss from removed bands. To tackle these issues, in this article, we propose a novel spatial–spectral feature extraction framework, multiscale 2-D singular spectrum analysis (2-D-SSA) with principal component analysis (PCA) (2-D-MSSP), for noise-robust feature extraction and data classification of HSI. First, multiscale 2-D-SSA is applied to exploit the multiscale spatial features in each spectral band of HSI via extracting the varying trends within defined windows. Taking the extracted trend signals at each scale level as the input features, the PCA is employed to the spectral domain for dimensionality reduction and spatial–spectral feature extraction. The derived spatial–spectral features in each scale are separately classified and then fused at decision-level for efficacy. As our 2-D-MSSP method can extract features and simultaneously remove noise in both spatial and spectral domains, which ensures it to be noise-robust for classification of HSI, even the uncorrected dataset. Experiments on three publicly available datasets have fully validated the efficacy and robustness of the proposed approach, when benchmarked with 10 state-of-the-art classifiers, including six spatial–spectral methods and four deep learning classifiers. In addition, both quantitative and qualitative assessment has validated the efficacy of our approach in noise-robust classification of HSI even with limited training samples, especially in classifying uncorrected data without filtering noisy bands.