Band Selection Techniques Research Articles

Rapid and high-throughput determination of sorghum (Sorghum bicolor) biomass composition using near infrared spectroscopy and chemometrics

Compositional characterization of biomass is vital for the biofuel industry. Traditional wet chemistry-based methods for analyzing biomass composition are laborious, time-consuming, and require extensive use of chemical reagents as well as highly skilled personnel. In this study, near-infrared (NIR) spectroscopy was used to quickly assess the composition of above-ground vegetative biomass from 113 diverse, photoperiod-sensitive, biomass-type sorghum (Sorghum bicolor) accessions cultivated under field conditions in Central Illinois. Biomass samples were analyzed using NIR spectra collected in the spectral range of 867–2536 nm, with their chemical compositions determined following the National Renewable Energy Laboratory (NREL) protocol. Advanced spectral pre-treatment and band selection techniques were utilized to develop calibration models using partial least squares regression (PLSR). The models' effectiveness was assessed through cross-validation and independent data tests. The predictions for moisture, ash, extractives, glucan, xylan, acid-soluble lignin (ASL), acid-insoluble lignin (AIL), and total lignin were accurate and reliable, demonstrating the capability of NIR spectroscopy to provide rapid and precise characterization of sorghum biomass. The results demonstrated that NIR spectroscopy is an efficient tool for rapidly characterizing sorghum biomass, making it a sustainable option for screening desirable feedstock for biofuel or bioproduct production.

A hyperspectral band selection strategy based on adjacency clustering and local structural correlation

ABSTRACT A popular strategy to reduce the spectral size of HyperSpectral Images (HSI) is to suppress all bands whose information content is largely duplicated in surviving bands. This is commonly termed Band Selection (BS). In this manuscript, a novel, unsupervised BS technique is proposed, consisting of three main steps: 1) Adjacent structural-similarity-based band partitioning, 2) for each partition, correlation-based, cluster-representative band selection, and 3) bad cluster removal using an Entropy measure. Initially, the adjacent structural similarity is measured, using Structural SIMilarity Index (SSIM); the next step is adjacency-partitioned cluster formation using a derivative-based approach. Then, a representative band with a high correlation is selected for each cluster. Finally, according to the Entropy measure, the cluster representatives are ranked to obtain the optimal subset of bands. The proposed Adjacency Clustering and Local Structural Correlation (ACLSC)-based BS is compared with other state-of-art approaches in terms of classification performance on benchmark HSI datasets. Overall Accuracy (OA), Average Accuracy (AA), and Kappa ( κ ) are computed on three datasets, namely, Indian Pines, Salinas, and Pavia University, to validate the proposed algorithm. Validation results support the use of adjacency clustering and incorporation of spatial-structural features for BS purposes.

A discriminatory groups-based supervised band selection technique for hyperspectral image classification

ABSTRACT Dimensionality reduction is a prevalent preprocessing step when analysing hyperspectral images (HSIs). It uses band selection (BS) and feature extraction (FE) techniques, but the former can retain the physical meaning and interpretability of the HSI datasets. Hence, these techniques are preferred to select a proper subset of bands without transformation. This letter introduces a straightforward supervised BS technique using discriminatory groups (DGs) that does not require complex computations. In this technique, total DGs are achieved equal to the spectral dimension of the HSI dataset, and each holds bands similar to the number of classes. The constituent bands of the DGs may present more than once, but the total number of bands in the group remains unchanged. The obtained DGs are ranked using high- to low-discrimination fractions, which indicate the uniqueness of the respective DG. The highest value that a DG can attain represents the group’s ability to distinguish among all classes (the maximum number of classes). Hence, the proposed technique has been termed ‘maximum class discriminatory group (MCDG)’. The discrimination fraction works as the control parameter for the selection of bands. Subsequently, the unique bands are selected from the ranked DGs and used for HSI classification. The proposed MCDG technique has an articulate and quick implementation without requiring any initialization of bands and outperforms various state-of-the-art methods while experimenting using real HSI datasets.

Uncovering Plastic Litter Spectral Signatures: A Comparative Study of Hyperspectral Band Selection Algorithms

This article provides insights into the optical signatures of plastic litter based on a published laboratory-scale reflectance data set (350–2500 nm) of dry and wet plastic debris under clear and turbid waters using different band selection techniques, including sparse variable selection, density peak clustering, and hierarchical clustering. The variable selection method identifies important wavelengths by minimizing a reconstruction error metric, while clustering approaches rely on the strengths of the correlation and local density of the spectra. Analyses of the data reveal three distinct absorption lines at 560, 740, and 980 nm that produce relatively broad reflectance peaks in the measured spectra of wet plastics around 475–490, 635–650, 810–815, and 1070 nm. The results of band selection consistently identify three important regions across 450–470, 650–690, and 1050–1100 nm that are close to the reflectance peaks of the mean of wet plastic spectra over clear and turbid waters. However, as the number of isolated important wavelengths increases, the results of the methodologies diverge. Density peak clustering identifies additional wavelengths in the short-wave infrared (SWIR) region of 1170–1180 nm) as a result of a high local density of the reflectance points. In contrast, hierarchical clustering isolates more wavelengths in the visible range of 365–400 nm due to weak correlations of nearby wavelengths. The results of the clustering methods are not consistent with the visual inspection of the signatures as peaks and valleys in the spectra, which are effectively captured by the variable selection method. It is also found that the presence of suspended sediments can (i) shift the important wavelength towards higher values in the visible part of the spectrum by less than 50 nm, (ii) attenuate the magnitude of wet plastic reflectance by up to 80% across the entire spectrum, and (iii) manifest a similar spectral signature with plastic litter from 1070 to 1100 nm.

Learning-Based Optimization of Hyperspectral Band Selection for Classification

Hyperspectral sensors acquire spectral responses from objects with a large number of narrow spectral bands. The large volume of data may be costly in terms of storage and computational requirements. In addition, hyperspectral data are often information-wise redundant. Band selection intends to overcome these limitations by selecting a small subset of spectral bands that provide more information or better performance for particular tasks. However, existing band selection techniques do not directly maximize the task-specific performance, but rather utilize hand-crafted metrics as a proxy to the final goal of performance improvement. In this paper, we propose a deep learning (DL) architecture composed of a constrained measurement learning network for band selection, followed by a classification network. The proposed joint DL architecture is trained in a data-driven manner to optimize the classification loss along band selection. In this way, the proposed network directly learns to select bands that enhance the classification performance. Our evaluation results with Indian Pines (IP) and the University of Pavia (UP) datasets show that the proposed constrained measurement learning-based band selection approach provides higher classification accuracy compared to the state-of-the-art supervised band selection methods for the same number of bands selected. The proposed method shows 89.08% and 97.78% overall accuracy scores for IP and UP respectively, being 1.34% and 2.19% higher than the second-best method.

Hyperspectral image-aided LiDAR point cloud labeling via spatio-spectral feature representation learning

Urban scene-level 3D point cloud labeling is a very laborious and expensive task compared to images. Conversely however, image processing techniques, deep learning or otherwise are more established and mature. Thus, in a multi-source data environment, the labeling of a point cloud scene via an automated image process as an initial step, followed by a manual human verification process is an effective way to save man hours and cost. With the above as the goal, this study presents a simple but robust spatio-spectral feature representation approach. In this approach, a class-aware band selection and reduction (CBSR) technique is developed for optimal hyperspectral feature representation. A double-branched convolutional Gaussian Bernoulli deep belief network (CGBDBN) is then used for hierarchical spatial feature extraction from LiDAR-derived data and the CBSR data. Using stacked ensemble learning, spatio-spectral features are generated from the two feature streams via a fusion rule and then classified — the results of which are used in labeling a raw 3D LiDAR point cloud through projection. To evaluate this study, extensive experiments were conducted on the IEEE 2018 Houston dataset — the only publicly available dataset with both hyperspectral image (HSI) and 3D point cloud covering the same area — for urban scene classification. The results indicate that the developed CBSR attained comparatively competitive results with state-of-the-art approaches, thus making it a robust spectral feature representation technique. Also, the weight-sharing property, probabilistic modeling, and hierarchical nature of CGBDBN gives our approach the ability to capture high-level contextual features. Furthermore, compared to the spatial- or spectral-only features, the generated spatio-spectral features are more discriminative and significantly aided in improving the proposed model’s efficacy. Overall, the proposed approach, based on the evaluation metrics, is a robust and effective approach for both coarse- and fine-grained raw LiDAR point cloud labeling tasks.

Investigation of Machine Learning Methods for Predictive Maintenance of the Ultra-High-Pressure Reactor in a Polyethylene-Vinyl Acetate Production Process

Ethylene-Vinyl Acetate (EVA) copolymer was synthesized from ethylene and vinyl acetate at high temperatures and ultra-high pressures. In this condition, any reactor disturbances, such as process or mechanical faults, may trigger the run-away decomposition reaction. This paper proposes a procedure for constructing a conditional health status prediction structure that uses a virtual health index (HI) to monitor the reactor bearing’s remaining useful life (RUL). The piecewise linear remaining useful life (PL-RUL) model was constructed by machine learning regression methods trained on the vibration and distributed control system (DCS) datasets. This process consists of using Welch’s power spectrum density transformation and machine learning regression methods to fit the PL-RUL model, following a health status construction process. In this research, we search for and determine the optimum value for the remaining useful life period (TRUL), a key parameter for the PL-RUL model for the system, as 70 days. This paper uses four-fold cross-validation to evaluate seven different regression algorithms and concludes that the Extremely randomized trees (ERTs) is the best machine learning model for predicting PL-RUL, with an average relative absolute error (RAE) of 0.307 and a Linearity of 15.064. The Gini importance of the ensemble trees is used to identify the critical frequency bands and prepare them for additional dimensionality reduction. Compared to two frequency band selection techniques, the RAE and Linearity prediction results can be further improved to 0.22 and 8.38.

Development of entropy measure for selecting highly sensitive WPT band to identify defective components of an axial piston pump

This paper proposes a novel tangent hyperbolic fuzzy entropy measure-based method for determining the highly (most) sensitive frequency band to easily identify defective components in an axial piston pump. The validation of proposed scheme was done by experimental and simulation analysis. In the proposed method, initially, the raw vibration signal is decomposed into 16 frequency bands. Second, the energy of the WPT band is computed. Third, lower bounds of energy readings at the 4th level of decomposition are extracted, and thereafter they are normalised and finally converted into the forms of fuzzy sets. Following that, the proposed measure's value is computed. The proposed measure is analogous to the quantity of available information. The WPT4 band with the highest measure value will have the most information and can be designated as sensitive. Finally, envelope demodulation of the sensitive band is computed to identify the defects by evaluating the magnitude of vibration at fundamental frequencies. The superiority of the proposed sensitive band selection technique has been justified by comparing the findings with the widely used Fast-Kurtogram, Protugram, and Spectral-Kurtosis methods. The authenticity of the proclaimed tangent hyperbolic fuzzy entropy measure is accomplished with the existing Deluca and Termini fuzzy entropy measures. The latter entropy measure exhibited some inconsequential and unspecified results and established the superiority of our defect identification methodology.

Unsupervised Cluster-Wise Hyperspectral Band Selection for Classification

A hyperspectral image provides fine details about the scene under analysis, due to its multiple bands. However, the resulting high dimensionality in the feature space may render a classification task unreliable, mainly due to overfitting and the Hughes phenomenon. In order to attenuate such problems, one can resort to dimensionality reduction (DR). Thus, this paper proposes a new DR algorithm, which performs an unsupervised band selection technique following a clustering approach. More specifically, the data set was split into a predefined number of clusters, after which the bands were iteratively selected based on the parameters of a separating hyperplane, which provided the best separation in the feature space, in a one-versus-all scenario. Then, a fine-tuning of the initially selected bands took place based on the separability of clusters. A comparison with five other state-of-the-art frameworks shows that the proposed method achieved the best classification results in 60% of the experiments.

Genetic Algorithm Captured the Informative Bands for Partial Least Squares Regression Better on Retrieving Leaf Nitrogen from Hyperspectral Reflectance

Nitrogen is a major nutrient regulating the physiological processes of plants. Although various partial least squares regression (PLSR) models have been proposed to estimate the leaf nitrogen content (LNC) from hyperspectral data with good accuracies, they are unfortunately not robust and are often not applicable to novel datasets beyond which they were developed. Selecting informative bands has been reported to be critical to refining the performance of the PLSR model and improving its robustness for general applications. However, no consensus on the optimal band selection method has yet been reached because the calibration and validation datasets are very often limited to a few species with small sample sizes. In this study, we address the question based on a relatively comprehensive joint dataset, including a simulation dataset generated from the recently developed leaf scale radiative transfer model (PROSPECT-PRO) and two public online datasets, for assessing different informative band selection techniques on the informative band selection. The results revealed that the goodness-of-fit of PLSR models to estimate LNC could be greatly improved by coupling appropriate band-selection methods rather than using full bands instead. The PLSR models calibrated from the simulation dataset with informative bands selected by genetic algorithm (GA) and uninformative variable elimination (UVE) method were reliable for retrieving the LNC of the two independent field-measured datasets as well. Particularly, GA was more effective to capture the informative bands for retrieving LNC from hyperspectral data. These findings should provide valuable insights for building robust PLSR models for retrieving LNC from hyperspectral remote sensing data.

Band selection technique based on binary modified equilibrium optimizer for hyperspectral image classification

The performance of hyperspectral image (HSI) classification can be enhanced by utilizing a metaheuristic technique with high exploration and exploitation capabilities for band selection. We design a framework for the improved classification of HSIs by utilizing a binary modified equilibrium optimizer (BMEO) for band selection. First, the proposed framework preprocesses the HSI data using the Gaussian filter to remove the undesired variations. Band selection is performed on the preprocessed HSI data using a BMEO, which exhibits balanced and high exploration and exploitation capabilities. The proposed V-shape transfer function is used for enhancing the exploration capabilities of the BMEO. An analysis for band selection is performed by comparing the reflectance spectra, correlation of each band with ground truth, and correlation between the bands before and after the band selection. The classification of HSI data by utilizing selected bands was done using the classifier selected by classifier analysis. The proposed framework outperforms 11 state-of-the-art techniques, including filter, wrapper, and deep learning-based techniques. The 95.85%, 98.13%, and 98.31% overall accuracies on the Indian pine, Pavia University, and Salinas datasets, respectively, show the significance of the proposed framework.

Discrimination of Chemical Oxygen Demand Pollution in Surface Water Based on Visible Near-Infrared Spectroscopy

Chemical oxygen demand (COD) is one of the indicators used to monitor the level of pollution in surface water. To recycle agricultural water resources, it is crucial to monitor, in a timely manner, whether COD in surface water exceeds the agricultural water control standard. A diagnostic model of surface water pollution was developed using visible near-infrared spectroscopy (Vis-NIR) combined with partial least squares discriminant analysis (PLS–DA). A total of 127 surface water samples were collected from Guangzhou, Guangdong, China. The COD content was measured using the potassium dichromate method. The spectra of the surface water samples were recorded using a Vis-NIR spectrometer, and the spectral data were pre-processed using four different methods. To improve the accuracy and simplicity of the model, the synthetic minority oversampling technique (SMOTE) and the competitive adaptive reweighted sampling (CARS) algorithm were used to enhance model performance. The best PLS–DA model achieved an accuracy of 88%, and the SMOTE–PLS–DA model had an accuracy of 94%. The SMOTE algorithm could improve the accuracy of the model despite the sampling imbalance. The CARS–SMOTE–PLS–DA model achieved 97% accuracy, and the CARS band selection technique improved the simplicity and accuracy of the discrimination model. The CARS–SMOTE–PLS–DA model improved the discrimination accuracy by 9% over that of the PLS–DA model. This method can not only save human and material resources but is also a new way for real-time online discrimination of COD in surface water.

Selection of shape-preserving, discriminative bands using supervised functional principal component analysis

ABSTRACT In this work, a band selection technique based on FDA and functional PCA is proposed. The method selects shape-preserving, discriminative bands which can highlight the important characteristics, variations as well as patterns of the hyperspectral data such that the differences between data from different classes become more apparent. The selection is performed in two stages. The first stage is selection of shape-preserving, keypoint bands which explain the trend characteristics of the reflectance curves in the hyperspectral data. These bands are able to preserve the functional representation of the curves as close as possible to their original representation using all the bands. The second stage involves selection of bands with good discriminative capabilities from the previously selected keypoint bands. For this, a novel iterative band selection technique using supervised functional principal component analysis (SFPCA) is proposed. The proposed band selection algorithm is the first approach reported in literature to explore the functional behaviour of HSI. An objective comparison of the proposed approach with several widely used, state-of-art band selection methods demonstrate a significant improvement in the classification accuracy.

Multiobjective Optimization-Based Hyperspectral Band Selection for Target Detection

Band selection can reduce information redundancy and improve application efficiency of hyperspectral images, such as classification. Traditional band selection methods usually weight different objectives and combine them together in a single function, making it difficult to balance conflict between various criteria. Recently, multiobjective optimization band selection techniques have got a lot of attention, which evaluate bands from different aspects separately, and search for a solution well balancing all the objectives simultaneously. However, few algorithms are focused on target detection, and most of them use just two objectives, which are not sufficient enough to select optimal band subset for detection. To alleviate these problems, this paper proposes an algorithm named target-oriented multiobjective optimization of band selection (TOMOBS). Firstly, the multiobjective optimization framework with three objective functions is modeled, involving information, noise, and correlation of the bands respectively. Secondly, in order to optimize the proposed model, the noninferior solution advantage matrix is designed based on the swarm intelligence optimization method, which can provide accurate solutions and improve the descriptiveness of multiobjective optimization problems. Thirdly, target-oriented evaluation mechanism is developed to guide selecting final result from the Pareto front, especially designed for target detection. Experiments on real hyperspectral datasets show that this algorithm can provide a subset of bands with strong representational capability for target detection, and achieve impressing results compared to the state-of-the-art methods.

DO-Net: Dual-Output Network for Land Cover Classification From Optical Remote Sensing Images

Land cover classification is the basic task of remote sensing image interpretation. Related methods have developed rapidly, especially the branch based on deep learning (DL). For high-resolution remote sensing images, the smaller inter-class difference and greater intra-class difference are two obstacles to improving the classification accuracy. For the former, the DL models generally use a deeper encoder to extract more powerful classification features. Considering that the scale of different land cover categories varies greatly, multi-scale feature extraction modules are also used to improve the classification accuracy. While the latter is always overlooked, and thus we propose a dual-output model, which uses a dense spatial pyramid pooling (DSPP) module to generate both the pixel-level and region-level predictions, to reduce the influence of intra-class differences. To further increase the classification accuracy, we investigate the band selection technique to apply the pre-trained encoder from the natural red green blue (RGB) dataset to multi-spectral remote sensing images. Extensive experiments on two datasets demonstrate the effectiveness of our model.

K-Means Embedded Deep Transform Learning for Hyperspectral Band Selection

In clustering-based hyperspectral band selection techniques, 2-D images of each band are usually taken as input samples. Some form of feature extraction on these images is performed before they are input to the clustering algorithm. The clustering algorithm returns the cluster centroids; the bands closest to the centroids are selected as representative bands for each cluster. In this work, we propose a joint representation learning and clustering framework. We embed the popular <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$K$ </tex-math></inline-formula> -means clustering loss into the newly developing framework of deep transform learning and solve the ensuing formulation via alternating direction method of multipliers (ADMM). We combine clustering with feature extraction. Application of our proposed solution to the hyperspectral band selection problem shows that we improve over the state of the art by a reasonable margin.

Featured Clustering and Ranking-Based Bad Cluster Removal for Hyperspectral Band Selection and Classification Using Ensemble of Binary SVM Classifiers

The rich spectral and spatial information of hyperspectral images are well known in the literature. The higher dimensionality of HSI creates Hughes's effect and increased computational complexity. This demands reduction for HS images as a pre-processing step. The necessary reduction of bands can be achieved by a proper band selection (BS) technique. The proposed features based unsupervised BS technique follows three subsequent steps: 1) for each band image statistical features are extracted, 2) bands are clustered with a k-means approach using the extracted features, 3) each cluster is ranked using mean entropy measure, 4) bad clusters are removed, and 5) for each selected cluster, a representative band is selected. The proposed method is validated over three widely used standard datasets and six state-of-the-art approaches using an ensemble of binary SVM classifiers. The obtained results strongly suggest the clustering is essential to reduce the redundancy, and removal of cluster is informative to keep the informative bands.

Selecting informative bands for partial least squares regressions improves their goodness-of-fits to estimate leaf photosynthetic parameters from hyperspectral data.

The plant photosynthetic capacity determines the photosynthetic rates of the terrestrial biosphere. Timely approaches to obtain the spatiotemporal variations of the photosynthetic parameters are urgently needed to grasp the gas exchange rhythms of the terrestrial biosphere. While partial least squares regression (PLSR) is a promising way to predict the photosynthetic parameters maximum carboxylation rate (Vcmax) and maximum electron transport rate (Jmax) rapidly and non-destructively from hyperspectral data, the approach, however, faces a high risk of overfitting and remains a high hurdle for applications. In this study, we propose to incorporate proper band selection techniques for PLSR analysis to refine the goodness-of-fit (GoF) in estimating Vcmax and Jmax. Different band selection procedures coupled with different hyperspectral forms (reflectance, apparent absorption, as well as derivatives) were examined. Our results demonstrate that the GoFs of PLSR models could be greatly improved by combining proper band selection methods (especially the iterative stepwise elimination approach) rather than using full bands as commonly done with PLSR. The results also show that the 1st order derivative spectra had a balance between accuracy (R2 = 0.80 for Vcmax, and 0.94 for Jmax) and denoising (when a Gaussian noise was added to each leaf reflectance spectrum at each wavelength with a standard deviation of 1%) on retrieving photosynthetic parameters from hyperspectral data. Our results clearly illustrate the advantage of using the band selection approach for PLSR dimensionality reduction and model optimization, highlighting the superiority of using derivative spectra for Vcmax and Jmax estimations, which should provide valuable insights for retrieving photosynthetic parameters from hyperspectral remotely sensed data.

A Review of Unsupervised Band Selection Techniques: Land Cover Classification for Hyperspectral Earth Observation Data

A hyperspectral image (HSI) is a collection of several narrow-band images that span a wide spectral range. Each band reflects the same scene, composed of various objects imaged at different wavelengths; the spatial information, however, remains generally consistent across bands. Both types of information, spectral and spatial, can be leveraged to identify and classify objects. Recently, the use of machine learning (ML) in object classification has become increasingly widespread. Regardless of the selected approach, object-specific spectral and spatial information is key to discriminating relevant categories. Whereas spatial information is usually repeated across bands, spectral information tends to be distributed more unevenly and often highly so. This poses the issue of removing redundancy, which is commonly called the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">band selection</i> ( <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">BS</i> ) <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">problem</i> and refers to identifying an optimal subset of bands for further HSI processing.

Whale optimization-based band selection technique for hyperspectral image classification

ABSTRACT Hyperspectral image (HSI) classification is one of the growing research areas in Remote Sensing. The high dimensionality of the data cube and a high correlation among the hyperspectral bands, poses a difficulty in HSI processing, and selection is one of the promising solutions to address the issue. In this research paper, a whale optimization-based band selection is proposed. Initially, the informative bands are retrieved via a search mechanism which resembles the hunting behaviour of the humpback whales. Later, a hybrid filter is proposed to extract the intrinsic and edge preserving spatial features. Finally, the smoothened bands are trained using a non-linear support vector machine for efficient classification. The exploitation and exploration in the local and global search for the selection of whale hyperspectral bands from the search space proved to be useful in achieving the high-quality classification maps. The overall accuracy reported on the three benchmark data sets – Indian Pines, University of Pavia and Salinas are 99.05, 97.88, and 97.56%, respectively. The proposed method accuracy is achieved with 5% fewer bands on Indian Pines, 70% on University of Pavia and 50% on Salinas Datasets over the other competing methods, which show the effectiveness of the whale optimization-based band selection.