Hyperspectral (HS) images are adjacent band images that are generally used in remote-sensing applications. They have numerous spatial and spectral information bands that are extremely useful for ...material detection in various fields. However, their high dimensionality is a big challenge that affects their overall performance. A new data normalization method was developed to enhance the variations and data distribution using the output of principal component analysis (PCA) and quantile transformation, called QPCA. This paper also proposes a novel HS images classification framework using the meta-learner technique to train multi-class and multi-size datasets by concatenating and training the hybrid and multi-size kernel of convolutional neural networks (CNN). The high-level model works to combine the output of the lower-level models and train them with the new input data, called meta-learner hybrid models (MLHM). The proposed MLHM framework with our external normalization (QPCA) improves the accuracy and outperforms other approaches using three well-known benchmark datasets. Moreover, the evaluation outcomes showed that the QPCA enhanced the framework accuracy by 13% for most models and datasets and others by more than 25%, and MLHM provided the best performance.
Groundwater fluoride is posing a health risk to humans, and analyzing groundwater quality is time-wasting and expensive. Statistical methods provide a valuable approach to study the spatial ...distribution of groundwater fluoride. Random Forest (RF), Artificial Neural Network (ANN), and Logistic Regression (LR) were used in this study for groundwater fluoride prediction in Datong Basin. The groundwater chemistry of 482 groundwater samples was collected and used to figure out the performance of three statistical technologies and extract the main factors controlling the enrichment of fluoride in groundwater. The data was separated into two parts for the statistical analysis, 80% for training and 20% for testing. The Chi-squared was applied to select the most relevant variables, and TDS, Cl−, NO3−, Na+, HCO3−, SO42−, K+, Zn, Ca2+, and Mg2+ were selected as best inputs for the fluoride prediction. Models were evaluated using the confusion matrix and The receiver operating characteristic area under the curve ROC (AUC). The results suggest that within ten input variables, the accuracies of RF, ANN, and LR were 0.89, 0.85, and 0.76, respectively. The mean decrease in impurity (MDI) and permutation feature demonstrates that eight of ten parameters, including TDS, Cl−, NO3−, Na+, HCO3−, SO42−, Ca2+ and Mg2+ are the variables influencing the groundwater fluoride in the study area. RF exhibited the best model with high conformity and confidence in predicting groundwater fluoride contamination in the study area.
•Three algorithms were employed for the prediction of groundwater fluoride.•The variables influencing the fluoride in the study area were evaluated.•Python 3.7 was used for the analysis of the three models.
Multi-class detection in remote sensing images (RSIs) has garnered wide attention and introduced several service applications in many fields, including civil and military fields. However, several ...reasons make detection from aerial images very challenging and more difficult than nature scene images: Objects do not have a fixed size, often appear at very various scales and sometimes appear in dense groups, like vehicles and storage tanks, and have different surroundings or background areas. Furthermore, all of this makes the manual annotation of objects very complex and costly. The powerful effect of the feature extraction methods on object detection and the successes of deep convolutional neural networks (CNN) extract deep features more than traditional methods. This study introduced a novel network structure and designed a unique feature extraction which employs squeeze and excitation network (SENet) and residual network (ResNet) to obtain feature maps, named a shallow-deep feature extraction (SDFE), that improves the resolution and the localization at the same time. Furthermore, this novel model reduces the loss of dense groups and small objects, and provides higher and more stable detection accuracy which is not significantly affected by changing the value of the threshold of the intersection over union (IoU) and overcomes the difficulties of RSIs. Moreover, this study introduced strong evidence about the factors that affect the detection of RSIs. The proposed shallow-deep and multi-scale (SD-MS) method outperforms other approaches for the given ten classes of the NWPU VHR-10 dataset.
ABSTRACTIn the rapidly evolving realm of remote sensing technology, the classification of Hyperspectral Images (HSIs) is a pivotal yet formidable task. Hindered by inherent limitations in ...hyperspectral imaging, enhancing the accuracy and efficiency of HSI classification remains a critical and much-debated issue. This review study focuses on a key application area in HSI classification: Land Use/Land Cover (LULC). Our study unfolds in fourfold approaches. First, we present a systematic review of LULC hyperspectral image classification, delving into its background and key challenges. Second, we compile and analyze a number of datasets specific to LULC hyperspectral classification, offering a valuable resource. Third, we explore traditional machine learning models and cutting-edge methods in this field, with a particular focus on deep learning, and spectral decomposition techniques. Finally, we comprehensively analyze future developmental trajectories in HSI classification, pinpointing potential research challenges. This review aspires to be a cornerstone resource, enlightening researchers about the current landscape and future prospects of hyperspectral image classification.
Hyperspectral images (HSIs) are typically utilized in a wide variety of practical applications. HSI is replete with spatial and spectral information, which provides precise data for material ...detection. HSIs are characterized by a high degree of variations and undesirable pixel distributions, providing major processing challenges. This article introduces the fast hyperspectral image classification (FHIC) model, a rapid model for classifying HSIs and resolving their associated challenges. It uses the enhancing transformation reduction (ETR) method to address the HSI difficulties and enhance classes' differentiation. It also uses exponential linear units (ELUs) to smooth and speed the classification processing. The structure of the FHIC model is designed to be very flexible and suitable for a range of HSIs. The model reduced execution time and RAM consumption, and provided superior performance compared to seven of the most advanced analysis models for three well-known HSIs. In some cases, it was 60% faster than other models. In addition, this work presents a new and highly effective method for measuring the performance of the compared models in terms of their accuracy and processing speed to provide an easy evaluation method. The code of the FHIC model is available at this link: https://github.com/DalalAL-Alimi/FHIC .
•A noverl feature extraction method to reduce HSI dimensions, called IDA is proposed.•IDA is applied to improve data correlation and distribution and classes variance.•Study the impact of feature ...extraction and preproccesing on the HSI classification.•Using different datasets with well-known exisiting methods to test the IDA.
Hyperspectral images (HSIs) are known for their high dimensionality and wide spectral bands that increase redundant information and complicate classification. Outliers and mixed data are common problems in HSIs. Thus, preprocessing methods are essential in enhancing and reducing data complexity, redundant information, and the number of bands. This study introduces a novel feature reduction method (FRM) called improving distribution analysis (IDA). IDA works to increase the correlation between related data, decrease the distance between big and small data, and correct each value's location to be inside its group range. In IDA, the input data passes through three stages. Getting rid of outliers and improving data correlation is the first step. The second stage involves increasing the variance. The third is to simplify the data and normalize the distribution. IDA is compared with four popular FRMs in four available HSIs. It is also tested and evaluated in various classification models, including spatial, spectral, and spectral-spatial models. The experimental results demonstrate that IDA performs admirably in enhancing data distribution, reducing complexity, and accelerating performance.
•Use the ETR method to increase the data stationariness.•Propose a new hybrid model for energy forecasting.•Use six different energy time series datasets.•The hybrid model with the ETR method ...provides the best results.•The proposed methods yield the best results for lengthy and short data periods.
Due to weather and political fluctuations that significantly impact the production and price of energy sources, enhancing data distribution and reducing data complexity is crucial to achieving accurate forecasting. Additionally, it is essential to provide a flexible forecasting model capable of handling rapid changes in the energy market and effectively anticipating energy supplies and demands. This study introduces a novel method to deal with energy market fluctuations in the long and short term and provide highly accurate forecasts for various energy data. It uses the Enhancing Transformation Reduction (ETR) method to improve the stationarity of the data, reduce seasonality and trend, and resolve rapid fluctuations. The output of ETR is then passed into a hybrid forecasting model referred to as “ Time-Series Forecasting Model using Long Short-Term Memory integrated with Artificial Neural Networks” (TLIA). The TLIA model benefits from transfer learning, which transmits the output of the LSTM layers into the ANN layers, enabling TLIA to base its work on the best performance and continue improving it. The study evaluates and tests its methods using six different datasets, including the electricity dataset of Victoria State, the oil price for the West Texas Intermediate, the Elia Grid load dataset, and wind power production. In addition to its characteristics, ETR accelerates and enhances the TLIA processing to achieve the highest accuracy compared to seven forecasting models in all six datasets. The TLIA is often 40 times or more superior to competing models. Compared to another model, the Mean Absolute Error (MAE) results of TLIA range between (0.008 and 0.088) versus (0.77 and 4318.544).
The vision transformer (ViT) has demonstrated performance comparable to that of convolutional neural networks (CNN) in the hyperspectral image classification domain. This is achieved by transforming ...images into sequence data and mining global spectral-spatial information to establish remote dependencies. Nevertheless, both the ViT and CNNs have their own limitations. For instance, a CNN is constrained by the extent of its receptive field, which prevents it from fully exploiting global spatial-spectral features. Conversely, the ViT is prone to excessive distraction during the feature extraction process. To be able to overcome the problem of insufficient feature information extraction caused using by a single paradigm, this paper proposes an MLP-mixer and a graph convolutional enhanced transformer (MGCET), whose network consists of a spatial-spectral extraction block (SSEB), an MLP-mixer, and a graph convolutional enhanced transformer (GCET). First, spatial-spectral features are extracted using SSEB, and then local spatial-spectral features are fused with global spatial-spectral features by the MLP-mixer. Finally, graph convolution is embedded in multi-head self-attention (MHSA) to mine spatial relationships and similarity between pixels, which further improves the modeling capability of the model. Correlation experiments were conducted on four different HSI datasets. The MGEET algorithm achieved overall accuracies (OAs) of 95.45%, 97.57%, 98.05%, and 98.52% on these datasets.
•Present new approach to classify hyperspectral images (HIS).•Propose a new feature extraction method, called ETR to boost HIS classification.•ETR decreases scattering caused by noise, mixed pixels, ...and outliers of HSI.•ETR smooths the classification process and accuracy and corrects pixels’ positions.
Recent improvements in remote sensing techniques (RST) allow for collecting hyperspectral images (HSIs) with enhanced spatial and spectral resolution. These enhancements boost the volume, velocity, and variety of obtaining HSIs. However, many HSIs comprise extensive variability and undesired pixel distribution because of different atmospheric scattering conditions or maybe because of a malfunction in the RST, called mixed pixels and noise, posing significant processing issues. Dimensionality reduction methods (DRM) have been developed to address this issue. They reduce the complexity of data while retaining the relevant information for the analysis. This study presents a novel DRM called enhancing transformation reduction (ETR) that can enhance and reduce the dimensionality of HSI. The data in ETR goes through two parts. The first part reduces dimensionality and increases the classes' variation. The second part aims to reduce the noise and the gap between values, and it corrects pixels' location and distribution. ETR is an HSI DRM that is quick. It was evaluated using four distinct HSIs and various state-of-the-art classification models. It enables faster extraction, more robust results, and prompt classification compared to other notable DRMs. The code of the ETR method is available on the GitHub site (https://github.com/DalalAL-Alimi/ETR).
In Hyperspectral images (HSI), dimensionality reduction methods (DRM) play a critical role in reducing the input data dimension and complexity. As much as the deep learning methods (DLM) have ...presented very aggressive achievements, preprocessing methods and DRM are very important to enhance the learning of DLMs. This study introduces a novel DRM called Compression and Reinforced Variation (CRV), which is used to reduce the input data dimension. The CRV minimizes the gap between the big and small related data in each class and omits the noise and redundant data. It selects the most informative features and normalizes them to enhance data distribution before inserting them into the learning model. The learning model of this study is multi-hybrid deep learning (MHDL) model to improve the extraction of multi-class HSI and spectral–spatial features. MHDL is a novel classification model that includes hybrid layers of conventional neural networks and batch normalization to avoid overfitting, normalizing the training, and extracting the spectral–spatial features for HSI. The proposed CRV provided highly efficient methods for reducing the HSI dimension and improving the classification accuracy of the MHDL model. In contrast to other conventional DRMs, CRV gave the highest accuracy in the shortest time. CRV-MHDL was also compared to seven existing classification models for three distinct datasets, and the findings demonstrated that the CRV-MHDL outperforms all of them by more than 2%. The code of this study is available at this link: https://github.com/DalalAL-Alimi/CRV.
•Propose hyperspectral image classification model using multi-hybrid deep learning model.•Propose a feature selection method by increasing the variance and filtering data.•Apply compression and reinforced variation to enhance data compression and classification.•Evaluate the proposed approach using different HS images with extensive comparisons.