Article
Geochemistry & Geophysics
Wenping Ma, Jianchao Shen, Hao Zhu, Jun Zhang, Jiliang Zhao, Biao Hou, Licheng Jiao
Summary: With the rapid development of earth observation technology, the article introduces a novel adaptive hybrid fusion network (AHF-Net) for multiresolution remote sensing image classification. The AHF-Net combines the adaptive weighted intensity-hue-saturation (AWIHS) strategy for data fusion and a correlation-based attention feature fusion (CAFF) module for feature fusion, resulting in improved classification accuracy. Experimental results demonstrate the effectiveness of the proposed algorithm.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2022)
Article
Remote Sensing
Wenxuan Wang, Leiming Liu, Tianxiang Zhang, Jiachen Shen, Jing Wang, Jiangyun Li
Summary: Convolutional neural networks have been dominating the downstream tasks on hyperspectral remote sensing images with their strong local feature extraction capability. However, they fail to effectively capture long-range dependencies, which the Transformer architecture can handle. This paper introduces a dual-branch Transformer architecture called Hyper-ES2T, which effectively utilizes spatial information and spectral correlations in hyperspectral images. The design also includes an efficient multi-head self-attention block to balance model accuracy and efficiency.
INTERNATIONAL JOURNAL OF APPLIED EARTH OBSERVATION AND GEOINFORMATION
(2022)
Article
Environmental Sciences
Xulun Liu, Shiping Ma, Linyuan He, Chen Wang, Zhe Chen
Summary: This paper proposes a hybrid network model, TransConvNet, to address the challenges in oriented object detection in remote sensing images. It integrates the advantages of CNN and self-attention-based network, pays attention to global and local information aggregation, and adapts to the object direction variability. The proposed method demonstrates effectiveness through extensive experimental results.
Article
Chemistry, Multidisciplinary
Jin Zheng, Tong Wang, Zhi Zhang, Hongwei Wang
Summary: This paper proposes an object detection method combining feature enhancement and hybrid attention to address the challenges in object detection in remote sensing images. The proposed method includes a feature enhancement fusion network and a hybrid attention mechanism, which improve the robustness and discriminability of features and focus on object features while suppressing background noises. Experimental results on the DOTA dataset demonstrate the effectiveness of the proposed method.
APPLIED SCIENCES-BASEL
(2022)
Article
Computer Science, Information Systems
Monika Sharma, Mantosh Biswas
Summary: This paper describes a unique semisupervised classification technique using a deep learning based hybrid framework (DL-HF) for hyperspectral images. The proposed method improves classification accuracy by pre-labeling unlabeled samples, expanding the training set, and utilizing self-arrangement based on deep learning. Evaluation results demonstrate that the proposed DL-HF algorithm outperforms other competing classification schemes on benchmark datasets.
MULTIMEDIA TOOLS AND APPLICATIONS
(2023)
Article
Environmental Sciences
Yinghui Quan, Yingping Tong, Wei Feng, Gabriel Dauphin, Wenjiang Huang, Wentao Zhu, Mengdao Xing
Summary: A novel feature fusion method, RTVSA, for urban area classification is proposed in this paper, combining features derived from HSI and LiDAR data. The method effectively extracts structural correlation, withstands noise well, and improves land cover classification accuracy, as demonstrated in experiments conducted on two urban Houston University datasets.
Article
Computer Science, Artificial Intelligence
Dalal AL-Alimi, Mohammed A. Al-qaness, Zhihua Cai, Eman Ahmed Alawamy
Summary: This study introduces a novel feature reduction method called improving distribution analysis (IDA) to enhance data distribution, reduce complexity, and accelerate performance in hyperspectral images (HSIs). The experimental results demonstrate that IDA performs admirably in achieving these goals.
PATTERN RECOGNITION
(2023)
Article
Environmental Sciences
Jing Liu, Zhe Yang, Yi Liu, Caihong Mu
Summary: A pixel frequency spectrum feature is introduced to CNNs to improve the classification accuracy of HRSIs through deep fusion feature extraction. Experimental results demonstrate better recognition results with more discriminant information using multi-branch CNNs.
Article
Geochemistry & Geophysics
Zixu Liu, Li Ma, Qian Du
Summary: This study investigates class-wise adversarial adaptation networks for the classification of hyperspectral remote sensing images. By adversarial learning between feature extractor and multiple domain discriminators, domain-invariant features are generated, and a probability-prediction MMD method is introduced to improve feature-alignment performance. The proposed CDA network can achieve unsupervised classification of target images and has demonstrated efficiency in experiments using Hyperion and AVIRIS hyperspectral data.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2021)
Article
Environmental Sciences
Haibin Wu, Huaming Zhou, Aili Wang, Yuji Iwahori
Summary: This paper proposes two novel classification frameworks based on multilayer perceptrons (MLPs), namely dilation-based MLP (DMLP) and DMLP with performance feature fusion using multi-branch residual blocks and principal component analysis (PCA) (DMLPFFN). Experimental results show that these methods outperform several state-of-the-art methods in hyperspectral precise crop classification.
Article
Geography, Physical
Chongxin Tao, Yizhuo Meng, Junjie Li, Beibei Yang, Fengmin Hu, Yuanxi Li, Changlu Cui, Wen Zhang
Summary: In the study of automatic interpretation of remote sensing images, semantic segmentation based on deep convolutional neural networks has been widely developed and applied. However, most current network designs focus on the visible RGB bands, neglecting the spectral information in the invisible light bands such as NIR. To address this issue, this paper proposes a novel deep neural network structure called the multispectral semantic segmentation network (MSNet), which achieves competitive performance for semantic segmentation of multi-classified feature scenes by leveraging the advantages of multispectral data and incorporating visible and invisible bands.
GISCIENCE & REMOTE SENSING
(2022)
Article
Computer Science, Artificial Intelligence
Dalal AL-Alimi, Zhihua Cai, Mohammed A. A. Al-qaness, Eman Ahmed Alawamy, Ahamed Alalimi
Summary: The study introduces a novel dimensionality reduction method (ETR) for enhancing and reducing the dimensionality of hyperspectral images. ETR is quick, provides more robust results, and enables prompt classification compared to other notable dimensionality reduction methods.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Environmental Sciences
Yue Qiu, Fang Wu, Haizhong Qian, Renjian Zhai, Xianyong Gong, Jichong Yin, Chengyi Liu, Andong Wang
Summary: To address the issue of intraclass heterogeneity and interclass homogeneity in remote sensing images, we propose an Attentional Feature Learning Network (AFL-Net) that accurately extracts buildings. The AFL-Net adapts the weights of multiscale features through attention mechanism and captures shape features to improve building recognition accuracy in complex environments.
Article
Environmental Sciences
Jiaqi Wang, Zhihui Gong, Xiangyun Liu, Haitao Guo, Donghang Yu, Lei Ding
Summary: Object detection plays an important role in remote sensing image interpretation. This study proposes an adaptive feature-aware method based on the single-shot detector architecture for object detection in remote sensing images. The approach utilizes self-attention and adaptive feature-aware module to improve localization and reduce the influence of complex backgrounds. Experimental results demonstrate that the proposed method achieves high accuracy and real-time detection of remote sensing images.
Article
Environmental Sciences
Jing Liu, Yang Li, Feng Zhao, Yi Liu
Summary: In this study, a spectral fractional-differentiation (SFD) feature is proposed to extract effective features for the terrain classification of hyperspectral remote-sensing images (HRSIs). A criterion for selecting the fractional-differentiation order based on maximizing data separability is also introduced. The effectiveness of the SFD feature is verified using four traditional classifiers and five network models, and the results show that the SFD feature can effectively improve the accuracy of terrain classification for HRSIs, especially in cases with small-size training samples.
Article
Computer Science, Artificial Intelligence
Mohamed Abd Elaziz, Laith Abualigah, Ahmed A. Ewees, Mohammed A. A. Al-qaness, Reham R. Mostafa, Dalia Yousri, Rehab Ali Ibrahim
Summary: In this paper, a modified version of Manta Ray Foraging Optimization (MRFO) called MRTMO is proposed to overcome the issue of trapping in local solutions in metaheuristic techniques. The proposed MRTMO integrates the triangular mutation operator and orthogonal learning strategy to achieve a balance between algorithm cores and guide the search agents effectively. Extensive experiments demonstrate the competitive performance of MRTMO in solving optimization and engineering problems.
APPLIED INTELLIGENCE
(2023)
Article
Computer Science, Artificial Intelligence
Laith Abualigah, Mohamed Abd Elaziz, Dalia Yousri, Mohammed A. A. Al-qaness, Ahmed A. Ewees, Raed Abu Zitar
Summary: This paper proposes a new data clustering method by using the advantages of metaheuristic optimization algorithms. A novel arithmetic optimization algorithm (AOA) is introduced to address complex optimization tasks. By integrating opposition-based learning (OLB) and Levy flight (LF) distribution, a new variant of AOA called Augmented AOA (AAOA) is developed to improve the exploration and exploitation trends of the traditional AOA. Extensive experiments and comparisons with other optimization algorithms demonstrate the superiority of AAOA in both benchmark functions and data clustering datasets.
JOURNAL OF INTELLIGENT MANUFACTURING
(2023)
Article
Computer Science, Artificial Intelligence
Ahmed A. Ewees, Mohammed A. A. Al-qaness, Laith Abualigah, Zakariya Yahya Algamal, Diego Oliva, Dalia Yousri, Mohamed Abd Elaziz
Summary: Feature selection techniques are crucial for improving the performance of data analysis and decision making. This paper introduces a modified method based on metaheuristic algorithms that increases convergence rate and avoids local optima. Results from testing on twenty datasets and comparing with other methods demonstrate the effectiveness of this approach in reducing dimension and improving prediction rates and performance metrics.
NEURAL COMPUTING & APPLICATIONS
(2023)
Article
Engineering, Multidisciplinary
A. M. Sadoun, I. M. R. Najjar, A. Fathy, Mohamed Abd Elaziz, Mohammed A. A. Al-Qaness, A. W. Abdallah, M. Elmahdy
Summary: Due to the lack of analytical solutions, a modified machine learning method named Dendritic Neural (DN) was used to predict the wear performance of copper-alumina (Cu-Al2O3) nanocomposites. The optimal weights of DN were determined by a new meta-heuristic technique named Artificial Hummingbird Algorithm (AHA) to improve its performance. The developed model using AHA algorithm showed excellent predictability of the wear rate and coefficient of friction for Cu-Al2O3 nanocomposites with reinforcement content up to 10%.
ALEXANDRIA ENGINEERING JOURNAL
(2023)
Article
Computer Science, Interdisciplinary Applications
Mohamed Abd Elaziz, Mohammed A. A. Al-qaness, Abdelghani Dahou, Rehab Ali Ibrahim, Ahmed A. Abd El-Latif
Summary: This paper proposes an efficient intrusion detection system for IoT-cloud based environments, using swarm intelligence algorithms and deep neural networks. Deep neural networks are used to obtain optimal features from IoT IDS data, and a feature selection technique based on the Capuchin Search Algorithm (CapSA) is proposed. The developed model, CNN-CapSA, is tested with four IoT-Cloud datasets and compared with other optimization algorithms, showing competitive performance.
ADVANCES IN ENGINEERING SOFTWARE
(2023)
Article
Environmental Sciences
Mohammed A. A. Al-qaness, Ahmed A. Ewees, Hung Vo Thanh, Ayman Mutahar AlRassas, Abdelghani Dahou, Mohamed Abd Elaziz
Summary: Decreasing fossil fuel utilization and anthropogenic greenhouse gases is a global goal to combat climate change and air pollution. Underground carbon storage (UCS) is a promising solution, but there are barriers to its global application. In this study, a hybrid algorithm called AOSMA was developed using swarm intelligence to enhance the prediction capability of the LSTM model. Evaluation experiments showed that AOSMA outperformed other algorithms in predicting CO2 storage efficiencies.
ENVIRONMENTAL SCIENCE AND POLLUTION RESEARCH
(2023)
Article
Mathematics
Mohammed A. A. Al-qaness, Abdelghani Dahou, Ahmed A. A. Ewees, Laith Abualigah, Jianzhu Huai, Mohamed Abd Elaziz, Ahmed M. M. Helmi
Summary: Many Chinese cities suffer from severe air pollution due to rapid economic development, urbanization, and industrialization. Particulate matter (PM2.5) is a major component of air pollutants and is associated with cardiopulmonary and other systemic diseases due to its ability to penetrate the human respiratory system. Forecasting PM2.5 concentration is vital for governments and local authorities to plan and take necessary actions.
Article
Mathematics
Mohamed Abd Elaziz, Abdelghani Dahou, Dina Ahmed Orabi, Samah Alshathri, Eman M. Soliman, Ahmed A. Ewees
Summary: The rapid spread of fake information and news related to the COVID-19 pandemic on social media platforms has raised serious concerns for public health and safety. This paper proposes a disinformation detection framework using multi-task learning and meta-heuristic algorithms to analyze Arabic social media posts. The experimental results show that the proposed framework achieves an accuracy of 59% and outperforms other algorithms in all evaluation measures.
Article
Computer Science, Artificial Intelligence
Dalal AL-Alimi, Zhihua Cai, Mohammed A. A. Al-qaness, Eman Ahmed Alawamy, Ahamed Alalimi
Summary: The study introduces a novel dimensionality reduction method (ETR) for enhancing and reducing the dimensionality of hyperspectral images. ETR is quick, provides more robust results, and enables prompt classification compared to other notable dimensionality reduction methods.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Computer Science, Theory & Methods
Ahmed M. Helmi, Mohammed A. A. Al-qaness, Abdelghani Dahou, Mohamed Abd Elaziz
Summary: In the era of smart life, human activity recognition (HAR) can play a significant role in advanced applications such as IoT, IoHT, smart homes, eldercare, and health informatics-based applications. This paper integrates the applications of deep learning and swarm intelligence to build a robust HAR system using wearable sensor data. A light feature extraction approach is developed using the residual convolutional network and RCNN-BiGRU. New feature selection methods based on the marine predator algorithm (MPA) are also developed. The results show that MPAV achieved the best performance compared to other MPA variants and other compared methods.
FUTURE GENERATION COMPUTER SYSTEMS-THE INTERNATIONAL JOURNAL OF ESCIENCE
(2023)
Article
Energy & Fuels
Dalal AL-Alimi, Ayman Mutahar AlRassas, Mohammed A. A. Al-qaness, Zhihua Cai, Ahmad O. Aseeri, Mohamed Abd Elaziz, Ahmed A. Ewees
Summary: To achieve accurate energy forecasting, it is important to enhance data distribution and reduce data complexity to deal with weather and political fluctuations. This study introduces a novel method that combines ETR and TLIA models to improve the accuracy of energy forecasts. The TLIA model demonstrates superior performance compared to other models, achieving higher accuracy in various datasets.
Article
Geochemistry & Geophysics
Dalal AL-Alimi, Zhihua Cai, Mohammed A. A. Al-qaness
Summary: This article introduces a fast hyperspectral image classification model (FHIC) that improves the performance of hyperspectral image classification through the use of enhancing transformation reduction and exponential linear units. The model has a flexible structure and is suitable for various hyperspectral images, with faster execution time and superior performance compared to other models.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)