Search

The near-surface air temperature (${{T}_{air}}$) is a principal variable describing energy exchange and water circulation between the land surface and the atmospheric environment. The estimation of ${{T}_{air}}$ by satellite land surface temperature (LST) is challenging due to the variable magnitude of the difference between ${{T}_{air}}$ and LST in both space and time, as well as the restriction of estimated ${{T}_{air}}$ to clear-sky conditions because of the penetration of infrared wavelengths. Moreover, the estimation suffers from low temporal resolution and primarily focuses on daily minimum, maximum, and two instantaneous ${{T}_{air}}$ per day. This study proposes a method for estimating all-sky gridded diurnal ${{T}_{air}}$ at regional scale from FY-4B/AGRI measurements. The multiscale geographically weighted regression model was investigated to establish the dynamic relationships between ground station observed ${{T}_{air}}$ and satellite LST under clear-sky conditions by employing different spatial values for each explanatory variable in localized regressions. A moving window loop based multiple linear regression was employed to establish the relationship between satellite-derived clear-sky ${{T}_{air}}$ and other variables to extrapolate ${{T}_{air}}$ in cloudy-sky pixels. The results showed that the proposed method captures the trend of ${{T}_{air}}$ variations well in hourly profiles with R values greater than 0.95. RMSE was 1.75 °C, 1.38 °C, 1.95 °C, and 2.19 °C in April, July, October, and January, respectively. The demonstration of heatwave monitoring showed that satellite-estimated ${{T}_{air}}$ provide an excellent representation of the spatial and temporal evolution of the heatwave.

Sugarcane is a significant crop in terms of annual biomass in the world. Timely and accurate mapping of sugarcane planting is important for food security and sustainability. However, accurately remote-sensing-based mapping sugarcane remains challenging due to two reasons: 1) the scarcity of sugarcane training samples, and 2) the diverse sugarcane fields planting dates. This article proposed a novel transfer learning algorithm for accurate sugarcane planting mapping through space and temporary in the U.S. This algorithm only depended on one place's training data to mapping other places’ growing sugarcane. We distilled burning sugarcane fields as training labels from Sentinel-2 in Palm Beach County and neighboring area in 2021. The time-invariant phenology features were calculated from composite Sentinel-2 normalized difference vegetation index (NDVI) series using linear cosine regression (LCR). They integrated as training samples to construct a one-class support vector machine (OCSVM) classifier, generating Jun.–Nov. growing sugarcane maps in Plam Beach County and Lafourche County 2022. Meanwhile, a postprocess method was used to improve mapping quality. These sugarcane maps were validated using surveyed and observed field dataset and compared against the cropland data layer (CDL). As a result, the sugarcane maps achieved the best accuracy in its mature stage (Sep. 2022), which exhibited nearly complete phenological characteristics. The maps show low misclassification rates and higher accuracy compared to the CDL 2022. Moreover, the LCR-OCSVM method was confirmed to have superior transfer learning capability for sugarcane classification across different regions and periods compared to the Harmonic-OCSVM method.

In this article, an integrated object detection and communication (IODC) system based on single-stage detection framework is first proposed for synthetic aperture radar (SAR) images in underwater environment. Specifically, the combination of the multiscale feature extraction module and semantic information enhance fusion module for SAR images in underwater environment is designed as the semantic encoder and the results prediction module with anchor-free detection method is designed as the semantic decoder. Considering the multiscale feature, the channel encoder composed by a multiscale fusion module and a redundancy module is designed, and the channel decoder is the inverse of the channel encoder. To adapt to the time-varying and complex wireless environment, an adaptive transmission (AT) mechanism based on attention mechanism and knowledge base is proposed for the IODC system. Moreover, considering the actual application requirements, a lightweight design for the IODC system with the AT mechanism is also conducted. The experiment results on the Sonar Common Target Detection dataset show that the proposed IODC system with AT mechanism can achieve nearly 91% average precision at 25 dB, which means the proposed system can achieve the effective integration of the objection detection and communication for SAR images in underwater environment. In the lightweight design, the model parameters of the proposed system can be reduced by up to 20%, with only a 2.22% sacrifice in performance.

Hyperspectral target detection technology is becoming more and more important in remote sensing. Most of the traditional methods of target sensing using spectral information treat hyperspectral image as 2-D matrix regardless of the structure information of hyperspectral image, resulting in insufficient separation of target and background and limited detection accuracy. In order to better utilize the hyperspectral intrinsic structural information, this article proposes a hyperspectral target sensing approach based on tensor adaptive reconstruction cascaded with global and local feature fusion (TRGLF). First, the Tucker decomposition and reconstruction method is utilized to alleviate the influence of noise and other elements in the complex background hyperspectral, which can maintain the intrinsic structural features of the data and improve the background and target separation. The principal components of the factor matrix are determined by a logarithmic singular value summation strategy, and then the energy difference between the spectra to be measured and the previous spectra is calculated to fine-tune the principal components and obtain more appropriate principal component values. Second, on the basis of reconstructed data, a global and local spatial spectral fusion method is adopted to obtain the target of interest. This includes using globally constrained energy minimization to obtain the target of interest and using local differential sensing to further obtain the target boundary. The detection performed on four realistically acquired hyperspectral water surface datasets demonstrate the excellent detection performance of the proposed target detection method.

The k-nearest neighbors (kNN) algorithm is a versatile tool for mapping forest attributes. However, the effects of using inadequate reference plots for modeling have not been thoroughly investigated. The interaction between topographic and biological factors results in a more complex distribution of forest attributes, leading to significant uncertainty and challenges in obtaining reliable information. This study presents a protocol that uses a 4-way factorial experimental design to establish appropriate sampling schemes aimed at reducing uncertainty and bias in estimating aboveground biomass (AGB) using the kNN technique. The research was conducted in a mixed forest within a subtropical region affected by wildfires, pine wilt disease, and agricultural activities. A total of 252 sampling schemes, incorporating various sampling methods, feature counts, neighbor counts (k), and reference-target distances (RTD), were utilized to generate corresponding AGB-kNN models. The performance of the models was assessed against measured AGB using a tree-based IPCC-compliant method with a high-resolution orthoimage and canopy-height-model data. Results indicated that these sampling schemes produced AGB estimations with average error rates ranging from 13% to 241%. The optimal kNN model utilized spectral, biophysical, and topographic features through a systematic sampling approach, with k set at 30 and RTD at 900 meters. Findings suggest that systematic sampling outperformed random and cluster sampling, with models that used moderate k and RTD generally providing less biased estimates. This protocol effectively identifies suitable kNN-AGB models, enhancing the ability to delineate areas with low biomass productivity for precision management, while also supporting forest improvement initiatives and biomass-related studies.

Semantic change detection (SCD) represents a challenging task in the interpretation of remote sensing images (RSIs), with the goal of identifying change regions and extracting semantic information from bitemporal RSIs simultaneously. The recent integration of deep neural networks leveraging multitask learning has shown promise in enhancing SCD performance. However, there is still a challenge in improving SCD performance, specifically in designing a fine-grained network structure that can handle the two subtasks of change region localization and semantic information recognition in parallel. In this context, a novel multitask Siamese network, termed EGMS-Net, is proposed to boost the performance of SCD, which consists of three core components. First, a coarse-to-fine multitask Siamese network is constructed to obtain semantic information and change information at multiple levels. Second, an adaptive change information enhancement method based on spatial-spectral collaborative attention mechanism is proposed, which can assist the accurate localization of change regions without significantly increasing the model parameters. Third, a change information guidance module is developed to strengthen the interaction between multitask branches and reduce the difficulty of network training. Experiments on three benchmark datasets demonstrate that the proposed EGMS-Net outperforms existing state-of-the-art methods in the SCD community.

Infrared moving small target detection is an important and challenging task in infrared search and track system, especially in the case of low signal-to-clutter ratio (SCR) and complex scenes. The spatial–temporal information has not been fully utilized, and there is a serious imbalance in their exploitation, especially the lack of long-term temporal characteristics. In this article, a novel method based on the spatial–temporal feature fusion tensor model is proposed to solve these problems. By directly stacking raw infrared images, the sequence can be transformed into a third-order tensor, where the spatial–temporal features are not reduced or destroyed. Its horizontal and lateral slices can be viewed as 2-D images, showing the change of gray values of horizontal/vertical fixed spatial pixels over time. Then, a new tensor composed of several serial slices are decomposed into low-rank background components and sparse target components, which can make full use of the temporal similarity and spatial correlation of background. The partial tubal nuclear norm is introduced to constrain the low-rank background, and the tensor robust principal component analysis problem is solved quickly by the alternating direction method of multipliers. By superimposing all the decomposed sparse components into the target tensor, small target can be segmented from the reconstructed target image. Experimental results of synthetic and real data demonstrate that the proposed method is superior to other state-of-the-art methods in visual and numerical results for targets with different sizes, velocities, and SCR values under different complex backgrounds.

The prevalence of zero values in zero-inflated time-series (ZI-TS) data poses significant challenges for traditional LSTM networks in learning long-term dependencies and trends. Specifically, the high proportion of zeros dilutes the influence of non-zero values, leading LSTM model to frequently predict zeros, which ultimately reduces prediction accuracy. A Weighted Zero-inflated Sensitive LSTM model (WZS-LSTM) which integrates both a stacked LSTM architecture and a Weighted Zero-inflated Sensitive (WZS) loss function was proposed to improve LSTM’s prediction accuracy in ZI-TS data. First, the stacked LSTM structure enhances the model’s capacity to capture long-term dependencies. Second, the model dynamically adjusts its focus towards non-zero values through the WZS loss function. Thorough experiments were conducted on UCI time-series dataset “WeatherAUS” and “Population”, and four typical time-series prediction model Prophet, ARIMA, LSTM, and the hurdle model are selected for comparative analysis of algorithm performance, while “WeatherAUS” is a zero-inflated time-series data, and “Population” is a normal time-series data. The results demonstrated that the WZS-LSTM model improved prediction accuracy, reducing errors by at least 2.38% when compared with the selected models: Prophet, ARIMA, LSTM, and the hurdle model in ZI-TS data, while the predictive performance of WZS-LSTM for general time-series data was comparable to ARIMA, Prophet and traditional LSTM. In addition, WZS reduced the error by an additional 0.21% compared to the best-performing loss function. Finally, it was demonstrated on real-world datasets that WZS-LSTM offers significant advantages in predicting ZI-TS data and holds great potential for broader application.

Reliability investigations are crucial for the advancement of light-emitting diode (LED) technology. Salty water vapor exposure poses a significant risk to the performance and longevity of AlInGaP LEDs, accelerating degradation. This study integrates multiple methods, including electrical and material analysis, machine vision, artificial intelligence (AI) recognition, and circuit design, to monitor the real-time degradation of LEDs. Our experiments reveal that exposure to saltwater vapor induces structural damage, including the detachment of the epitaxial layer and crack formation, leading to diminished of LED performance. Within 30 minutes of saltwater exposure, the device showed significant degradation in materials, electrical properties, and luminous areas, especially, the luminous area dropped to below 50% of its original brightness. Additionally, AI-based damage identification is implemented to monitor the LED’s state, triggering a protective circuit as the luminous area is less than 50%. This research offers valuable insights into LED degradation, real-time monitoring, and protective measures due to damage caused by saltwater vapor, and proposes a promising solution for early fault detection, paving the way for more robust LED-related systems in harsh environments.

We fabricate and demonstrate a thermal-radiation far-infared light source with enhanced thermal radiation efficiency, particularly within the crucial 8--14 μm wavelength range, which is vital for various applications. The device utilizes a metallic nanotube array (MeNTA) meticulously fabricated on a silicon wafer using stainless-steel material through sputter deposition. The simulation with Finite-Difference Time-Domain (FDTD) approach reveals significant alterations to the blackbodyradiated far-infrared spectrum. Experimental validation via Fourier Transform Infrared (FTIR) measurements confirms a pronounced wavelength filtering effect, primarily centered at 10.2 μm. The device incorporating stainless steel MeNTA exhibits 1.4 times improvement in luminous efficiency, reaching 7.45 × 10−3, accompanied by a radiated power of 11.034 mW and 0.681 mW/mm2 radiated power per area. These outcomes suggest the potential for expanding the fabrication process with alternative geometries and periods to tailor specific infrared emissions, which enables suitable thermal emitters for biomedical applications.

Recent years have seen a surge in service robots built on the ROS framework, where effective path planning algorithms are crucial for their successful operation. However, real-world operation presents challenges due to changes in environmental features and unexpected dynamic obstacles. If the path planning algorithms fail to update the state of the map layers in a timely manner, it can result in mission failures for the robots. Therefore, this study proposes a novel Motion Trajectory Analysis Regulator Navigation Framework within the ROS system. This framework incorporates path planning regulators for both global and local path planning algorithms, effectively mitigating path planning failures caused by dynamic environmental changes. For enhanced flexibility in path planning and dynamic obstacle avoidance, this framework proposes the ADRRT* algorithm for global path planning and the FGM-DWA algorithm for local path planning, replacing the traditional A* and DWA algorithms, respectively. Real-field experiments with a service robot demonstrate that the proposed methods not only ensure successful navigation with 100% success rate and a 23% improvement in time efficiency but also effectively handle dynamic and U-shaped obstacles, overcoming the challenge of path planning failures.

With the rapid development of vehicle intelligence and the Internet of Vehicles industry, passengers’ demand for travel experiences continues to increase. To provide more comfortable and convenient intelligent transportation services, it is necessary to effectively arrange routes that meet demand. However, the actual traffic environment is complex and changeable, and traditional path planning methods are no longer adaptable. To cope with the complex and changeable traffic environment, a new bus operation model called “flexible bus” is proposed. The flexible bus system is a bus model that responds to demand. Passengers can reserve shuttle buses according to their needs, and buses depart after a certain number of passengers is reached. Flexible bus vehicles are scheduled and planned in real time based on passenger needs and driving routes to improve operational efficiency and passengers’ boarding experiences. To achieve the goal of reducing bus operating costs and improving passengers’ boarding experience, a heuristic path planning algorithm called BusNav is proposed. BusNav includes two procedures: path extraction and trip allocation, which can efficiently dispatch buses to meet different operating costs and passengers’ boarding experience needs. Compared with traditional path planning methods, experimental results show that the BusNav algorithm performs well in terms of reduce operating costs and improve passengers’ boarding experience.

Image inpainting is an important task in computer vision, aiming at restoring missing or damaged areas in an image. The existing methods have certain problems such as texture blur and structure distortion, especially when the degenerated images exhibit complex structures and scenes. To address these issues, this paper proposes an image inpainting algorithm (called CTRF-GAN) based on the generative adversarial network (GAN) with CSWin Transformer module and Fast Fourier convolution residual module. The modified GAN framework replaces ordinary convolution in the encoder and decoder with gated convolution to reduce the training loss. The CSWin Transformer module is introduced into the generator to enhance the global dependency and to increase the receptive field. A fast Fourier convolution residual module (Res-AFFC) is proposed to extract high-frequency texture information. Finally, the HSV loss function is integrated to ensure that the color of the inpainting areas is consistent with the original image. Extensive experiments are conducted on the CelebA and Paris Street View datasets, verifying the superiority of the proposed algorithm.

The pathogenesis of essential tremor (ET) remains unclear, and the efficacy of related drug treatment is inadequate for proper tremor control. Hence, in the current study, consecutive low-frequency repetitive transcranial magnetic stimulation (rTMS) modulation on cerebellum was accomplished in a population of ET patients, along with pre- and post-treatment resting-state electroencephalogram (EEG) networks being constructed. The results primarily clarified the decreasing of resting-state network interactions occurring in ET, especially the weaker frontal-parietal connectivity, compared to healthy individuals. While after the rTMS stimulation, promotions in both network connectivity and properties, as well as clinical scales, were identified. Furthermore, significant correlations between network characteristics and clinical scale scores enabled the development of predictive models for assessing rTMS intervention efficacy. Using a multivariable linear model, clinical scales after one-month rTMS treatment were accurately predicted, underscoring the potential of brain networks in evaluating rTMS effectiveness for ET. The findings consistently demonstrated that repetitive low-frequency rTMS neuromodulation on cerebellum can significantly improve the manifestations of ET, and individual networks will be reliable tools for evaluating the rTMS efficacy, thereby guiding personalized treatment strategies for ET patients.

Advanced Persistent Threat (APT) groups pose significant cybersecurity threats due to their sophisticated and persistent nature. This study introduces a novel methodology to understand their collaborative patterns and shared objectives, which is crucial for developing robust defense mechanisms. We utilize MITRE ATT&CK Techniques, software, target nations, and industries as our primary features to understand the characteristics of APT groups. Since essential information is often buried within the unstructured data of Cyber Threat Intelligence (CTI) reports, we employ Natural Language Processing (NLP) and Named Entity Recognition (NER) to extract relevant data. To analyze and interpret the complex relationships between APT groups, we compute similarity among the features using weighted cosine similarity metrics and Machine Learning (ML) models, enhanced by feature crosses and feature selection strategies. Subsequently, hierarchical clustering is used to group APTs based on their similarity scores, helping to identify common behaviors and uncover deeper relationships. Our methodology demonstrates notable clustering performance, with a silhouette coefficient of 0.76, indicating strong intra-cluster similarity. The Adjusted Rand Index (ARI) of 0.63, though moderate, effectively measures agreement between our clustering and the ground truth. These metrics provide robust validation, surpassing commonly recognized benchmarks for effective clustering in cybersecurity. Our methodology successfully classifies 23 distinct APT groups into six clusters, highlighting the importance of techniques and industry features in the clustering process. Notably, techniques such as T1059 (Command and Scripting Interpreter) and T1036 (Masquerading) are prevalently deployed, observed in 18 out of 23 APT groups across all six clusters.

This paper proposes a physical charge-based analytical MOSFET threshold voltage model that explicitly incorporates interface-trapped charges which have been identified as playing a dominant role in defining threshold voltage trends in deep cryogenic temperatures. The model retains standard threshold voltage definition by various charges across the MOSFET capacitor while being analytical in its form, therefore, suitable for cryogenic CMOS VLSI design. Consequently, a model covering each and all above characteristics is proposed for the first time. Excellent fit between the model and measurement data from 180-nm bulk foundry devices is shown from room temperature to 4 K.

Spectral clustering (SC) is a hot research topic in graph theory today, which is centered on constructing similarity matrix. Many previous works have studied how to construct a higher-quality similarity matrix. However, most of these works ignore the connection between constructing similarity matrix and the overall distribution of the datasets. This problem will lead to low quality and lack of robustness in the similarity matrix when the data volume and dimensionality are expanded, which leads to poor clustering performance. Therefore, to solve the above problems, an Adaptive Fuzzy Spectral Clustering (AFSC) model is proposed in this paper. The model considers the local neighbor and global fuzzy affiliation information of the samples to construct the objective function, which can obtain the similarity matrix that satisfies the overall distribution of the datasets. Meanwhile, we propose a weight definition method to balance the above two kinds of information, which can prevent the adverse effect of information loss on the quality of the similarity matrix. The similarity matrix obtained by the AFSC model will have high quality and strong robustness, which leads to excellent clustering performance. In experiments, we verified that the AFSC model can obtain excellent clustering performance and outperform some state-of-the-art clustering models as it is applied to several datasets of different sizes and dimensions.

This research addresses the critical issue of traffic light detection and recognition in advanced driver assistance systems. In this paper, we propose a novel two-stage detection framework which is capable of recognizing different types of traffic lights, including arrow signals. Unlike most existing methods which detect entire traffic light boxes, our focus is on individual signal bulbs for detection and recognition. In the first stage, our approach aims to achieve an exceptionally low miss rate in detecting traffic signal bulbs, even the result contains many false positives. In the second stage, a classification network is employed to identify the correct traffic signal lights. Our method effectively tackles the diverse traffic light detection problems posed by varying arrangements of signal bulbs in different regions. Moreover, it can simultaneously detect individual traffic signals including various types of arrow lights. The experiments with real road scene images and public datasets have demonstrated the feasibility and effectiveness of our proposed technique.

Human action recognition has recently attracted much attention in computer vision research. Its applications are widely found in video surveillance, human-computer interaction, entertainment, and autonomous driving. In this study, we developed a system for evaluating online music performances. This system conducts experiments to assess performance of playing the erhu, the most popular traditional stringed instrument in East Asia. Mastering the erhu poses a challenge, as players often struggle to enhance their skills due to incorrect techniques and a lack of guidance, resulting in limited progress. To address this issue, we propose hybrid models based on graph convolutional networks (GCN) and temporal convolutional networks (TCN) for action recognition to capture spatial relationships between different joints or keypoints in a human skeleton, and interactions between these joints. This can assist players in identifying errors while playing the instrument. In our research, we use RGB video as input, segmenting it into individual frames. For each frame, we extract keypoints, encompassing both image and keypoint information, which serve as input data for our model. Leveraging our innovative model architecture, we achieve an impressive accuracy rate exceeding 97% across various classes of hand error modules, thus providing valuable insights into the assessment of musical performances and demonstrates the potential of AI-based solutions to enhance the learning and correction of complex human actions in interactive learning environments.

Multi-source sensor information fusion technology plays an important role in many application fields in the era of “Industry 4.0”. Generally, in practical engineering applications, the obtained information is inevitably uncertain due to the degradation of sensor performance, environmental interference, abnormal communication transmission process, etc. Compared with the strict requirements of probability density function for data quantity, the use of interval numbers to characterize the uncertainty of information has obvious advantages. Classical Dempster-Shafer(D-S) evidence theory can effectively deal with uncertain information, whereas its evidence structure is single-valued, making it difficult to fuse interval uncertainty information. Therefore, it is necessary to adopt an interval-valued evidence structure to carry out interval uncertainty information fusion. Current researches on interval-valued belief structure mainly focus on the modification of evidence combination rules but neglect the key problem of how to generate interval-valued evidence structure from original data obtained from sensors. Aiming at the problem of classification and discrimination based on multi-source interval-valued sensor information fusion, this paper adopts the theory of interval-valued evidence and focuses on the problem of generating a reasonable belief structure based on original interval-valued sensor information by using a multistage correction framework. Firstly, an interval-valued Lance and Williams Distance(ILWD) is proposed, and an initial belief structure is generated based on the proposed ILWD using the original sensor information. Secondly, by considering the interval length effect and based on Jaccard similarity coefficient, the initial interval-valued belief structure is corrected by a two-stage process. Then, in order to alleviate the impact of evidence conflict on the combination results, the weight of each evidence is assigned by an optimization method (the third stage of the belief structure modification). Further, the evidence is combined based on the method proposed by Wang et al. (2007). Finally, a practical case study is carried out to verify the rationality and effectiveness of the proposed method.

The quest for efficient and non-intrusive human activity recognition (HAR) in indoor environments has led to the burgeoning field of WiFi-based HAR. This method promises applications from healthcare monitoring to elderly care due to its cost-effectiveness and ease of deployment compared to traditional sensor-based systems. However, WiFi-based HAR faces significant challenges in maintaining performance across diverse environments and subjects, largely due to WiFi signal variability. Addressing this issue necessitates training models on extensive datasets. Recent studies have utilized conventional models, including Convolutional Neural Networks (CNNs), and sequence-to-sequence (Seq2Seq) models such as LSTM, GRU, or Transformer. Despite their precision, these models are computationally intensive and require more training data. To tackle these limitations, we propose a novel approach that leverages a Temporal Convolutional Network with Augmentations and Attention, referred to as TCN-AA. This model enhances computational efficiency and accuracy in the face of dataset variability by combining temporal convolutional layers with data augmentation strategies and an attention mechanism to focus on critical features. Our method is computationally efficient and significantly improves accuracy, even with a threefold increase in data size through augmentation techniques. Experiments on a public dataset indicate our approach outperforms state-of-the-art methods, achieving 99.42% accuracy. This work represents a significant step forward in the practical application of WiFi-based HAR, paving the way for broader adoption in critical areas like healthcare and security.

Automatically extracting meetup event information from the Internet can significantly enhance the discovery of activities. Existing methods for meetup event mining rely on the open APIs provided by event-based social networks (EBSN) to capture Meetup event data in designated regions and topics or a comprehensive crawling of the web to filter meetup events. Both approaches have limitations. In this study, we propose a novel four-stage framework to extract meetup events from event organizers’ websites, including event source page discovery, automatic pagination recognition, boilerplate removal, and event detection. From potential event organizer websites obtained from Facebook events, we built 7,012 profile APIs and obtained 520,909 published links from July 13, 2023, to June 24, 2024. Through the boilerplate remover, we extracted 289,541 pieces of valuable information and identified 69,284 event messages by the event detection module. The event page ratio of 13.3% of these event organizers’ websites is much higher than the 1% event page ratio of all websites, revealing the cost-effectiveness of the proposed approach.

The concept of human-centric lighting is gradually emerging. People no longer only consider the lighting function of light, but also begin to pay attention to the comfort and health of light. In this article, we perform an investigation on the circadian regulation of four-primary light-emitting diode (LED) based displays for the people with various ages from 1 year old to 100 years old. The double-Gaussian function is used to mimic the spectral power distribution of monochromatic LED. The main figures of merit of LED display performance are color gamut of Rec. 2020, luminous efficacy of radiation, correlated color temperature (CCT), and also non-visual parameters, such as circadian stimulus and melanopic efficacy of luminous radiation provided by Commission Internationale de l'Eclairage (CIE) standard. Then, the genetic algorithm is adopted for performing the spectral optimization of four-primary LED display while many constraint conditions are set, such as CCT values (2700 K and 6500 K), CCT tolerance, and color distance (Duv < 0.01) in the CIE 1960 UCS color space. Related results of visual and non-visual characteristics are also compared with those of three-primary LED displays. This work would pave a way for the development of healthy four-primary LED display in consideration of various ages of people.

The hippocampal region is one of the most affected brain areas observed as a landmark in Magnetic Resonance Imaging (MRI) images for Alzheimer’s disease (AD) diagnosis. The diminished alterations in the hippocampal and degeneration of cholinergic circuits have been conclusively correlated with a decline in memory and cognitive function. However, the hippocampal region may not appear as clearly defined as other brain regions, making it difficult for neurologists and researchers to identify by visual inspection. The application of deep learning models to pinpoint the hippocampal region was initially valued. We assessed the ability of a deep learning model, You Only Live Once (YOLO), to detect hippocampal regions in three MRI image views and categories. The Alzheimer’s Disease Neuroimaging Initiative-first (ADNI−1) dataset was used with 220 subjects in three categories using the three YOLO models. We obtained the YOLO performance for hippocampal region detection with accuracy in three views and categories. The average mean Average Precision (mAP) performance accuracy for YOLOv3 was 0.87, YOLOv4 was 0.85, and YOLOv5 was 0.96, respectively. The high accuracy of the detection of the hippocampal region was remarkable. We found that the sagittal view was higher than the axial and coronal views. Simultaneously, the Mild Cognitive Impairment (MCI) in the coronal view was lower among the three models. The results showed that YOLOv5 is a suitable model for detecting the hippocampal region in MRI images, and the sagittal view is the most reliable for detecting the hippocampal region in diagnosing AD. Our findings demonstrate the importance of detecting the hippocampal region to diagnose AD and accurately analyzing the hippocampal area within the region. The YOLOv5 model substantially affected performance metrics and interpretability across the three views and categories.

This paper reports on an all-oxide thin film piezotronic P-N heterojunction diode incorporating vertically-stacked structure of N+-ITO/P-type nickel oxide/N-type zinc oxide as a flexible energy scavenger and its diode characteristics on signal regulation which simplifies an essential element for harvesting which is signal rectification circuitry. An energy band diagram, theoretical modeling and equivalent small-signal circuit elaborate its working principle and device physics. Signal amplification due to introduction of in-series capacitances related to junction formation has also been addressed. A preliminary experimental study demonstrates applicability of such a flexible energy scavenger in various gratis non-stop thrusts originating from human body motions such as: simple tapping (as in typing) and walking actions for generating $\mu $ W-range power. Moreover, focusing on a simple power management system along with analysis of voltage waveforms in response to both resistive and capacitive loads unveils that the device is capable of quickly charging a capacitor and discharging it slowly allowing for possible energy storage. The estimation on generated power by a pixelated array that is obtainable due to ease of large-area fabrication processes and a single-pixel strip-based device exabits its feasibility as an energy source to power up some IoT nodes.

Generally, an inverter fed motor drive is often operated under low speed and/or light loads. Hence, except for good driving characteristics, having evenly high efficiencies over wide load range is also critical for a high-performance motor drive. A dual-inverter fed permanent-magnet synchronous motor (PMSM) drive achieving these goals is developed in this paper. First, the single-inverter fed drive is established with properly designed intelligent power module (IPM) based inverter schematic and control schemes. The dynamic modeling and systematic controller designs for current and speed control schemes are presented. Satisfactory dynamic and steady-state operating performances are verified experimentally. Next, the dual-inverter fed PMSM drive is constructed. The current sharing control and zero-sequence harmonic suppression approaches are proposed to yield improved performance. The robust current control is proposed to enhance the harmonic suppression response. Through the developed current robust error cancellation controller (CRECC), the robust and faster cancellation response is yielded. Finally, the motor drive efficiencies affected by power level, IPM type, inverter cell number, switching frequency, and DC-link voltage level are comparatively evaluated by measured results. Accordingly, the evenly high composite efficiencies over wide speed and load ranges are arranged.

The relationship between commercial vehicle travel patterns and urban functional areas reveals potential connections between urban form and human geographic flows, which provides critical information for optimizing urban transportation systems. Benefiting from the large-scale trajectory datasets, it would be possible to investigate deeper research by modeling the implied urban travel patterns. This study designs a framework to reveal the collective movement patterns of commercial vehicle trajectories inside the urban environment, focusing on their spatiotemporal variations within functional areas. Stopping behaviors of trajectories were identified to construct spatiotemporal origin-destination (OD) matrices, representing time-varying human geographic flows. The singular value decomposition (SVD) method was employed to quantify spatio-temporal OD matrice to obtain time and space travel features. Travel patterns’ dynamics and spatial interactions within functional areas were then analyzed. The experimental results obtained with real-life datasets from Changsha, China, uncovered three typical travel patterns depicting commercial vehicle activities in urban environment shifts from work-related locations on weekdays to leisure destinations on weekends, with central areas experiencing more short and medium-range trips. The findings provide scientific references for optimizing spatio-temporal travel patterns and functional distribution to meet the demands of urban development and traffic management strategies.

The vehicle license plate detection plays a key role in Intelligent Transportation Systems. Detecting license plates, such as cars, trucks, and vans, is useful for law enforcement, surveillance, and toll booth operations. How to detect license plates quickly and accurately is crucial for license plate recognition. However, the uneven light condition or the oblique shooting angle of the license plate to be detected changes dramatically in real-world complex capture scenarios and the detection difficulty increases. At the same time, distance, lighting, angle, and other requirements are quite high, which seriously affects the detection performance. Therefore, an improved YOLOv7 integrating the parameter-free attention module SimAM for license plate detection was proposed, namely YOLO-SLD. Without modifying the original ELAN architecture, which is the key component of YOLOv7, a SimAM mechanism was added at the end of the ELAN to better extract license plate features and increase computational efficiency. More importantly, the SimAM module does not require any parameters to be added to the original YOLOv7 network, reducing model computation, and simplifying the calculation process. The performance of the detection model with different attention mechanisms was tested on the CCPD dataset for the first time and the proposed method was proven to be effective. The experimental result shows that the YOLO-SLD model has higher detection accuracy and is more lightweight with mAP at 0.5 with the overall improvement in accuracy from 98.44% of the original YOLOv7 model to 98.91%, an increase of 0.47% in accuracy. The accuracy of the CCPD test subset in dark and light images has improved from 93.5% to 96.7%, an increase of 3.2% in accuracy. The parameter size of the model is reduced by 1.2 million parameters compared to the original YOLOv7 model. Its performance is better than the other prevalent license plate detection algorithms.

Sleep staging serves as a fundamental assessment for sleep quality measurement and sleep disorder diagnosis. Although current deep learning approaches have successfully integrated multimodal sleep signals, enhancing the accuracy of automatic sleep staging, certain challenges remain, as follows: 1) optimizing the utilization of multi-modal information complementarity, 2) effectively extracting both long- and short-range temporal features of sleep information, and 3) addressing the class imbalance problem in sleep data. To address these challenges, this paper proposes a two-stream encode-decoder network, named TSEDSleepNet, which is inspired by the depth sensitive attention and automatic multi-modal fusion (DSA2F) framework. In TSEDSleepNet, a two-stream encoder is used to extract the multiscale features of electrooculogram (EOG) and electroencephalogram (EEG) signals. And a self-attention mechanism is utilized to fuse the multiscale features, generating multi-modal saliency features. Subsequently, the coarser-scale construction module (CSCM) is adopted to extract and construct multi-resolution features from the multiscale features and the salient features. Thereafter, a Transformer module is applied to capture both long- and short-range temporal features from the multi-resolution features. Finally, the long- and short-range temporal features are restored with low-layer details and mapped to the predicted classification results. Additionally, the Lovász loss function is applied to alleviate the class imbalance problem in sleep datasets. Our proposed method was tested on the Sleep-EDF-39 and Sleep-EDF-153 datasets, and it achieved classification accuracies of 88.9% and 85.2% and Macro-F1 scores of 84.8% and 79.7%, respectively, thus outperforming conventional traditional baseline models. These results highlight the efficacy of the proposed method in fusing multi-modal information. This method has potential for application as an adjunct tool for diagnosing sleep disorders.

Successful production of metallic selective laser melting components requires a quality assurance process that can effectively and nondestructively assess internal defects. Ultrasound testing has emerged as a valuable modality for defect identification as well as the characterization of microstructural and material properties. In this study, a comprehensive quality assessment approach was developed, leveraging quantitative ultrasound parameters, images, and image texture analysis. The evaluation focused on Inconel 718 selective laser melting components manufactured using diverse combinations of process parameters. The investigation involved examining the correlation between ultrasound parameters, image texture features, and the overall quality of the fabricated components. To validate the findings obtained through ultrasound testing, micro-computed tomography was employed. The experimental results provided evidence that the ultrasound parameters and image texture features successfully detected nonhomogeneous microstructures present within the selective laser melting components. Notably, the speed of sound and attenuation exhibited a positive correlation of 0.64 and 0.74, respectively, with porosity, indicating their potential as indicators of porosity. Furthermore, image texture features such as entropy, contrast, and homogeneity demonstrated high correlations ( $R^{2}\gt 0.7$ ) with the quality of the components. Consequently, these identified features hold promising potential for aiding in the selection of optimal process parameters and the prediction of porosity in future research.

Deep learning technology has extensive application in the classification and recognition of medical images. However, several challenges persist in such application, such as the need for acquiring large-scale labeled data, configuring network parameters, and handling excessive network parameters. To address these challenges, in this study, we developed an adversarial-learning-based Taguchi convolutional fuzzy neural classifier (AL-TCFNC) for classifying malignant and benign lung tumors displayed in computed tomography images. In the framework of the developed AL-TCFNC, a fuzzy neural classifier replaces a conventional fully connected network, thereby reducing the number of network parameters and the training duration. To reduce experimental cost and training time, the Taguchi method was used. This method helps to identify the optimal combination of model parameters through a small number of experiments. The transfer learning of models across databases often results in subpar performance because of the paucity of labeled samples. To resolve this problem, we used a combination of maximum mean discrepancy and cross-entropy for adversarial learning with the proposed model. Two data sets, namely the SPIE–AAPM Lung CT Challenge data set and LIDC–IDRI Lung Imaging Research data set, were used to validate the AL-TCFNC model. When the AL-TCFNC model was used for transfer learning, it exhibited an accuracy rate of 89.55% and outperformed other deep learning models in terms of classification performance.

Objective: Blood circulation is an important indicator of wound healing. In this study, a tissue oxygen saturation detecting (TOSD) system that is based on multispectral imaging (MSI) is proposed to quantify the degree of tissue oxygen saturation (StO2) in cutaneous tissues. Methods: A wound segmentation algorithm is used to segment automatically wound and skin areas, eliminating the need for manual labeling and applying adaptive tissue optics. Animal experiments were conducted on six mice in which they were observed seven times, once every two days. The TOSD system illuminated cutaneous tissues with two wavelengths of light - red ( $\mathrm {\lambda } = 660$ nm) and near-infrared ( $\mathrm {\lambda } = 880$ nm), and StO2 levels were calculated using images that were captured using a monochrome camera. The wound segmentation algorithm using ResNet34-based U-Net was integrated with computer vision techniques to improve its performance. Results: Animal experiments revealed that the wound segmentation algorithm achieved a Dice score of 93.49%. The StO2 levels that were determined using the TOSD system varied significantly among the phases of wound healing. Changes in StO2 levels were detected before laser speckle contrast imaging (LSCI) detected changes in blood flux. Moreover, statistical features that were extracted from the TOSD system and LSCI were utilized in principal component analysis (PCA) to visualize different wound healing phases. The average silhouette coefficients of the TOSD system with segmentation (ResNet34-based U-Net) and LSCI were 0.2890 and 0.0194, respectively. Conclusion: By detecting the StO2 levels of cutaneous tissues using the TOSD system with segmentation, the phases of wound healing were accurately distinguished. This method can support medical personnel in conducting precise wound assessments. Clinical and Translational Impact Statement—This study supports efforts in monitoring StO2 levels, wound segmentation, and wound healing phase classification to improve the efficiency and accuracy of preclinical research in the field.

Stroke is a leading cause of global population mortality and disability, imposing burdens on patients and caregivers, and significantly affecting the quality of life of patients. Therefore, in this study, we aimed to explore the application of virtual reality technology in physical therapy by using immersive interactive training and designing rehabilitation modes for individual and group settings. We also aimed to provide patients with stroke with a comprehensive home-based rehabilitation and treatment plan, ultimately enhancing training effectiveness. Patients can engage in home-based rehabilitation through this system and undergo functional, mirror, and constraint-induced therapies tailored to different task contents. Simultaneously, using brain-computer interface technology, an emotion analysis mechanism was designed to map the patients’ brainwave signal data onto a two-dimensional space of positive-negative valence arousal; this approach can enable remote physical therapists to discern the patients’ emotional states during the rehabilitation process through virtual spaces, facilitating timely adjustments to rehabilitation tasks. Moreover, to prevent compromised effectiveness owing to improper training postures leading to compensation, the system offers real-time identification and recording, promptly issuing alerts when compensation occurs. The system provides a multiuser virtual rehabilitation space, enabling timely corrections and data observations, offering patients with stroke a home-based rehabilitation program, thereby realizing a localized aging care model.

The emergence of deep-learning technology has significantly improved the performance of super-resolution algorithms for a single remote sensing image; however, the number of deep-learning model parameters is large, which limits its real-time deployment. In addition, the reconstructed image quality still needs improvement. To deal with the aforementioned problems, a lightweight multiscale residual attention information distillation network is proposed in this article. It achieves high-quality and fast super-resolution processing of remote sensing images. First, a high- and low-frequency (HF and LF) separation reconstruction strategy is adopted that enables the network to improve the reconstructed details of HF components while keeping the number of model parameters low. Second, a novel multiscale residual attention information distillation group is designed as the key component to further extract richer regional features with different perceptual fields and HF information while reducing the number of network parameters. This is achieved by combining a multiscale residual information distillation block that consists of multiple residual convolutional sub-blocks and an HF channel aware attention block. Last, the experimental results show that, compared with existing mainstream methods, such as the MHAN, the number of model parameters can be reduced by 75%, and the edge details of the reconstructed images are richer and more complete. The corresponding peak signal-to-noise ratio and SSIM can reach 31.59 dB and 0.824, respectively, under the condition of a $ \times 4$ magnification factor, and 27.39 dB and 0.668, respectively, under the condition of a $ \times 8$ magnification factor.

In recent years, rapid advancements in hardware and deep learning technologies have paved the way for the extensive integration of image recognition and object detection into daily applications. As reliance on deep learning grows, so do concerns about the vulnerabilities of deep neural networks, emphasizing the need to address potential security issues. This research unveils the Entropy-boosted Loss, a novel loss function tailored to generate adversarial patches resembling potted plants. Specifically designed for the YOLOV2, YOLOV3, and YOLOV4 object detectors, these patches obscure the detectors’ ability to identify individuals. By enhancing the uncertainty in class probability, a person wearing an adversarial patch crafted using our proposed loss function becomes less identifiable by YOLO detectors, achieving the desired adversarial effect. This underscores the significance of comprehending the vulnerabilities of YOLO models to adversarial attacks, particularly for individuals aiming to obscure their presence from camera detection. Our experiments, conducted using the INRIA person dataset and under real-time network camera conditions, confirm the effectiveness of our method. Moreover, our technique demonstrates substantial success in virtual try-on environments.

The charge capture of color inks in chromatic electrowetting electronic paper, the contact angle hysteresis and inconsistent aperture ratio result in a narrower color gamut and reduced contrast of the display screen. To address the issue, we propose an algorithm of the adaptive color conversion and the dynamic enhancement display based on the aperture ratio of the chromatic electrowetting. Firstly, the relationship between the aperture ratio of electrowetting and the increase or decrease of voltage was tested through an experimental platform. Then, combined with the change of the aperture ratio, the reasonable grayscale allocation is made for color space conversion. Furthermore, considering the average grayscales of each color, we design a method of the dynamic contrast enhancement and histogram equalization based on the segmentation of aperture ratios. In addition, in view of the problem of grayscale loss in the reflective display, we propose an adaptive algorithm of domain error diffusion with saturation enhancement to compensate for color loss. The experiments demonstrate that the images processed by our algorithm have shown certain improvements in evaluation indicators compared to images processed by other methods. Most significant of all, it has better visual display effect on chromatic electrowetting electronic paper, displaying more details.

FengYun-3E (FY3E) is the world's first dust-dawn orbiting meteorological satellite for civil use, which has filled the vacancy of global early-morning-orbit satellite observation by working together with FengYun-3C and FengYun-3D satellites. The Medium Resolution Spectral Imager-Low Light (MERSI-LL) sensor carried by FY3E can detect surface temperature variation. The radiometric calibration conditions of MERSI-LL thermal infrared bands were evaluated using the collected field measurements and atmospheric transfer simulations during 5–23 December, 2022, at Lake Erhai. A thermal infrared radiometer equipped on an unmanned surface vehicle was used to continuously collect surface-emitted radiance and water temperature. Atmospheric conditions, surface emissivity, and aerosol optical depth measured near-field experiment site were adopted by atmospheric radiative transfer code to calculate the influence of atmosphere on long-wave radiation during the propagation from land surface to satellite aperture. The good in-orbit operational status could be detected according to our calibration experiments, and the calibration analysis suggested that the differences between the simulated brightness temperature and satellite-based brightness temperature are 0.71 K with an RMSE of 0.79 K and 0.11 K with an RMSE of 0.47 K for channels 6 and 7, respectively, which achieved better or similar calibration accuracy by comparing with TIR channels of other FengYun satellites.

We studied the characteristics of coupled-core four-core fibers (CC-4CFs) with different core pitches both theoretically and experimentally. The spatial mode dispersions of different fibers were analyzed, and the result agrees well with theoretical ones. In the experiment, we achieved a spatial mode dispersion as low as 3.72 ps/km1/2 for a 4 km-long CC-4CF with core pitch of 19 μm at a bending radius of 8 cm. The impulse response matrices of four core fibers with different core pitches was investigated using swept-wavelength interferometry over 100 nm bandwidth. The impulse responses were Gaussian-shaped and consistent between different inputs and outputs for CC-4CFs. Our results show that the CC-4CFs with core pitches of 17 and 19 μm exhibit narrower pulse broadening. In addition, the mode-dependent losses of different fibers were analyzed. For the fibers with core pitch of 19 μm, the estimated mode-dependent loss was lower than 3 dB for the entire 100 nm bandwidth. Our results are especially helpful for developing coupled-core multicore fibers for long-haul transmission applications.

According to the research status of fast beam pointing and scanning in space laser communication, this paper proposes a beam deflection method that cascades liquid crystal tunable phase retarder with polarization grating, and realizes fast beam scanning and pointing according to the polarization and diffraction characteristics of light and analyzes the relationship between voltage and delay, the voltage regulation accuracy is 0.001 V, and whether the spot intensity can be used in space laser communication is verified, and the fast pointing of four spots of ±6° and ±2° is realized. The pointing accuracy is 30.49 μrad, the beam divergence angle is about 0.0339°, and it is connected to the blazed grating, and the single beam is scanned at 4.062° and 16.06° by wavelength tuning from 1525 nm to 1600 nm, with a wavelength tuning accuracy of 0.01 nm and repeated positioning accuracy of about 0.01°, which has certain research significance for the miniaturization and lightweight design of space laser communication optical transceiver and lidar.

The China National Intellectual Property Administration (CNIPA) receives a tremendous volume of patent applications every year, with the detection of plagiarized patents entailing substantial costs. To overcome the limitations of existing models in handling the extended textual features of patent claim documents, this research presents the PatentGraph model, which integrates Pretrained Language Model (PLM) with Graph Neural Networks (GNNs). It is specifically designed to perform graph modeling based on the hierarchical citation structure of patent claims. The proposed model introduced a local-to-global strategy to extract full-text features for comprehensively understanding the claim text. By employing a Siamese Neural Network, PatentGraph calculates the similarity between patent pairs and gets precise plagiarism statuses. We collected 10,317 authorized Chinese patents from public database and constructed plagiarism sample sets by state-of-art AI tools and randomly rearranging the sentence order from similar authorized patents. The experiment results indicate that the PatentGraph surpasses baseline models across all evaluation metrics, with accuracy and F1 Score enhancements of 8.0% and 8.3%, respectively. The proposed PatentGraph furnishes a fresh perspective on automating the detection of patent plagiarism, and indicates the potential for enhanced efficiency and precision in the patent examination process.

Micro-expressions refer to brief, subtle facial movements often concealing genuine human emotions. However, the advancement of Micro-Expression Recognition (MER) is hindered by the low frame rates of frame-based cameras. Although the successor event camera has a high frame rate, there are currently difficulties in obtaining and unstable performance issues. Drawing inspiration from the operational principles of event camera, we introduces two event features. Beyond spatial information, these features encode temporal and polarity information of event. Following local normalization, we employ the temporal polar pixel-wise interaction module to extract local feature. Additionally, we construct a temporal polar dynamic network, merging local feature with dense global optical flow to map deeper features. Experimental results demonstrate the superiority of the proposed method across multiple datasets compared to state-of-the-art approaches. This work enriches the encoding of event features, enhancing their performance in micro-expression recognition tasks and contributing to the future proliferation of event camera technology.

Radiative cooling, a technique that operates without energy consumption, requires achieving ultra-broadband high absorption/emission within atmospheric window band (8- $13~\mu $ m), posing a challenge for the design of cooling structure. In this work, we propose and investigate an ultra-broadband near-perfect metasurface absorber comprising a titanium (Ti) metal layer, a magnesium fluoride (MgF2) dielectric layer, and a one-dimensional MgF2 grating overlaid with Ti grating. The absorber structure is optimized using an improved genetic algorithm. In the normal incidence range of 8- $13~\mu $ m wavelength, the optimized structure achieves 96.1% average absorption for transverse electric polarization and 92.2% for transverse magnetic polarization. Moreover, the average absorption for unpolarized light reaches 94.1%. Then, we also examine how each layer in the proposed absorber impacts the absorption characteristics. In addition, the relationship between absorption and the angle of incidence is studied, and the proposed structure exhibits strong angular insensitivity. The proposed structure’s electromagnetic field distribution is analyzed to probe its underlying physical mechanism. Additionally, the optimized structure exhibits excellent nighttime radiative cooling performance, with 121.8 W/m2 net cooling power at 300 K ambient temperature. The proposed structure exhibits excellent potential for application in the field of thermal management related to radiative cooling.

The advancement of intelligent manufacturing has significantly increased the complexity and demand for efficient automated logistics systems in multi-floor manufacturing and warehousing environments. This paper presents a novel integrated queuing network model for automated vehicle storage and retrieval systems (AVS/RS) that allows for rapid performance estimation in multi-level and multi-depth configurations. The model bridges the gap between traditional simulation and analytical approaches, enabling quick evaluations during the design phase. The study conducts numerical experiments and simulation verifications to assess the impact of different rack configurations on system performance. Key findings include the optimal horizontal and vertical design of racks, showing that increasing the number of aisles and sub-tiers can enhance throughput and reduce cycle time. Additionally, it is shown that a lower number of main tiers leads to better performance, even with the same total number of tiers. The results offer valuable insights for designing efficient AVS/RS, contributing to improved flexibility and efficiency in automated logistics systems. This research has practical implications for optimizing storage solutions in complex, multi-floor manufacturing environments, ultimately enhancing operational efficiency.

To enhance the navigation accuracy and interpretability of Unmanned Aerial Vehicles (UAVs) in sports analysis, this study proposes an improved model based on the Temporal Convolutional Network (TCN) algorithm, integrated with Explainable Artificial Intelligence. The model aims to analyze real-time data collected by Internet of Things sensors mounted on UAVs to capture athletes’ swing actions during table tennis matches, thereby providing transparent decision support. The research effectively addresses the vanishing gradient problem by replacing the traditional Rectified Linear Unit (ReLU) activation function with Leaky ReLU, while simplifying the network structure through the use of a Global Average Pooling layer to reduce model complexity. Additionally, the residual structure is fine-tuned to adapt to dynamic environmental features, thereby enhancing the model’s capability to recognize swing actions. During the experimental phase, the model is evaluated using a dataset of swing actions collected by UAVs, comprising 55,582 data samples, which are divided into training and testing sets in a 3:2 ratio. The results reveal that the proposed improved TCN achieves remarkable outcomes in table tennis swing action recognition, with an algorithmic recognition accuracy of 99.43%, as well as recall, precision, and F1 scores of 99.00%. This algorithm’s recognition accuracy surpasses that of TCN, Long Short-Term Memory, and Convolutional Neural Network-Long Short-Term Memory algorithms by 10.57%, 3.65%, and 2.70%, respectively. The enhancement in recognition performance provides robust support for the technical training and competitive performance of table tennis players. This research outcome offers new insights into the application of UAVs in sports analysis and lays the groundwork for optimizing models and feature extraction techniques in broader sports application scenarios.

In the era of generative artificial intelligence (GAI), preserving personal authenticity is crucial for maintaining trust in online communities. Blockchain technology, with its unique blend of anonymity and data transparency, has facilitated the development of decentralized identities (DIDs) on the internet. These DIDs serve as digital representatives for individuals online, enhancing user trust through self-sovereign identity, transparency, and user control, while strengthening privacy. Current DID solutions, such as Proof-of-Personhood and Proof-of-Reputation, have demonstrated technological maturity and paved the way for identity applications in Web3. Building on this foundation, the Twin3 project advances the concept further by introducing the innovative Proof-of-Authenticity (PoA) to represent pluralistic human characteristics in digital format. Specifically, Twin3 employs blockchain-based Soulbound Tokens (SBTs) to transform human identities into a programmable hexadecimal matrix with 256 characteristic spaces, representing a comprehensive spectrum of human attributes, akin to a digital DNA (Decentralized Nexus of Authenticity). This study presents the Twin3 system architecture, user data management,demonstrating how this approach can enhance personal digital identity verification and management. The potential applications of Twin3 SBTs in creating personal AI agents and facilitating participation in the agentic economy are explored, demonstrating how Twin3 can digitize human attributes comprehensively and empower individuals within emerging agent-based workflows and digital ecosystems.

Artificial intelligence (AI) ethics has emerged as a global discourse within both academic and policy spheres. However, translating these principles into concrete, real-world applications for AI development remains a pressing need and a significant challenge. This study aims to bridge the gap between principles and practice from a regulatory government perspective and promote best practices in AI governance. To this end, we developed the Ethical Regulatory Framework for AI (ERF-AI) to guide regulatory bodies in constructing mechanisms, including role setups, procedural configurations, and strategy design. The framework was developed through a systematic review, thematic analysis, and the integration of interdisciplinary concepts. A comprehensive search was conducted across four electronic databases (PubMed, IEEE Xplore, Web of Science, and Scopus) and four additional sources containing AI standards and guidelines from various countries and international organizations, focusing on studies published from 2014 to 2024. Thematic analysis identified and refined key themes from the included literature and integrated concepts from process control theory, computer science, organizational management, information technology, and behavioral psychology. This study adhered to the PRISMA guidelines and employed NVivo for thematic analysis. The resulting framework encompasses 23 themes, particularly emphasizing three feedback-loop processes: the ethical review process, the incentive and penalty process, and the mechanism improvement process, offering theoretical guidance for the construction of ethical regulatory mechanisms. Based on this framework, a seven-step process and case examples for mechanism design are presented, enhancing the practicality of ERF-AI in developing ethical regulatory mechanisms. Future research is expected to explore customization of the framework to remain responsive to emerging AI trends and challenges, supported by empirical studies and rigorous testing for further refinement and expansion.

Aiming at the ineffective tracking problem of photovoltaic tracking system caused by cloud shading, an efficient and high-precision cloud trajectory prediction method is proposed. Firstly, the transmission method is used to filter the associated cloud type factors to reduce the computational complexity, and a GP-DCGAN gradient penalty deep convolution generation adversarial network is proposed to realize the generalization of image data. Secondly, by combining the alpha-IoU loss function with CA and CBAM attention mechanisms, an $\alpha $ CCB-YOLOv5s model is proposed to achieve accurate identification of clouds. Finally, based on the $\alpha $ CCB-YOLOv5s network, the DeepSORT module is added to realize multi-target tracking. Using a series of historical location information fed back by the tracking process, the Kalman filter algorithm is used to correct and predict the moving trajectory of the tracking cloud. The experimental results show that the detection categories of cloud type factors are reduced by 50% and the computational complexity is reduced by transmission method. At the same time, compared with the original YOLOv5s algorithm in cloud detection, the precision P of $\alpha $ CCB-YOLOv5s model is increased by 4.56%, the recall rate R increased by 5.48%, the mean average precision mAP increased by 4.4%, and the target detection performance is significantly improved. In addition, $\alpha $ CCB-YOLOv5s is used as a detector in the prediction model. The mean absolute error (MAE) and root mean square error (RMSE) are significantly reduced, and the prediction accuracy is greatly improved, which can lay a foundation for the follow-up intelligent photovoltaic tracking strategy research.

This paper presents a new prototype of a single-ended single-phase class-E direct ac-ac converter for induction heating (IH) applications, which is the first proposal for the application of variable inductor (VI) technology to a direct ac-ac converter, thereby achieving fixed-frequency control with a wide soft-switching (SSW) range. Combining VI technology enhances IH system flexibility and resolves issues of mechanical noise and electromagnetic interference (EMI) caused by variable frequency modulation, which are prevalent in conventional IH methods. This paper analyzes the proposed converter’s operation, explaining the VI control strategy for fixed-frequency power regulation, with key VI parameters designed using a reluctance model. Furthermore, experimental verification with a 850W/50kHz prototype confirms the fundamental performance capabilities of the proposed direct ac-ac converter. This evaluation demonstrates the high-frequency load power regulation and power conversion efficiency, and further confirms the implementation of SSW. The proposed converter achieved a high efficiency of 96.0%, which is significant for reducing energy loss of IH systems.

The structure of a submersible permanent magnet tubular linear motor (SPPMTLM) can significantly affect its load spectrum. However, the existing models do not account for this effect. This study proposes an advanced SPPMTLM model that incorporates these considerations, offering enhanced accuracy and practicality. First, a dynamic sub-model of the SPPMTLM was developed to elucidate the dynamic load spectrum during one working cycle. Subsequently, a one-dimensional analytical sub-model and a lumped-parameter thermal network sub-model of the SPPMTLM were established to calculate the electromagnetic and thermal performance. The proposed model utilizes the Newton-Raphson method to address the nonlinear equations derived from the integration of the three constituent sub-models. This approach enables a detailed analysis of the coupling relationships among the stress, fluid, thermal, and electromagnetic fields. Based on the proposed model, the impacts of the motor structural parameters, materials, and operating conditions on the electricity consumption per ton of fluid were analyzed. Furthermore, the particle swarm optimization algorithm was applied to the proposed model to optimize the SPPMTLM performance under different well fluid supply capacities. Finally, prototype and field experimental validations are conducted to verify the accuracy and practicality of the proposed method.

Knowledge-based authentication techniques remain one of the proven ways of maintaining confidentiality, ensuring integrity, and guaranteeing the availability of an information system. They employ what a user knows (Passwords or PINs) to authorize or grant access to an information system. While passwords employ a fixed combination of characters, Personal Identification Numbers (PINs) are majorly numbers. Existing implementations of these authentication techniques involve the repetitive use of static passwords and PINs at every login instance. These have been exposed to various attacks, such as keyloggers, shoulder surfing, brute force, and dictionary attacks. To overcome these attacks, this study presents an authentication technique where users’ PINs are incremented during successive login attempts. Users are expected to choose a preferred incremental factor, which can be any number they can remember, that will be added to the default 6-digit PIN to produce a dynamic PIN that can be used in subsequent login sessions. Furthermore, an additional layer of security that involves the use of a dynamic 4 by 4 graphical grid was integrated into the proposed incremented PIN technique. At every login session, users are presented with a set of 16 possible PINs to choose from. The security analysis of the proposed authentication technique revealed that the proposed technique could resist existing password attacks, thereby enhancing security. A performance testing and usability analysis was also carried out among 1145 individuals who interacted with the web application that uses the incremental authentication technique. The questionnaire items were structured based on the constructs of the Unified Theory of Acceptance and Use of Technology (UTAUT) Model. Statistical analysis of the responses received showed an appreciable level of acceptance in terms of performance expectancy, effort expectancy, social influence, and facilitating conditions. The positive user acceptance results provide reassurance about the practicality and effectiveness of the proposed technique. It is believed that the proposed incremental graphical grid authentication technique will further enhance the security of our growing mobile and web applications.