The attack surface of Industrial IoT (IIoT) is enlarged by the interconnected devices and systems. Although many works have facilitated advanced approaches to help industrial entities against possible cyber threats, they may overlook the rich operational context of manufacturing processes, leaving the security evaluation context-agnostic. Recognizing that cyber attacks and production activities are increasingly intertwined, this paper introduces Meta-KadaSec, a novelMeta-Knowledge-drivenadaptiveSecurity provisioning approach that embeds security context within the natural operational fabric of manufacturing environments. Our approach integrates: (1) STKG-PPO, a reinforcement learning model that leverages manufacturing contextual knowledge through Knowledge Graphs with Spatial-Temporal associations; and (2) Reptile-CMDPs, a meta-reinforcement learning approach that enables rapid adaptation across diverse manufacturing contexts with theoretical guarantees for convergence. Evaluations are driven by realistic attack vectors from the Edge-IIoTset dataset and an open source factory simulator, demonstrating that our context-embedded model STKG-PPO improves production efficiency by 52.5% and reduces convergence time by 6.7% compared to context-agnostic baselines. Furthermore, our meta-learning approach Reptile-CMDPs accelerates adaptation, achieving 95.2% higher average rewards compared to training from scratch.

Vehicular Fog Computing (VFC) is significantly enhancing the efficiency, safety, and computational capabilities of Intelligent Transportation Systems (ITS), and the integration of Unmanned Aerial Vehicles (UAVs) further elevates these advantages by incorporating flexible and auxiliary services. This evolving UAV-integrated VFC paradigm opens new doors while presenting unique complexities within the cooperative computation framework. Foremost among the challenges, modeling the intricate dynamics of aerial-ground interactive computing networks is a significant endeavor, and the absence of a comprehensive and flexible simulation platform may impede the exploration of this field. Inspired by the pressing need for a versatile tool, this paper provides a lightweight and modular aerial-ground collaborative simulation platform, termedAirFogSim. We present the design and implementation of AirFogSim, and demonstrate its versatility with five key missions in the domain of UAV-integrated VFC. A multifaceted use case is carried out to validate AirFogSim's effectiveness, encompassing several integral aspects of the proposed AirFogSim, including UAV trajectory, task offloading, resource allocation, and blockchain. In general, AirFogSim is envisioned to set a new precedent in the UAV-integrated VFC simulation, bridge the gap between theoretical design and practical validation, and pave the way for future intelligent transportation domains. Our code will be available athttps://github.com/ZhiweiWei-NAMI/AirFogSim.

The highly dynamic Low Earth Orbit (LEO) environment poses significant challenges for efficient and heterogeneous service provision. A promising approach is to use the communication links of LEO satellites as signals of opportunity, where opportunistic offloading techniques appear to improve the overall performance. However, current solutions often overlook the sophisticated predictive capabilities to exploit spatio-temporal correlation across multiple dimensions (e.g., link quality, node capacity, task requirements), failing to evaluate the quality of transient opportunities for offloading decisions. This paper proposes a Spatio-Temporal Transformer-based Opportunistic Task Offloading (STOTO) approach for heterogeneous tasks in dynamic LEO. The core of STOTO is a spatio-temporal Transformer prediction model, designed to achieve superior awareness of evolving service requirements and resource availability. Then, the scheme dynamically evaluates the situation and resource availability, allocating tasks to the optimal nodes in real time. Experimental results demonstrate that our approach significantly outperforms existing methods, achieving 8.4% higher task completion rates on average.

Achieving advanced UAV autonomy is increasingly pivotal in complex low-altitude environments, while developing such AI-driven autonomous capabilities critically depend on benchmarking suites. However, existing solutions often address fundamental, isolated domains (e.g., network-centric, traffic-centric), failing to capture the intricate, comprehensive mission-level interactions in low-altitude operations. This paper introduces AirFogSim, a high-fidelity simulation platform for validating UAV autonomy in low-altitude missions. The cores of AirFogSim are a novel workflow-agent-state (WAS) modeling approach and standardized APIs for external data sources, empowering AirFogSim to simulate diverse UAV behaviors under realistic, dynamic environmental conditions. To demonstrate the practical utility, we generate a benchmark dataset for UAV situation awareness with varied weather and electromagnetic interference conditions. Multiple AI techniques are implemented and evaluated on this dataset, including traditional machine learning algorithms and large language models (LLMs). The platform is currently open-sourced on GitHub.

Modern satellite networks are transitioning from monolithic designs to microservice architectures, introducing complex spatio-temporal patterns, varied on-board processing capabilities, and high mobility with dynamic topologies. These features require accurate prediction of inter service dependencies for optimal resource allocation and system management. To address these challenges, this paper introduces the Spatio-Temporal Graph Neural Network with Orbital contextual features (STGNN-O). This model incorporates orbital information as contextual features, processes spatial dependencies through multi-head graph attention networks, and captures temporal patterns at three different timescales to complete satellite service performence metrics prediction, which refers to forecasting key performance indicators of microservices running on satellite platforms. Also, due to the lack of real-world relavent dataset, a comprehensive satellite service benchmark dataset is created based on real-world parameters and service patterns across multiple orbital configurations. Experiments demonstrate that STGNN-O significantly outperforms state-of-the-art baselines, achieving substantial improvements in prediction accuracy. Ablation studies confirm that the integration of orbital information and multi-scale temporal features significantly contributes to prediction accuracy across all orbital regimes.

Mobile Edge Computing (MEC) is a promising technology for future 6G communication systems. However, the dynamic network environment and the selfish nature of devices pose challenges to task offloading. Therefore, it is very critical to design an effective cooperative offloading scheme in dynamic environments. In this paper, a social-assisted two-stage cooperative task offloading and resource allocation algorithm based on Stackelberg game and DRL (SAC-SDRL) is proposed to maximize the system utility. The problem is formulated as a mixed integer non-linear programming (MINLP) problem that jointly determined the edge server selection, and offloading rate, resource price, and resource allocation. To address this problem, two stage-solutions are introduced. In the first stage, given a fixed resource price and offloading rate, the edge server selection and resource allocation scheme based on hybrid actor-critic algorithm is designed to solve the problem of hybrid action space. In order to avoid invalid decision space, a clustering method based on social relationship and spectral clustering is developed. In the second stage, based on the obtained edge server selection and resource allocation decision, a dynamic pricing and offloading incentive scheme based on the Stackelberg game is proposed, in which the optimal resource price and optimal offloading rate can be determined with the proposed gradient-based iterative search method. Moreover, it is proved that the game can achieve the Stackelberg equilibrium. Finally, simulation results show that the proposed SAC-SDRL algorithm can achieve higher system utility compared with other concerned algorithms.

3D terrain reconstruction with satellite imagery achieves cost-effective and large-scale earth observation and is crucial for safeguarding natural disasters, monitoring ecological changes, and preserving the environment. Recently, learning-based multi-view stereo (MVS) methods have shown promise in this task. However, these methods simply modify the general learning-based MVS framework for height estimation, which overlooks the terrain characteristics and results in insufficient accuracy. Considering that the Earth’s surface generally undulates without drastic changes and can be measured by slope, integrating slope considerations into MVS frameworks could enhance the accuracy of terrain reconstruction. To this end, we propose an end-to-end slope-aware height estimation network named TS-SatMVSNet for large-scale remote sensing terrain reconstruction. To effectively obtain the slope representation, drawing from mathematical gradient concepts, we innovatively proposed a height-based slope calculation strategy to first calculate a slope map from a height map to measure the terrain undulation. To fully integrate slope information into the MVS pipeline, we separately design two slope-guided modules to enhance reconstruction outcomes. Specifically, we designed a slope-guided interval partition module for refined height estimation using slope values. And, a height correction module is proposed, using a learnable Gaussian smoothing operator to amend the inaccurate height values. Additionally, to enhance the efficacy of height estimation, we proposed a slope direction loss for implicitly optimizing height estimation results. Extensive experiments on the WHU-TLC dataset and MVS3D dataset show that our proposed method achieves state-of-the-art performance and demonstrates competitive generalization ability compared to all listed methods. Our code will be available at https://github.com/StriveZs/TS-SatMVSNet.

The evolving Industrial Internet of Things (IIoT) is shifting towards decentralized collaborative manufacturing, posing heightened network security issues within interconnected value chains, thus requiring advanced Network Intrusion Detection (NID) systems to identify potential threats. In this context, traditional centralized NID systems are insufficient due to cross-industrial privacy concerns and interconnected secure threats. Federated Learning (FL) has emerged as a promising solution to enable the sharing of security insights without compromising privacy across participants. However, establishing an FL-based NID framework in realistic IIoT scenarios faces several hurdles, including the limited availability of large-scale devices and heterogeneous attack data distributions. The former leads to inconsistent client participation and degraded performance, while the latter hinders model convergence. To address these, we propose a novel Federated Learning-based Industrial Network Intrusion Detection (FedIn-NID) framework, incorporating a multidimensional client selection strategy and a dynamic global aggregation strategy. The selection strategy synergistically considers multidimensional factors including client availability, local dataset distribution, and dataset size. This approach accommodates clients with varying availability and avoids the selection of biased clients with data concentrated in a few categories. During model aggregation, the proposed strategy leverages the concept of exponential moving average to dynamically balance the holistic yet slightly older knowledge in the global model with the partial but relatively newer knowledge in local models, ensuring effective aggregation and convergence of the global NID model. Experiments demonstrate that FedIn-NID outperforms baselines by 10% to 30%, showcasing remarkable robustness with increasing data distribution heterogeneity and device count.

Federated semi-supervised learning (FSSL) has recently emerged as a promising approach for enhancing the performance of federated learning (FL) using ubiquitous unlabeled data. However, this approach encounters challenges when learning a global model using both fully labeled and fully unlabeled clients. Previous works overlook the dissimilarities between labeled and unlabeled clients, predominantly using shared parameters for local training across these two types of clients, thereby inducing intertask interference during local training. Moreover, these works typically adopt a single-perspective aggregation strategy, primarily focusing on data-volume-aware aggregation (i.e., FedAvg), leading to a lack of comprehensive consideration in model aggregation. In this article, we propose a novel FSSL method termed $\text {D}^{{2}}\text {Fed}$ , which addresses these issues by rethinking the roles of labeled clients and unlabeled ones to mitigate intertask interference during local training and by integrating client-type-aware with data-volume-aware to provide a more comprehensive perspective for model aggregation. Specifically, in local training, our proposed $\text {D}^{{2}}\text {Fed}$ distinguishes between the primary and accessory roles of labeled and unlabeled clients, respectively, performing dual-role additive local training (DALT) accordingly. In global aggregation, $\text {D}^{{2}}\text {Fed}$ uses a dual-perspective global aggregation (DGA) strategy, transitioning from data-volume-aware aggregation to client-type-aware aggregation. The proposed method simultaneously improves both local training and global model aggregation for FSSL without compromising privacy. We demonstrate the effectiveness and robustness of the proposed method through extensive experiments and elaborate ablation studies conducted on the CIFAR-10/100, SVHN, FMNIST, and STL-10 datasets. Experimental results show that $\text {D}^{{2}}\text {Fed}$ outperforms state-of-the-arts on five datasets under diverse data settings.

Network slicing is the key enabler for the 5G Industrial Internet of Things (IIoT), allowing tailored services and security guarantees for vertical industries. With the advent of 5G-Advanced (5G-A) and 6G era, the number of slices will increase significantly, leading to more diverse security requirements given different slice features. To provide adaptive security management spanning multiple slices in IIoT, this paper proposes a novel slice-specific secure IIoT (SSIOT) architecture with an AI-enabled solution. The SSIOT architecture separates the control and data planes, where the control plane orchestrates the Security Service Function Chains (SSFC) across network slices and the data plane analyzes the slice-specific features like traffic patterns, resource SLA guarantees, and Virtual Security Network Function (VSNF) dependencies. To extract these spatial-temporal features from the dynamic IIoT environments, we facilitate the powerful deep reinforcement learning (DRL) methods and propose a structural GS2L approach. GS2L is maliciously designed with the core principles of graph convolutional network (GCN) and Gated Recurrent Unit (GRU), enabling a thorough understanding of physical resource distribution and the request dynamics across slices. Extensive experiments are conducted in diverse IIoT slices with the real-world USNet and fat-tree topologies. Simulation results demonstrate that GS2L outperforms state-of-the-art learning and heuristic benchmarks, showcasing an overall 15.2% improvement with efficient and stable resource utilization.

In this paper, we focus on the task offloading problem for zero-trust vehicular fog computing (VFC) to promote trustworthy collaborative computing among vehicles. We propose a blockchain-enabled zero-trust VFC framework (BlockZT-VFC) to continuously verify and dynamically authorize the vehicle nodes. To overcome the reliability and efficiency issues in BlockZT-VFC, a multi-attribute task offloading and group-based continuous verification (MTOCV) scheme is designed. In particular, multiple attributes are extracted from both vehicles and tasks, ensuring that tasks are offloaded to FVs with authorized trustworthiness, and the group-based continuous verification efficiently guarantees the resulting authenticity by evaluating independent probability mathematically. Experimental simulations demonstrate an 18% increase in throughput and a 34% decrease in latency across two different scenarios, showcasing the superiority of our mechanism in improving system performance and reliability.

The Industrial Internet of Things (IIoT) continues to evolve alongside advancements in 5G and beyond 5G communication technologies, and network slicing has emerged as a promising technique to offer isolated slices and tailored services across industrial use cases, which profoundly affects the traditional security solution. As the future IIoT evolves towards the post-5G era, the anticipated growth in network slices will unavoidably exacerbate challenges in security service management, but developing an adaptive and effective security strategy remains a significant challenge. This study introduces a novel slice-specific IIoT (SSIOT) architecture, meticulously crafted to address the unique security needs of each slice based on network function virtualization. To adapt to the SSIOT, an AI-driven GS2L model is presented, which combines graph convolutional network (GCN) with sequence-to-sequence (Seq2Seq) deep reinforcement learning (DRL). The GS2L offers the network topology explanation module, service request analysis module, and slice-specific distribution extraction module, and adeptly navigates the multifaceted Security Service Function Chain (SSFC) embedding conundrum and resource allocation in slice-specific systems. Comprehensive experimental evaluations underscore GS2L's superiority, showcasing its proficiency in delivering augmented QoS satisfaction while ensuring prudent resource utilization over the learning-based and heuristic benchmark.

Deep neural networks have achieved promising progress in remote sensing (RS) image classification. However, the training process requires abundant samples for each class, and it is unrealistic to annotate labels for each RS category, especially considering that the RS target database is increasing dynamically. Therefore, we introduce an innovative prototype network tailored for Generalized Category Discovery (GCD) in remote sensing scene classification. This network consists of two essential modules: one dedicated to representation learning and the other to prototype learning. Through extensive experiments conducted on three benchmark datasets, i.e., RSS-DIVCS, NWPU-RESISC45, and AID, we demonstrate that the proposed model achieves remarkable performance gain up to 20%, effectively addressing the challenges inherent in classifying dynamically varying remote sensing images.

The rise of collaborative manufacturing, driven by the rapid proliferation of Industrial Internet of Things (IIoT) technologies, has markedly enhanced agility and productivity in industrial environments. However, this advancement has also significantly broadened the attack surface and uncovered unique vulnerabilities intrinsic to these interconnected systems. This paper introduces a novel Hierarchical Federated Incremental Learning Network Intrusion Detection (HFIN) approach. To our knowledge, this is the first attempt to address the ever-evolution network intrusion detection (NID) challenges in IIoT landscapes from the continuous attack-defense perspective. Our proposed HFIN capitalizes on decentralized model training across multifarious IIoT devices, ensuring data privacy and empowering continuous learning capabilities. It utilizes distributed data sources for secure experience sharing, collaboratively enhancing the continuous detection performance of the global model. Furthermore, regarding the inherent resource constraints of IIoT devices, we proposed a novel edge-client Weighted Transmission Optimization strategy (WTO). This strategy adeptly balances effective intrusion detection with the operational constraints of IIoT devices. By holistically considering detection capabilities and data volume across different attack types, it prioritizes the transmission of more critical and scarce attack data for training within bandwidth constraints. This maintains the comprehensive detection capability of the global model against various network attacks. To validate the effectiveness of HFIN, we conduct extensive experiments using the NF-UQ-NIDS-v2 and NF-ToN-IoT-v2 datasets. Experimental results demonstrate that our method outperforms baselines by approximately 10% in terms of accuracy and F1-score, highlighting the applicability and effectiveness of HFIN in enhancing security against sophisticated industrial environments and ever-evolving cyber threats.

Vehicular fog computing (VFC) has emerged as a promising solution to mitigate vehicular network computation load. In the hierarchical VFC, vehicles are employed as mobile fog nodes at the edge to provide reliable and low-latency services. Particularly, since privately-owned vehicles are rational nodes, their intentions for both computation provision and service demand should be considered instead of overestimating their willingness. To remunerate the participation intentions of vehicles as well as improve vehicular fog resource utilization in the large-scale VFC, the trading-based mechanism is a potential solution. In this article, we propose a many-to-many task offloading framework based on the vehicular trading paradigm. This framework enables computational resource trading across different VFC subsystems and decides the multi-tier task offloading results based on the trading consensus. The trading process is viewed as a partially observable Markov decision process (POMDP) and a Multi-Agent Gated actor Attention Critic (MA-GAC) approach is designed to reach an effective and stable offload-and -serve cooperation among vehicles. Theoretical analyses and experiments verify the feasibility and efficiency of the proposed framework, and simulation results demonstrate that the coordinated MA-GAC approach not only benefits vehicles with higher long-term rewards but also optimizes the system social welfare in a distributed manner.

Vehicular fog computing (VFC) has been expected as a promising architecture that can make full use of computing resources of idle vehicles to increase computing capability. However, most current VFC architectures only focus on the local region and ignore the spatio-temporal heterogeneity of computing resources, resulting in that some regions have idle computing resources while others cannot satisfy the requirements of tasks. To further improve the overall computing resource utilization in the whole network, in this work, we propose a hierarchical VFC architecture, where neighboring regions can share their idle computing resources. Considering the high complexity of both inter- and intra-region cooperative task offloading in such a hierarchical VFC architecture, we put forward a distributed task offloading strategy based on multi-agent reinforcement learning in which the multi-agent reinforcement learning method is designed to learn each task vehicle’s offloading strategy in a distributed manner. Moreover, to tackle the inefficiency caused by the multi-agent credit assignment problem, we provide the counterfactual multi-agent reinforcement learning approach which exploits a counterfactual baseline to evaluate the action of each agent. Simulation results validate that the proposed hierarchical VFC architecture can effectively improve the global task computing efficiency and the proposed mechanism outperforms the baseline algorithms.

Vehicular fog computing (VFC) can make full use of computing resources of idle vehicles to increase computing capability. However, most current VFC architectures only focus on the local region and ignore the spatio-temporal distribution of computing resources, resulting that some regions have idle computing resources while others cannot satisfy the requirements of tasks. Therefore, we propose a hierarchical VFC architecture, where neighboring regions can share their idle computing resources. Considering that the existing centralized offloading mode is not scalable enough and the high complexity of cooperative task offloading, we put forward a distributed task offloading strategy based on multi-agent reinforcement learning. Moreover, to tackle the inefficiency caused by the multi-agent credit assignment problem, we provide the counterfactual multi-agent reinforcement learning approach which exploits a counterfactual baseline to evaluate the action of each agent. Simulation results validate that the hierarchical architecture and the distributed algorithm improves the efficiency of global performance.

Electric vehicles (EVs), one of the most effective solutions to reduce gas emission and realize fossil fuels replacement, are enjoying growing popularity from governments to customers. The development of EVs leads to significant advances in vehicle automation and electrification, but meanwhile poses additional heavy charging and data processing burden on current smart grid. Considering the mutual demand and supply relationship between EVs and smart grid in both charging and computing tasks, we integrate vehicular fog computing (VFC) and smart EV charging for joint optimization and propose an integrated charging and computing (IC2) architecture for EV-included smart grid. In the proposed IC2 architecture, charging stations are profit-driven third-party power prosumers that also help compute tasks offloaded by smart grid while EVs act as both energy consumers and computation providers. We employ the contract theory to provide a multiattribute contract-based charging protocol for EVs and charging stations in an information asymmetry scenario. To obtain the optimal contract, we derive KKT conditions and design a convex–concave-procedure-based contract optimization algorithm. We also design a heuristic offloading algorithm to assign heterogeneous tasks toward different EVs. Numerical results indicate that the proposed multiattribute contract-based charging-computing scheme can effectively benefit both the charging stations and EVs, and meanwhile improves the task computation capability in EV-integrated smart grid.

Vehicular fog computing (VFC) is regarded as a promising framework for vehicular computing applications by utilizing local spare resources of nearby vehicles to conduct ubiquitous time-critical and data-intensive tasks. Meanwhile, how to provide stable and low-latency services through real-time task offloading has become a heated issue. Opposite to the traditional task-to-individual offloading manner, in this article, we propose a novel task-block (TB)-based offloading paradigm for VFC, in which the tasks are merged into blocks to be assigned and offloaded. This TB-based offloading paradigm effectively alleviates the offloading decision-making burden and, thus, reduces the overall computation latency in the dynamic vehicular environment. Faced with transmission-reliable and time-intensive requirements of TBs, we turn to cooperation among vehicles and further propose a TB-based overlapping matching-coalition (TBOMC) scheme integrating overlapping coalition formation (OCF) game with matching theory to address the complicated offloading problem. The OCF game framework encourages vehicular fog nodes to devote their resources and form collaborative computing groups in a distributed method. Numerical results demonstrate that the TBOMC scheme better exploits local computing capabilities and outperforms from 5% to 12% over other existing benchmarks.

With increasing time-critical and computation-intensive tasks generated by mobile applications, vehicular fog computing (VFC) has emerged as a promising solution to relieve the overload on roadside units (RSUs) or cloud centers. In VFC, tasks are offloaded to vehicular fog nodes local to the client devices, which exploits the under-explored computational resources of nearby vehicles. In this paper, we propose a novel cooperative vehicular fog computing architecture from an overlapping perspective, termed Overlapping-enabled Cooperative Vehicular fog Computing (OCVC) to fully utilize vehicular fog nodes' local potential resources. Different from traditional cooperative VFC architecture where each vehicle only works in one fog computing group at one time, the proposed OCVC architecture enables vehicles to participate in different computing groups simultaneously, and thus is able to fully exploit potential computational resources in an overlapping manner. In addition, we provide a distributed OCVC scheme to solve the complicated computing group formation, overlapping resource allocation, and task assignment problem by employing the overlapping coalition formation (OCF) game framework and a heuristic offloading algorithm. We conduct simulations for performance comparison in terms of diversified performance metrics and numerical results show that the proposed OCVC scheme performs better than other benchmarks under different conditions.

Electric vehicles (EVs) are enjoying growing popularity from governments to customers. However, the development of EVs inevitably poses heavy charging and data processing burden on current smart grid. Considering the mutual demand and supply relationship between EVs and smart grid in both charging and computing tasks, we integrate vehicular fog computing and EV charging for joint optimization and propose an integrated charging-computing

Confronted with the increasing computation-intensive requirements of vehicular applications, vehicular fog computing (VFC) has emerged as the promising solution to mitigate the load at the edge of vehicular network. In VFC, vehicles are employed as vehicular fog nodes to provide reliable services with applicability. However, considering the individual serving and offloading intentions of the privately-owned vehicles, the many-to-many task offloading in dynamic vehicular environment becomes a challenging problem. In this paper, we propose a distributed dynamic many-to-many task offloading framework based on vehicle-to-vehicle (V2V) trading paradigm to improve the fog resource utilization in VFC. In order to reach an effective and stable offload-and-serve cooperation between vehicles as service demanders and vehicles as computation providers in the proposed framework, we formulate the trading process as a partially observable Markov decision processes (POMDP) and design a Multi-Agent Gated actor Attention Critic (MA-GAC) approach, leading to an efficient offloading optimization process in a distributed manner. Theoretical analysis and experiments verify the feasibility and efficiency of the proposed framework, and simulation results demonstrate that the proposed MA-GAC approach outperforms other benchmarks in the dynamic environment.

Vehicular fog computing (VFC) has emerged as a promising solution to relieve the overload in vehicular network. Since individual vehicular fog node is incapable of providing ultra-reliable and low-latency services constrained by limited resources, cooperation among vehicles becomes an attractive attempt to promote quality of service (QoS). In this paper, we propose a novel Overlapping-enabled Cooperative Vehicular Computing architecture in VFC, termed OCVC, to fully utilize vehicular fog nodes' local potential resources. The proposed OCVC architecture enables vehicles to participate in different fog groups simultaneously different from traditional cooperative computing architecture. In addition, we propose a distributed OCVC scheme to solve the complicated computing group for-mation, overlapping resource allocation, and task assignment problem based on overlapping coalition formation (OCF) game framework. We conduct experiments in several metrics and numerical results show that the proposed OCVC scheme per-forms at least 5 % better than other benchmarks under different conditions.

The existing methods of the automatic extraction of topographic feature lines from terrain representation either have too high sensitivity to terrain noise or lose significant branches. In this study, we present new multi-criterion methods to extract topographic feature lines from contours on different topographic gradients according to the negative or positive signs of the curvature and neighboring feature points on the contours, the hierarchical structure of these feature points, and the spatial relationships between topographic feature lines and contours. First, the digitization directions of source contours were automatically detected and adjusted (when necessary) to establish the spatial relationships among the contours before we extract and group the feature points in the terrain. Second, we determine the ‘mainstreams’ and ‘tributaries’ of the topological structure trees according to the relationships among the previously identified feature point groups. Finally, a key aspect of our paper is the proposition of multi-criterion methods to extract topographic feature lines. Compared with the regular square grids (RSG)-based and Voronoi skeleton-based methods, the proposed methods can extract topographic feature lines with higher accuracy, better continuity, lower spatial logical conflicts between topographic feature lines and contours.

As a research hotspot of computer vision, crowd counting methods have achieved success in natural images. But crowd counting in aerial images are rarely explored, and existing methods do not perform well because of the higher resolution, smaller object scale and more complex scene. Therefore, this paper proposes a lightweight dual-task network (LDNet) for crowd counting, which only uses bifurcated structure to overcome these new challenges in aerial images without complicated pipelines. To realize this, a complete but efficient Guidance Branch is proposed to assist Counting Branch in fitting crowd distribution. Furthermore, a scene attention mechanism is used to consider the complex scene information, which are never considered by existing methods. Our LD-Net outperforms existing methods on aerial crowd counting dataset (Visdrone), and gets better or comparable results on natural crowd counting datasets (UCF_CC_50, UCF_QNRF, ShanghaiTech Part A).