Search

At present, terrorism has become an important factor affecting world peace and development. As the time series data of terrorist attacks usually show a high degree of spatial–temporal correlation, the spatial–temporal prediction of casualties in terrorist attacks is still a significant challenge in the field of counter-terrorism. Most of the existing terrorist attack prediction methods lack the ability to model the spatial–temporal dynamic correlation of the time series data of terrorist attacks, so they cannot yield satisfactory prediction results. In this paper, we propose a novel Attention-based spatial–temporal multi-graph convolutional network (AST-MGCN) for casualty prediction of terrorist attacks. Specifically, we construct the spatial adjacency graph and spatial diffusion graph based on the different social-spatial dynamic relationships of terrorist attacks and determine the multi-scale period of time series data of terrorist attacks by using wavelet transform to model the temporal trend, period and closeness properties of terrorist attacks. The AST-MGCN mainly consists of spatial multi-graph convolution for extracting social-spatial features in multi-views and temporal convolution for capturing the transition rules. In addition, we also use the spatial–temporal attention mechanism to effectively capture the most relevant spatial–temporal dynamic information. Experiments on public datasets demonstrate that the proposed model outperforms the state-of-the-art baselines.

Although numerous spatiotemporal approaches have been presented to address the problem of missing spatiotemporal data, there are still limitations in concurrently capturing the underlying spatiotemporal dependence of spatiotemporal graph data. Furthermore, most imputation methods miss the hidden dynamic connection associations that exist between graph nodes over time. To address the aforementioned spatiotemporal data imputation challenge, we present an attention-based message passing and dynamic graph convolution network (ADGCN). Specifically, this paper uses attention mechanisms to unify temporal and spatial continuity and aggregate node neighbor information in multiple directions. Furthermore, a dynamic graph convolution module is designed to capture constantly changing spatial correlations in sensors utilizing a new dynamic graph generation method with gating to transmit node information. Extensive imputation tests in the air quality and traffic flow domains were carried out on four real missing data sets. Experiments show that the ADGCN outperforms the state-of-the-art baseline.

Purpose The purpose of this study is to establish an effective tracking algorithm for small unmanned aerial vehicles (UAVs) based on interacting multiple model (IMM) to take timely countermeasures against illegal flying UAVs. Design/methodology/approach In this paper, based on the constant velocity model (CV), the maneuvering adaptive current statistical model (CS) and the angular velocity adaptive three-dimensional (3D) fixed center constant speed rate constant steering rate model, a small UAV tracking algorithm based on adaptive interacting multiple model (AIMM-UKF) is proposed. In addition, an adaptive robust filter is added to each model of the algorithm. The linear Kalman filter algorithm is attached to the CV model and the CS model and the unscented Kalman filter algorithm (UKF) is attached to the CSCDR model to solve the nonlinearity of the 3D turning model. Findings Monte-Carlo simulation comparison with the other two IMM tracking algorithms shows that in the case of different movement modes and maneuvering strength of the UAV, the AIMM-UKF algorithm makes a good trade-off between the amount of calculation and filtering accuracy, which can maintain more accurate and stable tracking and has strong robustness. At the same time, after testing the actual observation data of the UAV, the results show that the AIMM-UKF algorithm state estimation trajectory can be regarded as an actual trajectory in practical engineering applications, which has good practical value. Originality/value This paper presents a new small UAV tracking algorithm based on IMM and the advantages and practicability of this algorithm compared with existing algorithms are proved through experiments.

The dynamic recommender system realizes the real-time recommendation for users by learning the dynamic interest characteristics, which is especially suitable for the scenarios of rapid transfer of user interests, such as e-commerce and social media. The dynamic recommendation model mainly depends on the user-item history interaction sequence with timestamp, which contains historical records that reflect changes in the true interests of users and the popularity of items. Previous methods usually model interaction sequences to learn the dynamic embedding of users and items. However, these methods can not directly capture the excitation effects of different historical information on the evolution process of both sides of the interaction, i.e., the ability of events to influence the occurrence of another event. In this work, we propose a Dynamic Graph Hawkes Process based on Linear complexity Self-Attention (DGHP-LISA) for dynamic recommender systems, which is a new framework for modeling the dynamic relationship between users and items at the same time. Specifically, DGHP-LISA is built on dynamic graph and uses Hawkes process to capture the excitation effects between events. In addition, we propose a new self-attention with linear complexity to model the time correlation of different historical events and the dynamic correlation between different update mechanisms, which drives more accurate modeling of the evolution process of both sides of the interaction. Extensive experiments on three real-world datasets show that our model achieves consistent improvements over state-of-the-art baselines.

Graph Convolutional Networks (GCNs) are powerful deep learning methods for non-Euclidean structure data and achieve impressive performance in many fields. But most of the state-of-the-art GCN models are shallow structures with depths of no more than 3 to 4 layers, which greatly limits the ability of GCN models to extract high-level features of nodes. There are two main reasons for this: 1) Overlaying too many graph convolution layers will lead to the problem of over-smoothing. 2) Graph convolution is a kind of localized filter, which is easily affected by local properties. To solve the above problems, we first propose a novel general framework for graph neural networks called Non-local Message Passing (NLMP). Under this framework, very deep graph convolutional networks can be flexibly designed, and the over-smoothing phenomenon can be suppressed very effectively. Second, we propose a new spatial graph convolution layer to extract node multiscale high-level node features. Finally, we design an end-to-end Deep Graph Convolutional Neural Network II (DGCNNII) model for graph classification task, which is up to 32 layers deep. And the effectiveness of our proposed method is demonstrated by quantifying the graph smoothness of each layer and ablation studies. Experiments on benchmark graph classification datasets show that DGCNNII outperforms a large number of shallow graph neural network baseline methods.

For the speed of traditional SIFT algorithm in the feature extraction and matching is slow, the article proposes an improved RANSAC features image matching method based on speeded up robust features (SURF). First of all, detect images features and extract with SURF method, use the fast library for approximate nearest neighbours-based matcher method to perform initial matching on image feature points. Improve the RANSAC algorithm to increase the probability of correct matching points being sampled. Experimental results show that the improved RANSAC algorithm has high matching accuracy, good robustness, and short running time. It lays the foundation for the subsequent fast image stitching.

Key population control is related to national security and social stability. Aiming at the difficulty of extracting key population information entities, this paper proposes a sequence labeling method based on dynamic word vectors. Replace the Word2vec pre-training model with the BERT model, which strengthens the feature extraction capabilities of the traditional entity extraction model, and more fully describes the multiple semantic and syntactic information of words. The improvement ideas for the BiLSTM-CRF model are as follows: Embed the BERT model upstream of the model, which is responsible for converting the original corpus into a dynamic vectorized representation, and the trained word vector is input into the BiLSTM layer for semantic encoding, and further mining the semantic related features of the entity context, and finally, the CRF layer outputs the sequence label with the maximum probability. After training with key population information as a data set, the F1 value of the model reached 0.90.

Aiming at the problem of event prediction in large-scale event network, a collapse subgraph convolution (CSGCN) algorithm is proposed, which uses event subgraph to predict the subsequent events of event group. CSGCN algorithm collapses the edge induced event subgraph in large-scale event network, removes the irrelevant event nodes from the subgraph, and forms a new event subgraph. GCN algorithm is used to learn the graph embedding representation of the event subgraph, and the subsequent events of the event group are predicted by comparing the similarity between the graph embedding representation of the event group and the subsequent events. Because only some related nodes are processed each time, the application of the model in large-scale data graph is feasible. Through experiments, we explore and verify the effectiveness of extracting features from subgraphs of large-scale graph by using graph convolution training to obtain graph embedding representation. We find that GCN has better event prediction effect than Euclidean distance and co rotation similarity, which further shows that graph convolution algorithm has good performance in the field of graph feature extraction.

The occurrence of mass incidents has an important impact on social and public security. The inspiring and polarization of people’s emotions is an important factor affecting the generation and deterioration of mass incidents. Put forward an emotional interaction computation model of actors in the mass incidents combining the group emotional contagion theory with the agent-based modeling method. In the modeling process, pick up some major factors such as emotional value, emotional expression degree, sensitivity, and simulate the emotional interaction of actors. A large number of simulation results show that in the progress of emotional interaction of actors in the mass incidents there is individual emotional spiral phenomenon. It directly leads to group emotional polarization, and affects the actors’ violent behavior. Group emotional contagion is an important factor to promote the development of the incident.

In order to analyze the network structure of the “East Turkistan” organization and identify members with higher individual threats, so as to conduct a better attack on the “East Turkistan” organization. We use Gephi, a network analysis software, and collect open source information on three batches of 25 terrorists announced by the Ministry of Public Security and use networkX, a Python package to study the structure of complex networks, to conduct social network analysis(SNA) of the “East Turkistan” organization personnel. The “East Turkistan” organization presents a combination of scale-free network and hierarchical network in the topological structure. To manifest the analysis of social network methods' importance in combating terrorism, we simulate the effect of removing the core members of the “East Turkistan” organization. The construction of SNA requires a large deal of open source information. How to obtain intelligence is an important factor that restricts precision strikes. In addition, how to combine social network analysis with computer deep learning knowledge is also a new hot spot and direction in the field of anti-terrorism.

With the popularization of big data, public security organs are facing problems such as incomplete information collection, difficulty in fusion of multi-source heterogeneous data, and poor data association collected by front-end sensing devices in the management and control of targeted person. Combining the current situation and needs of public security management, this paper analyzes the targeted person's full-element information based on the Apriori association rule algorithm to conduct frequent item mining on the targeted person's attribute characteristics and behavior characteristics, which can effectively identify the abnormal behavior of targeted person and obtain deduction information, and finally implements a holographic archive system based on JavaEE design that can completely describe the targeted person. Its main functions include add files, file details, advanced search, “my focus”, “my supplement” and “my search”. After testing, the various functions of the system are operating well, which is conducive to the public security organs in real-time grasp of targeted person's basic information, social information, behavior information and other full-element information.

Fringe projection profilometry technique has attracted more and more attention in the computer vision and found wide applications in the areas of high-precision measurement. This paper presents an accurate and robust system calibration method to formulate the phase-to-height conversion. This method uses the Levenberg-Marquardt nonlinear optimization algorithm to fit the phase value of a virtual reference plane, which is parallel to the line between the optical centers of the camera and the projector. The final expression just has three coefficients. With no need to manually adjust the position of components, several experiments were conducted to evaluate the performance of proposed method and demonstrated that the novel system calibration outperforms state-of-the-art methods in terms of stability and accuracy.

In this paper, we proposed a weighted KNN algorithm based on random forests. The proposed algorithm fully measures the differences in the importance of each feature, and overcomes the shortcoming of k-nearest neighbor (KNN) algorithm in classifying unbalanced data sets and data sets of different feature importance. The classification accuracy of the KNN algorithm is effectively improved, and the performance of the proposed algorithm is verified through experiments.

The accuracy of camera calibration depends significantly on the control point detection which customarily subjects to image noise, distortion and resolution. This paper uses concentric circle grids as the calibration pattern and accordingly proposes a robust and accurate control point detection method for camera calibration. Each set of concentric circles on the grids pattern consists of multiple circles whose radii decrease progressively. The geometric structure of multi-hierarchies can not only robustly extract control points from complicated scenarios but also build a voting scheme to avoid numerical instability resulting from the estimation of projected centers. According to the common self-polar triangle of concentric circles and the local sub-pixel edge detection algorithm, we efficiently estimated sub-pixel coordinates of control points under projective transformation. Simulations and real experiments were conducted to demonstrate that the performance of calibration using concentric circle grids is superior to the popular grid patterns with respect to the image noise, distortion and resolution.

Mass incidents pose a serious threat to public order and national security, building a knowledge base in the field of mass incidents, visualizing the relationship between entities from incidents, and scientifically reasoning about key information about incidents have become key links in effectively responding to mass violence crimes. In view of the lack of existing mass incident knowledge graph analysis methods and strong subjective reasoning methods, based on the BiLSTM-CRF model, this paper analyzes event elements and entity associations by extracting entities from the text data of mass events, the protege software is used to construct the mass event ontology model, and the visualization of the mass event knowledge graph is realized based on the neo4j graph database. The research results show that the knowledge graph established in this paper can better reflect the correlation between the elements in the mass incident, and is of great significance to the national security department to improve the emergency decision-making ability.

Based on the development technology of WebGIS, the mobile terminal is accessed through the public mobile communication network, a cross-regional visual command and dispatch communication platform is set up. The platform can realize police officers' real-time location display, track replay and optimal path planning. At the same time, the platform integrates the Web instant messaging functions with geographic information system, and it can directly operate the target on the map to achieve more convenient and efficient command and dispatch functions. With the excellent regional coverage of the public mobile communication network, the platform can achieve cross-regional staff dispatching missions. The use of visual command and dispatch communication platform can effectively utilize the current limited manpower and material resources in emergency situations and maximize the capability of emergency handling and provide effective solution for cross-regional emergency command and dispatch tasks.

This paper mainly proposed a kind of algorithm based on moving object detection and direction recognition, with a variety of data in video surveillance system, constructing a moving objects trajectories system with a combination of artificial and video surveillance system, to find escape routes of suspects accurately and make work easy for investigators. This system was created by Python language, MySQL database language and Open CV direction recognition technology, completing usage of map, creating database structure and recognizing the direction of a moving object, to predict object moving direction.

At present, a person-specific 3D model generation technology has been an active research topic for scientists in the field of computer vision and computer graphics. The paper proposes a 3D modeling method based on single photograph, by analysis of individual difference like geometric features which present the shape and locations of facial components. An extension of the basic Active Appearance Model is proposed to localize some facial feature points in the frontal image and all these feature points are aligned to fit the generic 3D face model to a specialized one to reflect the given person's face, being both more robust and faster. Then an optimized texture mapping of cylindrical projection and model adjustment technologies are presented, to synthesize personal virtual face realistically. This method rectifies these previous defects, such as the difficulty in gaining access to acquiring the accurate data of a 3D face model, slowly manual location methods, incomplete facial information mining. The approach gives a significant improvement in both the reliability and the overall accuracy of 3D modeling.

This paper uses Agent-based modeling and simulation to study collective violent events from the perspective of individual's risk perception, by abstracting individual's risk preference, emotion factor, risk level and other factors we analyze and explore the influence of each factors to collective violent events. Considerable simulations indicate that, individual's density risk preference determines assemble morphology of the crowd and the event-relevance and emotion factor together determine the trend of events, while police's timely and effective intervention could prevent the spreading of the events. These conclusions provide meaningful theoretical basis for relevant organs.

Based on emotional contagion theory and combined emotional contagion scale with modeling and simulating research methods then put forward a computing model of individual emotion in the mass incidents with venting anger. In the model, emotional contagion scale is used to measure emotional contagion sensitivity of the agents. The agents have parameters such as five basic emotions, explicit emotions, comprehensive emotions and so on. A large number of simulation results show that the model can simulate the changes of individual basic emotions, the group basic emotions and the group comprehensive emotions in the process of emotional contagion.

This paper uses the method of Agent-Based Modeling and Simulation to analyze the interactions among participants in Mass Violence Events. We study the affection degree of instigators, activists, cops and civilians to individual civilians to attend the events. The results show that the influencing factors which could affect civilians to attend the events are agitation power, number of activists, number of cops and risk-avoidance and so on. It shows that the law of development of Mass Violence Events. These results could provide public security organs with some certain theoretical support.

This paper uses base-Agent method to build a new model of researching the evolution of discontent in mass violence event, in which model we introduce emotion factor, relation factor, risk factor and relevance factor as the attribute value of the individual Agent, then using psychophysics’ correlation theory to quantize the change of individual discontent. In the paper, by setting different initial values, we make relevant experiments to study the process of evolution of discontent of a typical mass violence event that has happened and three other different circumstances. All of what we achieved from this paper can offer help to the police to understand the inherent mechanism and law of mass violence event, also can provide guide for taking pointed measures to prevent and manage mass violence event.

This paper presents an agent-based collective model to study “non-organization and direct interest” collective events. The purpose of this model is to analyze the spread phenomenon emerging in this type of events and explain some important characters which affect the spread. The experimental results show that the degree of spread is influenced by some variables, and these variables also influence individual's motion intensity. Simulations reproduce the spread phenomenon of this type of events and indicate the weight of different factors which impact on individual's motion intensity, revealing the regular pattern of this event.

This paper introduces a new method for analyzing Mass Violence Events, and then we make use of Intelligent Artificial and Agent-Based System to deal with them. In the meanwhile, we describe and defy three kinds of agents (cops, activists and civilians), what comes next is that we apply these theories to model and simulate such events. The simulation results show that the behaviors of activists can easily affect whether civilians will join into Mass Violence Events, if they attend events, it will lead cops to face more difficulties when dealing with them. The new method can provide certain theoretical meanings for public security organs to address Mass Violence Events.

This paper uses Analytic Hierarchy Process(AHP) to analyze the probability of individual appearing acts of violence by the influence of different roles in the crowd of Mass Violence Event, obtaining the weight of influence of organization leaders, agitators, general participants and bystanders to individuals. This paper also introduces spatial heterogeneity to reflect different roles' influence decaying with distance. By means of experiment, we obtain the approximate probabilistic data of individuals appearing violence in different locations and personnel situations, depict the roughly curve of probability affected by crowd density in different personnel situations. Then we can determine the areas of high violence probability in certain environment, which provide data support for the police to dispose analogous events.

To formulate a reasonable and effective urban seismic traffic evacuation plan, this paper applied modified equivalent length, analysis of the factors which influence the seismic evacuation process, using the improved equivalent length to instead of the actual road network, this paper put forward an improved seismic evacuation dynamic traffic assignment model which is based on CTM. Experimental results show that the model can either be multiple origin-destination network optimization problem simplified into single origin-destination traffic assignment network optimization problem, and can be properly reflected the dynamic characteristics of the evacuation traffic flow, and can be satisfied the dynamic allocation demand of traffic evacuation simultaneously.

Before city emergencies happen, one of the most important tasks is to develop contingency plans, and reasonable analysis of space evacuation vulnerability is an important part of contingency plans. Based on the analysis of evacuation vulnerability of emergency area, this article develops a critical evacuation cluster model, and searches for the critical cluster of evacuation area by heuristic searching algorithm. At the end of the article, it generates the maps of space evacuation vulnerability in city emergency with the technology of GIS, which provide technical support for the making of contingency plans of city emergency.

This paper bases on the research of cellular automaton and Agent, proposing a Grid-Agent model. This model uses the grid as the basic unit, and the individual Agents in the grid implement their movement by judging the state of their own and the grid around to achieve the purpose of simulating the crowd behavior. This will overcome the insufficient of the cellular automaton by simulating the crowed behavior, and also overcome shortcomings that the lack of temporal logic the Agent-based. At last, this paper simulates the crowd-gathered behavior by using the model, and the results show that the Grid-Agent model is more accurate and logic in the simulation of human behavior, and we can observe and research the regulation of crowd-gathered behavior better.

During any extreme event, especially the evacuation in crowded places. The keys to saving lives are orderly and efficient evacuations. We have improved the traditional particle swarm optimization. It can simulate non-adaptive crowd behavior in the crowd facilities during the emergency evacuation. We introduce the concept of the leader (trained staff) and the parameter of influence, and analyzed the influence of the leader on the evacuation process, especially the factors of the leader's distribution, number and influence. It will provide theoretical support for the evacuation of High-density facilities.

This paper studied the regulation of mass event from the perspective of its individual, and proposed an agent-based mass event modeling. The individual behavior of aggregation, riots, fled in mass event was modeled. Then the model was applied to computer simulation for observing crowd behavior of all the individuals in the three states. By this, we can study the behavior rules of mass event.

This paper puts forward a method based on Logistic regression of emergency evacuation traffic trip generation forecasting model. Based on analysis of Xiamen residents earthquake evacuation survey data, the software of applied statistic SPSS could establish emergency evacuation traffic trip generation forecasting model by Logistic regression. This model could predict the probability of all evacuation modes of residents to obtain the better emergency processing plan of traffic evacuation in earthquake disasters. Subsequently, for the purpose of making the more reasonable and effective of city emergency management plan, the test sample would be simulated to accuracy and fitting degree of the prediction model result.

When city emergencies happen, one of the most important tasks of emergency rescue is to find the shortest path to the emergency in shortest time, scientific and rational path plays an important role in improving rescue efficiency. According to the needs of city emergency rescue and basing on the analysis of the classical algorithm for solving the shortest path problem (Dijkstra algorithm), this article discusses an improved algorithm within dynamic restricted searching area which dynamically changes the direction of the search and restricts the searching area. This new method reduces time and space complexity and improves the efficiency of the algorithm. At the end of the article, it improves the practicality and reliability of this improved algorithm by simulation.

During and after any extreme event of natural or man-made origin, the primary objectives are getting people out of harms way and to an safety area. The keys to saving lives are orderly and efficient evacuations. We have proposed a new model based on the traditional lattice gas automate, in our paper, it can simulate the human crowd behavior in the crowd facilities, such as shopping malls, stadiums, school buildings etc, under the condition that many people rush to exits at the same time. In this model, people move by their instinct and the information what they obtained. In order to reappearance the scenes of the events and reach a better evacuation. We introduce psychological factors, velocity levels of pedestrians and exit levels in our model.

On the basis of the characters of footprint shape, a toe shape description method based on geometric characteristic values of toe image is proposed. Corner detection is carried out on toe region, and the characteristic points which can describe the toe shape are confirmed by the edge of toe image. Through finding characteristic points whose distances to the center are stable and which can distinguish different toe shapes and the correlation among them, the toe shape characteristic vector is built. On the basis of toe shape characteristic vector, a toe shape recognition method based on fuzzy neural networks is proposed. Firstly, according to the statistical distribution disciplinarians of different shape toe images, membership functions are constructed respectively from the angle, length and region parameters, and these values are used as single judgment factors. The comprehensive judgment vector can be obtained through the operation among these single judgment factors. At last, the distance vector between comprehensive judgment vector and four model vectors are computed to feed into neural networks in order to judge. Because the method based on fuzzy neural network can reflect subjectively and correctly the different shapes of toe image, so the total automatic recognition rate amounts to 92.80%.

A feature-level image fusion method based on segmentation region and neural networks is proposed in this paper. Firstly, the source images are segmented and merged into a set of common regions which are used for guiding the whole fusion process; then selecting the corresponding segmentation regions from the source images respectively and extracting features representing clarity in the two regions; at last the features are fed into a neural networks to judge clear region to reconstruct the final fusion image. The experimental results show that the fusion effect is better.

The estimation of harmonic to noise ratio (HNR) in transmitted sound signals with a wavelet transform was proposed. When subjects phonate sustained vowels in breathy, falsetto, leakage and pressed mode in normal loudness, the HNR in the human voice and transmitted sound signals was estimated and compared. The results indicated that the HNR in transmitted sound signals could accurately image vocal fold vibration characteristics. It would be a useful measurement for clinical laryngeal disease diagnosis.