Security and protection method and device based on full-range feature recognition and four-dimensional trajectory tracking

Abstract

The invention provides a security and protection method and device based on full-range feature recognition and four-dimensional trajectory tracking, and relates to the technical field of security and protection, and the method comprises the steps: collecting a multi-modal data set of a target object in a target region through a data collection module; performing data fusion on the multi-modal data set through a feature recognition module to obtain a plurality of feature sequences, and performing individual feature matching and identity authentication to obtain individual behavior data of each target object; performing behavior anomaly detection on each target object based on the individual behavior data through a behavior analysis module; visualizing the four-dimensional moving track through a four-dimensional track reconstruction module, and adding a specified mark to the four-dimensional moving track of the target object with the abnormal behavior; and through the intelligent operation linkage module, determining the grade of the abnormal detection result and executing a corresponding target action. According to the technical scheme, various data sources can be integrated, and full-time-space full-scene monitoring is supported, so that the safety protection level of the park is improved.

Description

Technical Field

[0001] The present invention relates to the field of security technology, and in particular to a security method and device based on full-range feature recognition and four-dimensional trajectory tracking. Background Art

[0002] In park management, traditional security systems often have difficulty coping with complex and ever-changing security challenges. Especially with the continuous expansion of park scale and the increase in traffic density, how to effectively identify and track individual behavior has become an urgent problem to be solved. Most of the existing security methods rely on video surveillance and manual patrols with fixed cameras, which are not only inefficient, but also difficult to achieve real-time monitoring and rapid response of the entire park. In addition, when faced with emergencies, there is a lack of effective tools to quickly locate the location of suspicious persons and their activity trajectories, which limits the overall effectiveness of the security system. Summary of the invention

[0003] In view of this, the purpose of the present invention is to provide a security method and device based on full-range feature recognition and four-dimensional trajectory tracking, and to provide a technical solution that can integrate multiple data sources and support full-time, full-space and full-scene monitoring to improve the security protection level of the park.

[0004] In a first aspect, the present invention provides a security method based on full-range feature recognition and four-dimensional trajectory tracking, which is applied to a security device, wherein the security device includes a data acquisition module, a feature recognition module, a behavior analysis module, a four-dimensional trajectory reconstruction module, and an intelligent operation linkage module. The method includes: Through the data collection module deployed in the target area, data is collected from one or more target objects entering the target area to obtain a multimodal data set; Through the feature recognition module, the multimodal data set is fused to obtain multiple feature sequences, and individual feature matching and identity authentication are performed on each feature sequence to obtain the individual behavior data corresponding to each target object; Through the behavior analysis module, behavior anomaly detection is performed on each target object based on individual behavior data to obtain the behavior anomaly detection result corresponding to each target object. The behavior anomaly detection result is used to characterize whether the target object has abnormal behavior; Through the four-dimensional trajectory reconstruction module, the four-dimensional movement trajectory of the target object in the target area is visualized, and the four-dimensional movement trajectory corresponding to the target object with abnormal behavior is marked with a specified mark; Through the intelligent operation linkage module, the alarm and response levels corresponding to the abnormal detection results are determined, and the target actions corresponding to the alarm and response levels are executed. The target actions include one or more of alarm actions, multi-level response actions and equipment linkage disposal actions.

[0005] In one implementation, the multimodal data set includes access control sensing data, video surveillance data, and sensor data; and the multimodal data set is subjected to data fusion to obtain multiple feature sequences, including: Extracting a first facial feature corresponding to the access control sensing data and a second facial feature corresponding to the video surveillance data, and performing fuzzy matching on the first facial feature and the second facial feature to determine the correlation relationship between the access control sensing data and the video surveillance data; Based on the association relationship, the access control perception data, video surveillance data and sensor data are fused to obtain multiple feature sequences.

[0006] In one implementation, based on the association relationship, data fusion is performed on the access control sensing data, the video surveillance data, and the sensor data to obtain multiple feature sequences, including: Matching access control sensing data with sensor data, and matching video surveillance data with sensor data; If there is a correlation between the access control perception data and the video surveillance data matched by the sensor data, data fusion is performed on the sensor data and the access control perception data and the video surveillance data matched by the sensor data to obtain multiple feature sequences.

[0007] In one embodiment, individual feature matching and identity authentication are performed on each feature sequence to obtain individual behavior data corresponding to each target object, including: For any two feature sequences, determine the feature similarity between the two feature sequences, and determine that the two feature sequences belong to the same target object when the feature similarity is higher than a preset threshold; Based on the feature sequence belonging to the same target object, the individual behavior data corresponding to each target object is determined.

[0008] In one embodiment, a behavior abnormality detection is performed on each target object based on the individual behavior data to obtain a behavior abnormality detection result corresponding to each target object, including: Preprocessing the individual behavior data corresponding to each target object, and extracting key features corresponding to each target object from the preprocessed individual behavior data, the key features including behavior frequency and behavior duration; Determine the target behavior pattern to which each target object belongs through a pre-trained behavior pattern model based on a plurality of pre-configured candidate behavior patterns and key features corresponding to each target object; Based on the key features corresponding to each target object and the target behavior pattern to which it belongs, behavior anomaly detection is performed on each target object to obtain the behavior anomaly detection result corresponding to each target object.

[0009] In one implementation, based on the key features corresponding to each target object and the target behavior pattern to which it belongs, behavior anomaly detection is performed on each target object to obtain a behavior anomaly detection result corresponding to each target object, including: For any key feature corresponding to each target object, determine whether the value of the key feature is higher than the behavior threshold corresponding to the target behavior pattern to which the target object belongs, and determine that the target object has abnormal behavior if the judgment result is yes.

[0010] In one embodiment, visualizing the four-dimensional moving trajectory of the target object in the target area includes: Obtain the three-dimensional coordinate data of the target object moving in the target area and its associated timestamp information, and construct the four-dimensional movement trajectory corresponding to the target object; A three-dimensional model corresponding to the target area is established, and the four-dimensional movement trajectory is mapped into the three-dimensional model to realize the visualization of the four-dimensional movement trajectory.

[0011] In a second aspect, the present invention further provides a security device based on full-range feature recognition and four-dimensional trajectory tracking, the security device comprising a data acquisition module, a feature recognition module, a behavior analysis module, a four-dimensional trajectory reconstruction module and an intelligent operation linkage module: The data collection module deployed in the target area is used to: collect data from one or more target objects entering the target area to obtain a multimodal data set; The feature recognition module is used to: perform data fusion on the multimodal data set to obtain multiple feature sequences, and perform individual feature matching and identity authentication on each feature sequence to obtain the individual behavior data corresponding to each target object; The behavior analysis module is used to: perform behavior anomaly detection on each target object based on individual behavior data, and obtain a behavior anomaly detection result corresponding to each target object. The behavior anomaly detection result is used to characterize whether the target object has abnormal behavior; The four-dimensional trajectory reconstruction module is used to: visualize the four-dimensional moving trajectory of the target object in the target area, and add a specified mark to the four-dimensional moving trajectory corresponding to the target object with abnormal behavior; The intelligent operation linkage module is used to: determine the alarm and response levels corresponding to the abnormal detection results, and execute the target actions corresponding to the alarm and response levels. The target actions include one or more of alarm actions, multi-level response actions, and equipment linkage disposal actions.

[0012] In a third aspect, the present invention further provides an electronic device, comprising a processor and a memory, wherein the memory stores computer executable instructions that can be executed by the processor, and the processor executes the computer executable instructions to implement any one of the methods provided in the first aspect.

[0013] In a fourth aspect, the present invention further provides a computer-readable storage medium, which stores computer-executable instructions. When the computer-executable instructions are called and executed by a processor, the computer-executable instructions prompt the processor to implement any one of the methods provided in the first aspect.

[0014] The present invention provides a security method and device based on full-range feature recognition and four-dimensional trajectory tracking, and the security device includes a data acquisition module, a feature recognition module, a behavior analysis module, a four-dimensional trajectory reconstruction module and an intelligent operation linkage module. First, through the data acquisition module deployed in the target area, data is collected for one or more target objects entering the target area to obtain a multimodal data set; then, through the feature recognition module, data fusion is performed on the multimodal data set to obtain multiple feature sequences, and individual feature matching and identity authentication are performed on each feature sequence to obtain individual behavior data corresponding to each target object; then, through the behavior analysis module, behavior anomaly detection is performed on each target object based on the individual behavior data to obtain a behavior anomaly detection result corresponding to each target object, and the behavior anomaly detection result is used to characterize whether the target object has abnormal behavior; then, through the four-dimensional trajectory reconstruction module, the four-dimensional movement trajectory of the target object in the target area is visualized, and a specified mark is added to the four-dimensional movement trajectory corresponding to the target object with abnormal behavior; finally, through the intelligent operation linkage module, the alarm and response level corresponding to the abnormal detection result is determined, and the target action corresponding to the alarm and response level is executed, and the target action includes one or more of an alarm action, a multi-level response action and an equipment linkage disposal action. The above method automatically collects multimodal data sets and analyzes the individual behavior data corresponding to each target object, which is used to detect abnormal behavior of the target object, accurately restore the four-dimensional movement trajectory of each target object, and add specified marks to the four-dimensional movement trajectory of the target object with abnormal behavior. In addition, intelligent operation linkage is introduced to realize real-time monitoring and trajectory tracking of target personnel with abnormal behavior. The present invention provides a technical solution that can integrate multiple data sources and support full-time, full-space and full-scene monitoring, so as to improve the security protection level of the park and provide strong technical support for maintaining public safety.

[0015] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention are realized and obtained by the structures particularly pointed out in the description, claims and drawings.

[0016] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0018] Figure 1 A schematic flow chart of a security method based on full-range feature recognition and four-dimensional trajectory tracking provided by an embodiment of the present invention; Figure 2 An overall flow chart of a security method based on full-range feature recognition and four-dimensional trajectory tracking provided by an embodiment of the present invention; Figure 3 A technical framework diagram of a data acquisition module provided by an embodiment of the present invention; Figure 4 A technical framework diagram of a feature recognition module provided by an embodiment of the present invention; Figure 5 A technical framework diagram of a four-dimensional trajectory reconstruction module provided by an embodiment of the present invention; Figure 6 A technical architecture diagram of an intelligent operation linkage module provided by an embodiment of the present invention; Figure 7 A schematic diagram of the structure of a security device based on full-range feature recognition and four-dimensional trajectory tracking provided by an embodiment of the present invention; Figure 8 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution of the present invention will be clearly and completely described in combination with the embodiments below. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0020] At present, most of the existing security means rely on video surveillance and manual patrols of fixed cameras, which are not only inefficient, but also difficult to achieve real-time monitoring and rapid response of the entire park. In addition, when facing emergencies, there is a lack of effective tools to quickly locate the location of suspicious persons and their activity trajectories, which limits the overall effectiveness of the security system. The embodiments of the present invention are deeply analyzed in view of the above problems. First, the activities of individuals within the park are highly uncertain and mobile, and a method that can obtain and process a large number of access control records in real time is required to facilitate timely detection of abnormal behaviors. Secondly, due to the complex environment of the park and the diversity of individual characteristics, the security system is required to have the ability of full-range fuzzy search and be able to quickly locate specific targets in massive data. Finally, in order to improve the speed and accuracy of security incident response, it is necessary to build a three-dimensional map model to intuitively present the activity trajectory of individuals, and combine intelligent analysis technology to achieve effective tracking of suspicious behaviors. Therefore, it is particularly important to develop an intelligent security system that integrates full-time and full-scene access control record query, full-range feature search, and four-dimensional trajectory reconstruction functions.

[0021] Based on this, the present invention provides a security method and device based on full-range feature recognition and four-dimensional trajectory tracking, and provides a technical solution that can integrate multiple data sources and support full-time and full-scene monitoring to improve the security protection level of the park.

[0022] To facilitate understanding of this embodiment, a security method based on full-range feature recognition and four-dimensional trajectory tracking disclosed in an embodiment of the present invention is first described in detail. The security device includes a data acquisition module, a feature recognition module, a behavior analysis module, a four-dimensional trajectory reconstruction module and an intelligent operation linkage module. Figure 1 The flowchart of a security method based on full-range feature recognition and four-dimensional trajectory tracking is shown, and the method mainly includes the following steps S102 to S110: Step S102: Data is collected from one or more target objects entering the target area through a data collection module deployed in the target area to obtain a multimodal data set.

[0023] Among them, the data collection module may include access control sensing devices, video surveillance devices and behavior trajectory recognition sensors. The multimodal data set includes data collected by access control sensing devices (referred to as access control sensing data), data collected by video surveillance devices (referred to as video surveillance data), and data collected by behavior trajectory recognition sensors (referred to as sensor data). The target object is employees or visitors.

[0024] Step S104, through the feature recognition module, data fusion is performed on the multimodal data set to obtain multiple feature sequences, and individual feature matching and identity authentication are performed on each feature sequence to obtain individual behavior data corresponding to each target object.

[0025] In one example, access control sensing data, video surveillance data, and sensor data are matched separately to analyze the access control sensing data, video surveillance data, and sensor data that are suspected to belong to the same target object. By fusing the access control sensing data, video surveillance data, and sensor data that are suspected to belong to the same target object, multiple feature sequences can be obtained. The similarity between the feature sequences is then used to identify the feature sequences belonging to the same target object, to achieve individual feature matching and identity authentication, and to obtain individual behavior data. The individual behavior data will contain all the data of the target object (including access control sensing data, video surveillance data, and sensor data).

[0026] Step S106, using the behavior analysis module, conducts behavior anomaly detection on each target object based on individual behavior data to obtain a behavior anomaly detection result corresponding to each target object. The behavior anomaly detection result is used to characterize whether the target object has abnormal behavior.

[0027] In one example, key features such as behavior frequency and behavior duration can be extracted from individual behavior data, and a pre-trained neural network can be used to determine the target behavior pattern to which the behavior frequency and behavior duration belong. Based on the behavior threshold, behavior frequency, and behavior duration corresponding to the behavior pattern, a determination is made as to whether the target object has abnormal behavior, and a corresponding behavior anomaly detection result is obtained.

[0028] Step S108 : The four-dimensional moving trajectory of the target object in the target area is visualized through a four-dimensional trajectory reconstruction module, and a designated mark is added to the four-dimensional moving trajectory corresponding to the target object with abnormal behavior.

[0029] The four-dimensional movement trajectory includes three-dimensional coordinate data and its associated timestamp information. In one example, a three-dimensional model corresponding to the target area can be constructed, and the four-dimensional movement trajectory can be mapped into the three-dimensional model to achieve visualization of the four-dimensional movement trajectory. Preferably, the four-dimensional movement trajectory corresponding to the target object with abnormal behavior can be visualized by highlighting, special color, magnification and other designated marks.

[0030] Step S110, through the intelligent operation linkage module, determine the alarm and response level corresponding to the abnormal detection result, and execute the target action corresponding to the alarm and response level. The target action includes one or more of the alarm action, multi-level response action and equipment linkage disposal action.

[0031] In one example, the intelligent operation linkage module is configured with an alarm mechanism, a multi-level response strategy and a linkage disposal process, which is used to execute corresponding alarm actions, multi-level response actions and equipment linkage disposal actions according to the alarm and response levels corresponding to the abnormal detection results.

[0032] The security method based on full-range feature recognition and four-dimensional trajectory tracking provided by the embodiment of the present invention automatically collects multimodal data sets and analyzes the individual behavior data corresponding to each target object, which is used to detect abnormal behavior of the target object, and accurately restores the four-dimensional movement trajectory of each target object, and adds specified marks to the four-dimensional movement trajectory of the target object with abnormal behavior. In addition, intelligent operation linkage is introduced to realize real-time monitoring and trajectory tracking of target personnel with abnormal behavior. The present invention provides a technical solution that can integrate multiple data sources and support full-time and full-scene monitoring, so as to improve the security protection level of the park and provide strong technical support for maintaining public safety.

[0033] For ease of understanding, the present invention provides a specific implementation of a security method based on full-range feature recognition and four-dimensional trajectory tracking, see Figure 2 The overall flow chart of a security method based on full-range feature recognition and four-dimensional trajectory tracking shown in the figure includes a data acquisition module, a feature recognition module, a behavior analysis module, a four-dimensional trajectory reconstruction module, and an intelligent operation linkage module, etc. The embodiment of the present invention explains each module separately.

[0034] (i) Data acquisition module: The multimodal data set collected by the data acquisition module includes access control perception data, video surveillance data and sensor data.

[0035] As the cornerstone of intelligent security devices, the data acquisition module not only gathers multiple real-time data sources, but also uses advanced algorithms to achieve full-range feature recognition and four-dimensional trajectory tracking. The module integrates and analyzes data from access control, monitoring and behavior trajectory recognition sensors to build an intelligent security system and behavior recognition system, while providing real-time data support for subsequent decision-making.

[0036] For example, see Figure 3 A technical framework diagram of a data acquisition module is shown.

[0037] The data sources of the data collection module include access control sensing devices, video surveillance equipment and behavior trajectory recognition sensors. Specifically: The access control sensing device uses identity recognition technology: based on face recognition, RFID (Radio Frequency Identification) and other technologies, it creates ID card data of employees and visitors (i.e., target objects), and records the target object's timestamp (T), location (L), and identity certificate (ID) when the target object enters and exits the target area. The formula used is: [D_{entry}={ID,L,T}].

[0038] Video surveillance equipment performs real-time data collection: high-definition surveillance equipment is deployed to obtain real-time video streams. The image recognition algorithm is set to use a convolutional neural network (CNN) to detect moving targets and generate a feature vector (F): [F=CNN(I_t)\quad(I_t\{is a real-time video frame})]. Preferably, the storage algorithm of the video surveillance equipment is also improved: a fast indexing algorithm such as Hash mapping is introduced to optimize the storage and call speed when introducing video frames. The specific storage strategy is: [S={H(T),F}\quad(H\{is a hash function})].

[0039] Behavior trajectory recognition sensor: This sensor detects surrounding objects by transmitting and receiving radio waves. The reflected signal is transformed and analyzed to identify the object's movement trajectory, speed, direction, distance and other parameters. If the received signal is set to R(t), it can be expressed as: [R(t)=H(S(t))\quad(S(t)\{is the transmitted signal})]. Through the static detection function, the sensor can determine the static state of the human body. The algorithm built into the sensor analyzes the changes in the reflected signal. If there is no obvious movement, the security device will identify it as static and send out relevant signals.

[0040] Data processing and storage are performed on the multimodal data sets collected from the above data sources, including data preprocessing, missing value filling, data normalization, data storage, etc. Specifically: Data preprocessing: The collected data is preprocessed to improve the quality and usability of the data. The preprocessing steps include: Denoising: Use a high-pass or low-pass filter to remove noise signals. The denoising of E(t) can be expressed as: [E_{filtered}(t)=E(t)\asth(t)\quad(h\{is the filter response})].

[0041] Missing value filling: Apply interpolation techniques (such as linear interpolation) to fill in missing time series data based on neighboring data points.

[0042] Data normalization: Scale the feature values ​​to between 0 and 1 for ease of model processing, the formula is: [E_{norm}(i)=\frac{E(i)-E_{min}}{E_{max}-E_{min}}].

[0043] Data storage: The storage mechanism uses a cloud database that adapts to high concurrency, such as MongoDB or Cassandra, and stores data in a JSON structure.

[0044] (II) Feature recognition module: The core task of the feature recognition module is to extract high-dimensional features from data from various sources to achieve accurate identity recognition and behavior classification of individuals. By comprehensively utilizing technologies such as deep learning, fuzzy matching, and multimodal data fusion, the accuracy and efficiency of feature recognition can be improved, providing reliable support for intelligent security devices. Figure 4 The technical framework diagram of a feature recognition module shown includes training module, feature extraction and comparison processes.

[0045] The embodiment of the present invention provides a specific processing flow of a feature recognition module, including: (2.1) Extract the first facial features corresponding to the access control sensing data and the second facial features corresponding to the video surveillance data.

[0046] In one example, facial features are extracted from access control sensing data and video surveillance data using a convolutional neural network (CNN). CNN can automatically learn hierarchical features in images, and the convolution operation of each layer can be expressed by the following formula: [Z=f(W*X+b)] where (Z) is the output feature map, (W) is the convolution kernel, (X) is the input image, (b) is the bias, and (f) is the activation function (such as ReLU). At the same time, large-scale annotated datasets (such as LFW, VGGFace, etc.) are used in the training phase to enhance the generalization ability of the model. Data enhancement (such as random cropping, flipping, etc.) is used to further improve the robustness of the model.

[0047] (2.2) Perform fuzzy matching on the first facial feature and the second facial feature to determine the correlation between the access control sensing data and the video surveillance data.

[0048] In one example, fuzzy matching is performed on the first facial feature and the second facial feature to improve the recognition rate and accuracy.

[0049] The fuzzy matching algorithm processing flow includes: calculating the similarity between feature vectors, usually using cosine similarity or Euclidean distance: [\{CosineSimilarity}=\frac{A\cdotB}{||A||\cdot||B||}][\{EuclideanDistance}=||AB||_2=\sqrt{\sum(A_i-B_i)^2}]. Comparing the detected image with the image in the database, determining the correlation between the access control perception data and the video surveillance data.

[0050] (2.3) Based on the association relationship, the access control perception data, video surveillance data and sensor data are fused to obtain multiple feature sequences.

[0051] In one example, a more comprehensive description of individual characteristics is formed by fusing access control data (such as time, location, and behavior records), video surveillance data (facial features, action recognition), and sensor data (location, behavior).

[0052] In a specific implementation, access control sensing data and sensor data are matched, and video surveillance data and sensor data are matched; if there is a correlation between the access control sensing data and video surveillance data matched by the sensor data, the sensor data and its matched access control sensing data and video surveillance data are fused to obtain multiple feature sequences.

[0053] Fusion methods include weighted averaging and principal component analysis (PCA). PCA reduces high-dimensional features to lower dimensions and eliminates the correlation between features. Its transformation formula is as follows: [Z=\{X}W], where (Z) is the reduced-dimensional feature, (X) is the original data, and (W) is the feature weight matrix.

[0054] (2.4) For any two feature sequences, determine the feature similarity between the two feature sequences, and determine that the two feature sequences belong to the same target object when the feature similarity is higher than a preset threshold. Based on the feature sequences belonging to the same target object, determine the individual behavior data corresponding to each target object.

[0055] In one example, feature comparison uses the Softmax classification algorithm to calculate feature similarity, thereby achieving individual feature matching and identity confirmation. The Softmax formula is as follows: [P(y_i)=\frac{e^{f(x_i)}}{\sum_{j=1}^{K}e^{f(x_j)}}].

[0056] Where (P(y_i)) represents the probability that the feature vector (x_i) belongs to the identity (y_i), (f(x_i)) is the classification score of the feature vector, and (K) is the number of categories.

[0057] The classification decision is based on: if (P(y_i)>\theta) (threshold), the identity is confirmed to be (y_i); otherwise, it is judged to be an unknown identity.

[0058] (III) Behavior Analysis Module: Conduct in-depth analysis of individual behavior patterns to identify and determine abnormal behavior. This module is an important part of the intelligent security device. Through behavior pattern analysis, it can monitor potential safety hazards in real time and increase awareness of abnormal situations, thereby effectively protecting the safety of the park.

[0059] The specific implementation mainly includes the following steps: (3.1) Preprocess the individual behavior data corresponding to each target object, and extract the key features corresponding to each target object from the preprocessed individual behavior data, where the key features include behavior frequency and behavior duration.

[0060] Preprocessing: Clean and standardize the collected individual behavior data to ensure the quality and consistency of the data. This includes removing noise data, filling missing values, and normalizing the data to a uniform scale.

[0061] Feature extraction: Extract key features from the preprocessed individual behavior data, such as behavior frequency and behavior duration. These features will serve as input to the behavior pattern model. Behavior frequency is the number of times a specific behavior occurs per unit time, reflecting the activity of the behavior. Behavior duration is the time an individual stays in a certain activity state, reflecting the persistence of the behavior.

[0062] (3.2) Determine the target behavior pattern to which each target object belongs through a pre-trained behavior pattern model based on a plurality of pre-configured candidate behavior patterns and key features corresponding to each target object.

[0063] In order to accurately analyze the behavior patterns of individuals, we use the long short-term memory (LSTM) neural network. This architecture based on the recurrent neural network (RNN) is particularly suitable for sequence data and can effectively solve the shortcomings of traditional models in dealing with long-term dependencies.

[0064] Before executing (3.2), LSTM model construction, model training, model verification and testing are required.

[0065] LSTM model construction: Use deep learning frameworks such as TensorFlow or PyTorch to build an LSTM neural network model. The model should include an input layer, an LSTM layer, a fully connected layer, and an output layer. The LSTM layer is responsible for capturing the temporal dependencies in the behavior data.

[0066] Model training: Use historical behavior data to train the LSTM model, and optimize the model parameters through the back propagation algorithm so that it can accurately predict individual behavior patterns. During the training process, methods such as cross-validation are used to evaluate the performance of the model and make necessary adjustments.

[0067] Model verification and testing: Verify the performance of the LSTM model on an independent test set to ensure that it can accurately identify and predict individual behavior patterns. At the same time, perform robustness testing on the model to ensure that it can maintain stable performance in different scenarios and conditions.

[0068] The pre-configured multiple candidate behavior patterns and the key features corresponding to each target object are input into the verified and tested LSTM model to determine the target behavior pattern to which each target object belongs using the LSTM model. The behavior patterns can be walking, wandering, staying, etc., which reflect the activity status of the individual.

[0069] (3.3) Based on the key features corresponding to each target object and its target behavior pattern, a behavior anomaly detection is performed on each target object to obtain a behavior anomaly detection result corresponding to each target object.

[0070] In one example, for any key feature corresponding to each target object, it is determined whether the value of the key feature is higher than the behavior threshold corresponding to the target behavior pattern to which the target object belongs, and when the judgment result is yes, it is determined that the target object has abnormal behavior. The process of setting the behavior threshold is as follows: Use statistical methods (such as standard deviation) and machine learning methods to set the behavior threshold. For example, define the mean and standard deviation of the behavior, and judge the abnormality based on this: [z=\frac{x-\mu}{\sigma}]; where (z) is the standardized value, (x) is the current behavior feature, (\mu) is the sample mean, and (\sigma) is the sample standard deviation. When (z>z_{{threshold}}), it is marked as abnormal behavior.

[0071] Specifically, behavioral anomaly detection is a key step in identifying potential risks. Combined with the LSTM model, the following two detection methods are performed: active detection and passive detection. Active detection: Real-time evaluation of whether individual behavior exceeds the set normal range. Once abnormal behavior is detected, the security device will immediately sound an alarm and generate a log. Passive detection: When abnormal frequencies or data peaks are detected in the data stream, the relevant data is recorded and analyzed in detail. In this process, the data is further analyzed through clustering algorithms or anomaly detection algorithms (such as isolation forests) to identify potential abnormal patterns.

[0072] (IV) Four-dimensional trajectory reconstruction module: The main function of the 4D trajectory reconstruction module is to effectively visualize the movement trajectory of individuals in the smart park, so that managers can monitor and analyze dynamic activities. This module helps managers understand the flow of people in the park, as well as potential safety hazards and abnormal behaviors in real time by accurately capturing and analyzing individual behaviors.

[0073] In one implementation manner, the three-dimensional coordinate data of the target object moving in the target area and its associated timestamp information are first obtained to construct a four-dimensional movement trajectory corresponding to the target object; then a three-dimensional model corresponding to the target area is established, and the four-dimensional movement trajectory is mapped to the three-dimensional model to achieve visualization of the four-dimensional movement trajectory.

[0074] Specifically, in the process of collecting and processing trajectory data, various sensors and data collection technologies are first used to obtain individual location information. Figure 5 The technical framework diagram of a four-dimensional trajectory reconstruction module shown in FIG. includes the processes of collecting position information, collecting and constructing trajectory data, data coordinate conversion, and dynamic visualization of trajectory. The following are the key steps of trajectory reconstruction: (4.1) Location information acquisition: The geographical location of individuals in the park can be obtained in real time through GPS, base station signals, Wi-Fi positioning and other technologies. For indoor environments, Bluetooth positioning devices or ultra-wideband (UWB) technology can be used. During the collection process, the location information is presented as a timestamp and the corresponding three-dimensional coordinate data, that is, ((t,x,y,z)), where (t) is the time, (x), (y), and (z) are the spatial positions of the three-dimensional coordinates.

[0075] (4.2) Trajectory construction: By processing the position information in the time series, the three-dimensional trajectory of the individual is constructed. The trajectory can be represented as a series of connected points, that is, [{trajectory}={(t_1,x_1,y_1,z_1),(t_2,x_2,y_2,z_2),\ldots,(t_n,x_n,y_n,z_n)}].

[0076] (4.3) Data cleaning and interpolation: Remove noise in data acquisition and fill in missing data caused by unstable signals. For missing data points, linear interpolation or spline interpolation methods can be used to complete them to ensure the continuity of the trajectory.

[0077] Furthermore, 3D modeling and visualization technology can be used to provide managers with intuitive dynamic trajectory display. The following are several important aspects of the visualization process: Choose visualization tools: Use modern visualization tools such as Three.js and Unity to visualize trajectory data as 3D models. Three.js is a web-based 3D JavaScript library that is very suitable for web applications, while Unity provides powerful 3D rendering and real-time interaction capabilities.

[0078] Data rendering: Map the trajectory points to the 3D scene to generate a dynamic trajectory diagram. It can be expressed in a variety of ways, such as particle system, line drawing or point cloud display. The dynamic transformation of the trajectory can display important information such as the individual's moving direction and speed.

[0079] User interaction: Provide a user-friendly interactive interface that allows managers to freely rotate, zoom and observe the three-dimensional perspective to further enhance the understanding of dynamic activities. It can also embed event trigger mechanisms, such as clicking a specific point on the track to pop up relevant information or analytical data.

[0080] (V) Intelligent operation linkage module: The intelligent operation linkage mechanism aims to integrate all security modules in the park to form a highly integrated security response system. This mechanism can realize real-time data communication and collaboration between modules, which not only improves the overall security efficiency, but also enhances the ability to respond to emergencies. Through efficient data exchange, security devices can make quick judgments and implement appropriate response measures, which is conducive to building a safer park environment.

[0081] See also Figure 6 The technical architecture diagram of an intelligent operation linkage module shown in the figure includes the process of integrating data sources, behavior analysis modules, data input acquisition interface, behavior detection, triggering alarms and notifying relevant personnel. Please refer to the following steps for details: (5.1) Data fusion and analysis: Real-time data fusion: The security device integrates information from different data sources in real time, including access control systems, surveillance cameras, sensors, and alarm systems. Through deep learning and data mining algorithms, combined with feature recognition and behavior analysis data, it generates a comprehensive view and builds an individual behavior and identity template. This template can not only reflect the individual's behavioral characteristics, but also identify possible abnormal behaviors, thus providing a basis for subsequent security decisions.

[0082] Data visualization: To ensure that operators can quickly understand complex data, security devices provide a real-time monitoring interface to display the dynamics and behavior patterns of people in the area. Visualization tools display through maps, charts and dynamic graphics, which facilitates security personnel to quickly analyze and evaluate data. For example, high-risk areas can be displayed through heat maps, or individual movement paths can be displayed using dynamic trajectory maps. At the same time, security devices also allow users to customize visualization indicators to more accurately monitor key areas and targets.

[0083] (5.2) Real-time alert mechanism: Alarm mechanism design: To ensure timely and effective response to suspicious behaviors, this mechanism adopts a hierarchical alarm system. The system divides different alarm levels according to the degree of abnormal behavior and triggers corresponding emergency response measures. For example, for low-level abnormal behaviors, the system may only record and mark them; while for high-level threats, security personnel will be notified immediately to take action. The alarm threshold can be dynamically adjusted based on historical data analysis and actual application scenarios to achieve the best response effect.

[0084] Multi-level response strategy: In addition to graded alarms, a multi-level response strategy is also designed. For example, when suspicious behavior is detected, the system will first automatically send a text message or email to alert relevant personnel to pay attention; if the situation continues to deteriorate, it will trigger a higher level of response, such as on-site sound and light warnings, remote voice warnings, etc.; in the most extreme cases, the system will directly contact the park's emergency response team or even external law enforcement agencies.

[0085] Linkage handling process: Once an alarm is triggered, the system will automatically start a series of linkage handling processes. This includes but is not limited to locking the relevant area, adjusting the viewing angle of the surveillance camera to obtain more details, launching drones for aerial reconnaissance, etc. These linkage operations can greatly improve the response efficiency and ensure that the development of the situation is controlled in the shortest time.

[0086] The embodiment of the present invention proposes an intelligent security method based on full-range feature recognition and four-dimensional trajectory tracking. The method realizes all-round monitoring of individual activities in the park by integrating advanced image recognition technology and big data analysis algorithms. The method can automatically collect and analyze the access records of the access control system, use fuzzy matching algorithms to search for features, and accurately restore the individual's movement trajectory with the help of three-dimensional map technology. More importantly, the present invention also introduces an intelligent operation linkage mechanism that can automatically associate facial capture information with other systems (such as feature recognition, image recognition and access records), thereby realizing real-time monitoring and trajectory tracking of suspicious personnel behavior. This "follow the map to find the trace" method has greatly improved the security protection level of the park and provided strong technical support for maintaining public safety.

[0087] The embodiment of the present invention provides a specific example of a security method based on full-range feature recognition and four-dimensional trajectory tracking. An emergency occurred in a large park, and an unauthorized person attempted to enter a sensitive area in the park. The security device first recorded the person's timestamp (T), location (L) and identification (ID) through the access control system, and captured the person's facial image through video surveillance. The feature recognition module extracts facial features through CNN and performs fuzzy matching with the identity information in the database, and finds that the person does not have access rights.

[0088] At this point, the behavior analysis module starts working, analyzing the person's behavior pattern through the LSTM neural network and detecting abnormal behavior (such as long-term wandering). The four-dimensional trajectory reconstruction module updates the person's movement trajectory in real time and displays it on the visual interface for the park security personnel to view.

[0089] The intelligent operation linkage mechanism was immediately activated, and the security device automatically sent an alarm to the security personnel, and reminded relevant personnel to pay attention through text messages and emails. At the same time, the security device adjusted the angle of the surveillance camera to obtain more details, and launched a drone for aerial reconnaissance. Finally, the security personnel quickly locked the person based on the information provided by the security device and took corresponding measures to successfully prevent potential security threats.

[0090] It can be seen from the above embodiments that the intelligent security method provided by the present invention can effectively improve the security management level of the park, ensure rapid response in emergencies, and protect the safety of the park.

[0091] In summary, the technical key points of the security method based on full-range feature recognition and four-dimensional trajectory tracking provided by the embodiment of the present invention include: (a) Full range feature recognition and multimodal data fusion: Innovation: Comprehensive use of multiple data sources such as face recognition, RFID, and behavioral trajectory recognition sensors, and identity authentication through fuzzy matching technology, combined with multimodal data fusion technology (such as weighted average method, PCA) to achieve all-round feature recognition of individuals. Advantages: Compared with single feature recognition, this invention can more comprehensively describe individual characteristics and improve recognition accuracy and robustness.

[0092] (b) Four-dimensional trajectory tracking and data processing: Innovation: Use GPS, Wi-Fi positioning and other technologies to obtain individual location information in real time, build a four-dimensional trajectory in combination with the time dimension, and use data cleaning and interpolation technology (such as linear interpolation) to ensure trajectory continuity. Advantages: Accurately capture and analyze individual movement trajectories, helping managers understand the flow of people in the park and potential safety hazards in real time.

[0093] (c) Real-time data fusion and intelligent analysis: Innovation: Data from access control systems, surveillance cameras, sensors and alarm systems are integrated in real time, comprehensive views are generated through deep learning and data mining algorithms, and behavior pattern analysis is performed using LSTM neural networks. Advantages: Through real-time data integration and intelligent analysis, security devices can make quick judgments, detect abnormal behaviors in a timely manner, and improve overall security efficiency.

[0094] (d) Four-dimensional trajectory reconstruction and dynamic visualization: Innovation: Use modern visualization tools (such as Three.js and Unity) to visualize trajectory data in three dimensions, and provide a user-friendly interactive interface that supports free rotation, zooming, and observing three-dimensional perspectives. Advantages: Provide managers with intuitive dynamic trajectory display, enhance understanding of dynamic activities, and assist decision-making.

[0095] (e) Intelligent operation linkage mechanism: Innovation: Establish a hierarchical alarm system and design a multi-level response strategy to ensure timely and effective response to suspicious behaviors. Once the alarm is triggered, the system automatically starts the linkage disposal process, including locking the area, adjusting the viewing angle of the surveillance camera, launching drone reconnaissance, etc. Advantages: Through the intelligent linkage mechanism, the system can respond quickly and take appropriate countermeasures, improve the efficiency of event handling, and reduce security risks.

[0096] Based on the above technical points, the embodiments of the present invention have at least the following advantages: Comprehensively improve security efficiency: By integrating a variety of advanced technical means, all-round monitoring of personnel activities in the park can be achieved, and the overall effectiveness of security devices can be improved; real-time response and rapid decision-making: real-time data fusion and intelligent analysis technology enable security devices to make quick judgments and take appropriate countermeasures to shorten response time; accurate behavior recognition and anomaly detection: using advanced behavior analysis technology to accurately identify individual behavior patterns, timely discover and deal with potential safety hazards; enhance user experience and decision support: through four-dimensional trajectory reconstruction and dynamic visualization technology, provide managers with intuitive dynamic trajectory display, enhance understanding of dynamic activities, and assist decision-making.

[0097] Based on the above embodiments, the present invention provides a security device based on full-range feature recognition and four-dimensional trajectory tracking, see Figure 7 The schematic diagram of the structure of a security device based on full-range feature recognition and four-dimensional trajectory tracking is shown, the security device includes a data acquisition module 702, a feature recognition module 704, a behavior analysis module 706, a four-dimensional trajectory reconstruction module 708 and an intelligent operation linkage module 710: The data collection module 702 deployed in the target area is used to: collect data from one or more target objects entering the target area to obtain a multimodal data set; The feature recognition module 704 is used to: perform data fusion on the multimodal data set to obtain multiple feature sequences, and perform individual feature matching and identity authentication on each feature sequence to obtain individual behavior data corresponding to each target object; The behavior analysis module 706 is used to: perform behavior anomaly detection on each target object based on individual behavior data, and obtain a behavior anomaly detection result corresponding to each target object, and the behavior anomaly detection result is used to characterize whether the target object has abnormal behavior; The four-dimensional trajectory reconstruction module 708 is used to: visualize the four-dimensional moving trajectory of the target object in the target area, and add a specified mark to the four-dimensional moving trajectory corresponding to the target object with abnormal behavior; The intelligent operation linkage module 710 is used to: determine the alarm and response level corresponding to the abnormal detection result, and execute the target action corresponding to the alarm and response level. The target action includes one or more of the alarm action, multi-level response action and equipment linkage disposal action.

[0098] The security device based on full-range feature recognition and four-dimensional trajectory tracking provided by the embodiment of the present invention automatically collects multimodal data sets and analyzes the individual behavior data corresponding to each target object, so as to detect abnormal behavior of the target object, accurately restore the four-dimensional movement trajectory of each target object, and add designated marks to the four-dimensional movement trajectory of the target object with abnormal behavior. In addition, intelligent operation linkage is introduced to realize real-time monitoring and trajectory tracking of target personnel with abnormal behavior. The present invention provides a technical solution that can integrate multiple data sources and support full-time and full-scene monitoring, so as to improve the security protection level of the park and provide strong technical support for maintaining public safety.

[0099] The device provided in the embodiment of the present invention has the same implementation principle and technical effects as those of the aforementioned method embodiment. For the sake of brief description, for matters not mentioned in the device embodiment, reference may be made to the corresponding contents in the aforementioned method embodiment.

[0100] An embodiment of the present invention provides an electronic device. Specifically, the electronic device includes a processor and a storage device. The storage device stores a computer program, and when the computer program is executed by the processor, it executes the method described in any one of the above-mentioned implementation methods.

[0101] Figure 8 A structural diagram of an electronic device provided in an embodiment of the present invention, the electronic device 100 includes: a processor 80, a memory 81, a bus 82 and a communication interface 83, wherein the processor 80, the communication interface 83 and the memory 81 are connected via the bus 82; the processor 80 is used to execute an executable module stored in the memory 81, such as a computer program.

[0102] The memory 81 may include a high-speed random access memory (RAM), and may also include a non-volatile memory, such as at least one disk memory. The communication connection between the system network element and at least one other network element is realized through at least one communication interface 83 (which may be wired or wireless), and the Internet, wide area network, local area network, metropolitan area network, etc. may be used.

[0103] The bus 82 may be an ISA bus, a PCI bus, or an EISA bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 8 Only one bidirectional arrow is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0104] Among them, the memory 81 is used to store programs, and the processor 80 executes the program after receiving the execution instruction. The method executed by the device for flow process definition disclosed in any embodiment of the above-mentioned embodiments of the present invention can be applied to the processor 80 or implemented by the processor 80.

[0105] The processor 80 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the hardware integrated logic circuit or software instructions in the processor 80. The above processor 80 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods, steps and logic block diagrams disclosed in the embodiments of the present invention can be implemented or executed. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor. The steps of the method disclosed in the embodiment of the present invention can be directly embodied as a hardware decoding processor to execute, or can be executed by a combination of hardware and software modules in the decoding processor. The software module can be located in a mature storage medium in the field such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory 81, and the processor 80 reads the information in the memory 81 and completes the steps of the above method in combination with its hardware.

[0106] The computer program product of the readable storage medium provided in the embodiment of the present invention includes a computer-readable storage medium storing program code, and the instructions included in the program code can be used to execute the methods described in the previous method embodiments. The specific implementation can be referred to the previous method embodiments, which will not be repeated here.

[0107] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program codes.

[0108] Finally, it should be noted that the above-described embodiments are only specific implementations of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit them. The protection scope of the present invention is not limited thereto. Although the present invention is described in detail with reference to the above-described embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above-described embodiments within the technical scope disclosed by the present invention, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A security method based on full-range feature recognition and four-dimensional trajectory tracking, characterized in that: Applied to a security device, the security device includes a data acquisition module, a feature recognition module, a behavior analysis module, a four-dimensional trajectory reconstruction module and an intelligent operation linkage module, and the method includes: By using the data collection module deployed in the target area, data is collected from one or more target objects entering the target area to obtain a multimodal data set; Through the feature recognition module, the multimodal data set is subjected to data fusion to obtain a plurality of feature sequences, and individual feature matching and identity authentication are performed on each of the feature sequences to obtain individual behavior data corresponding to each of the target objects; Through the behavior analysis module, behavior anomaly detection is performed on each of the target objects based on the individual behavior data to obtain a behavior anomaly detection result corresponding to each of the target objects, wherein the behavior anomaly detection result is used to characterize whether the target object has abnormal behavior; The four-dimensional moving trajectory of the target object in the target area is visualized by the four-dimensional trajectory reconstruction module, and a designated mark is added to the four-dimensional moving trajectory corresponding to the target object with abnormal behavior; Through the intelligent operation linkage module, the alarm and response level corresponding to the abnormal detection result is determined, and the target action corresponding to the alarm and response level is executed. The target action includes one or more of an alarm action, a multi-level response action and an equipment linkage disposal action.

2. The security method based on full-range feature recognition and four-dimensional trajectory tracking according to claim 1 is characterized in that: The multimodal data set includes access control perception data, video surveillance data and sensor data; The multimodal data set is subjected to data fusion to obtain multiple feature sequences, including: Extracting a first facial feature corresponding to the access control sensing data and a second facial feature corresponding to the video surveillance data, and performing fuzzy matching on the first facial feature and the second facial feature to determine an association relationship between the access control sensing data and the video surveillance data; Based on the association relationship, data fusion is performed on the access control sensing data, the video surveillance data and the sensor data to obtain multiple feature sequences.

3. The security method based on full-range feature recognition and four-dimensional trajectory tracking according to claim 2 is characterized in that: Based on the association relationship, the access control sensing data, the video surveillance data and the sensor data are fused to obtain multiple feature sequences, including: Matching the access control sensing data with the sensor data, and matching the video surveillance data with the sensor data; If the association relationship exists between the access control sensing data and the video surveillance data matched by the sensor data, data fusion is performed on the sensor data and the access control sensing data and the video surveillance data matched by the sensor data to obtain multiple feature sequences.

4. The security method based on full-range feature recognition and four-dimensional trajectory tracking according to claim 1 is characterized in that: Performing individual feature matching and identity authentication on each of the feature sequences to obtain individual behavior data corresponding to each of the target objects, including: For any two of the feature sequences, determining the feature similarity between the two feature sequences, and determining that the two feature sequences belong to the same target object when the feature similarity is higher than a preset threshold; Based on the feature sequence belonging to the same target object, individual behavior data corresponding to each target object is determined.

5. The security method based on full-range feature recognition and four-dimensional trajectory tracking according to claim 1 is characterized in that: Performing behavior anomaly detection on each of the target objects based on the individual behavior data to obtain a behavior anomaly detection result corresponding to each of the target objects includes: Preprocessing the individual behavior data corresponding to each of the target objects, and extracting key features corresponding to each of the target objects from the preprocessed individual behavior data, wherein the key features include behavior frequency and behavior duration; Determine the target behavior pattern to which each target object belongs through a pre-trained behavior pattern model based on a plurality of pre-configured candidate behavior patterns and the key features corresponding to each target object; Based on the key features corresponding to each target object and the target behavior pattern to which it belongs, behavior anomaly detection is performed on each target object to obtain a behavior anomaly detection result corresponding to each target object.

6. The security method based on full-range feature recognition and four-dimensional trajectory tracking according to claim 5 is characterized in that: Based on the key features corresponding to each target object and the target behavior pattern to which it belongs, behavior anomaly detection is performed on each target object to obtain a behavior anomaly detection result corresponding to each target object, including: For any of the key features corresponding to each of the target objects, determine whether the value of the key feature is higher than the behavior threshold corresponding to the target behavior pattern to which the target object belongs, and determine that the target object has abnormal behavior if the judgment result is yes.

7. The security method based on full-range feature recognition and four-dimensional trajectory tracking according to claim 1 is characterized in that: Visualizing the four-dimensional moving trajectory of the target object in the target area includes: Acquire the three-dimensional coordinate data of the target object moving within the target area and its associated timestamp information, and construct a four-dimensional movement trajectory corresponding to the target object; A three-dimensional model corresponding to the target area is established, and the four-dimensional movement trajectory is mapped into the three-dimensional model to realize visualization of the four-dimensional movement trajectory.

8. A security device based on full-range feature recognition and four-dimensional trajectory tracking, characterized in that: The security device includes a data acquisition module, a feature recognition module, a behavior analysis module, a four-dimensional trajectory reconstruction module and an intelligent operation linkage module: The data acquisition module deployed in the target area is used to: collect data from one or more target objects entering the target area to obtain a multimodal data set; The feature recognition module is used to: perform data fusion on the multimodal data set to obtain multiple feature sequences, and perform individual feature matching and identity authentication on each of the feature sequences to obtain individual behavior data corresponding to each of the target objects; The behavior analysis module is used to: perform behavior anomaly detection on each of the target objects based on the individual behavior data to obtain a behavior anomaly detection result corresponding to each of the target objects, wherein the behavior anomaly detection result is used to characterize whether the target object has abnormal behavior; The four-dimensional trajectory reconstruction module is used to: visualize the four-dimensional movement trajectory of the target object in the target area, and add a specified mark to the four-dimensional movement trajectory corresponding to the target object with abnormal behavior; The intelligent operation linkage module is used to determine the alarm and response level corresponding to the abnormal detection result, and execute the target action corresponding to the alarm and response level, and the target action includes one or more of an alarm action, a multi-level response action and an equipment linkage disposal action.

9. An electronic device, characterized in that: The method comprises a processor and a memory, wherein the memory stores computer executable instructions that can be executed by the processor, and the processor executes the computer executable instructions to implement the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are called and executed by a processor, the computer-executable instructions prompt the processor to implement the method according to any one of claims 1 to 7.

Images