An intelligent security system, method, device and storage medium based on AI analysis

Abstract

The present application provides an intelligent security system, method, device and storage medium based on AI analysis, which relates to the field of intelligent security technology. The system collects surveillance video information in an intelligent security area and extracts behavioral information of a target person in the surveillance video information; constructs a behavioral topology map based on the behavioral information, extracts abnormal behavior nodes from the behavioral topology map, and determines the degree of abnormal behavior deviation through the behavioral deviation characteristics between each abnormal behavior node; obtains an audio data stream, converts the audio data stream into a security audio segment of the target person, and constructs an audio feature matrix based on the security audio segment; determines the probability of the target person invading a sensitive area in the intelligent security area based on the abnormal behavior deviation and the audio feature matrix, and sends an intrusion report to the security center based on the intrusion probability. The present application can accurately identify the behavior pattern of the target person and, in combination with the audio features, evaluate the intrusion threat of the target person to improve the safety and reliability of the intelligent security system.

Description

Technical Field

[0001] The present application relates to the field of intelligent security technology. More specifically, the present application relates to an intelligent security system, method, device and storage medium based on AI analysis. Background Art

[0002] Smart security systems based on AI analysis utilize advanced artificial intelligence technologies to process and analyze surveillance data and sensor information in real time, improving the efficiency and accuracy of security monitoring. Smart security systems typically integrate multiple functions, including video surveillance, audio recognition, intrusion detection, and behavioral analysis. They can automatically identify potential security threats and issue timely alerts, reducing labor costs and response times.

[0003] In practical applications, intelligent security systems use deep learning and machine learning algorithms to analyze behavioral patterns in surveillance video, identify suspicious activity and abnormal behavior, and even perform facial recognition and object tracking. Furthermore, the incorporation of audio sensors enables the system to detect changes in ambient sound, such as sudden arguments or the sound of breaking glass, further enhancing security capabilities. The combination of these technologies not only enhances the intelligence of security monitoring but also provides users with more comprehensive and in-depth security protection. However, existing technologies still experience false positives and false negatives in complex and dynamic environments. For example, intelligent security systems may mistakenly identify harmless activity as a potential threat, resulting in unnecessary alerts or missing real security risks. Therefore, accurately identifying the behavioral patterns of a target individual and combining audio features to assess their intrusion threat remains a challenge facing the industry. Summary of the Invention

[0004] The present application provides an intelligent security system, method, device, and storage medium based on AI analysis, which can accurately identify the behavior patterns of target persons and combine audio features to assess the intrusion threat of the target person, thereby improving the safety and reliability of the intelligent security system.

[0005] In a first aspect, the present application provides an intelligent security method based on AI analysis, the security method comprising the following steps:

[0006] Collecting surveillance video information in the intelligent security area and extracting behavioral information of the target person in the surveillance video information;

[0007] constructing a behavior topology map of the target person in the intelligent security area based on the behavior information, extracting abnormal behavior nodes from the behavior topology map, and then determining the abnormal deviation degree of the target person's behavior in the intelligent security area based on the behavior deviation characteristics between each abnormal behavior node;

[0008] Acquire an audio data stream of a target person in an intelligent security area, convert the audio data stream into a security audio segment of the target person, and construct an audio feature matrix of the target person in the intelligent security area based on the security audio segment;

[0009] The probability of a target person invading a sensitive area in an intelligent security area is determined according to the abnormal behavior deviation and the audio feature matrix, and an intrusion report is sent to a security center based on the intrusion probability.

[0010] In this embodiment, extracting the behavior information of the target person from the surveillance video information specifically includes:

[0011] Extracting facial feature information from the surveillance video information;

[0012] Based on the comparison between the facial feature information and the candidate facial feature information, a target person entering the intelligent security area is determined;

[0013] Obtaining image information of a target person in the surveillance video information;

[0014] Behavioral features are extracted from the image information to obtain behavioral information of the target person in the monitoring video information.

[0015] In this embodiment, constructing a behavior topology map of the target person in the intelligent security area based on the behavior information specifically includes:

[0016] Divide the intelligent security area into intervals to obtain a grid map of the intelligent security area;

[0017] The behavior information is mapped to a grid map of the intelligent security area to obtain a behavior topology map of the target person in the intelligent security area.

[0018] In this embodiment, extracting abnormal behavior nodes from the behavior topology graph specifically includes:

[0019] Determine the behavior characteristics corresponding to each behavior topology node in the behavior topology graph;

[0020] Determine the target person's behavioral characteristics based on all behavioral characteristics;

[0021] For each sensitive area in the behavior topology map, determining the behavior risk of each behavior topology node in the sensitive area based on the behavior characterization feature;

[0022] The behavior topology node with the largest behavior risk is used as the abnormal behavior node corresponding to the sensitive area, and then the abnormal behavior nodes corresponding to each sensitive area in the behavior topology graph are obtained.

[0023] In this embodiment, converting the audio data stream into a security audio segment of the target person specifically includes:

[0024] Determining the short-term energy and zero-crossing rate of the audio data stream;

[0025] The audio data stream is converted according to the short-time energy and the zero-crossing rate to obtain a security audio segment of the target person.

[0026] In this embodiment, constructing the audio feature matrix of the target person in the smart security area based on the security audio segment is to extract features from the security audio segment, and then constructing the audio feature matrix of the target person in the smart security area based on the extracted audio features.

[0027] In this embodiment, determining the probability of a target person intruding into a sensitive area in the intelligent security area based on the abnormal behavior deviation and the audio feature matrix specifically includes:

[0028] Get pre-built Bayesian network probability models;

[0029] The abnormal behavior deviation and the audio feature matrix are input as input data into the Bayesian network probability model for prediction, so as to obtain the probability of the target person invading the sensitive area in the intelligent security area.

[0030] In a second aspect, the present application provides an intelligent security system based on AI analysis for executing an intelligent security method based on AI analysis, the security system comprising:

[0031] The acquisition module is used to collect surveillance video information in the intelligent security area and extract the behavior information of the target person in the surveillance video information;

[0032] an abnormal behavior determination module, configured to construct a behavior topology map of the target person in the intelligent security area based on the behavior information, extract abnormal behavior nodes from the behavior topology map, and then determine the abnormal deviation degree of the target person's behavior in the intelligent security area based on the behavior deviation characteristics between each abnormal behavior node;

[0033] An audio feature extraction module is used to obtain an audio data stream of a target person in an intelligent security area, convert the audio data stream into a security audio segment of the target person, and construct an audio feature matrix of the target person in the intelligent security area based on the security audio segment;

[0034] The intrusion security module is used to determine the probability of a target person invading a sensitive area in the intelligent security area based on the abnormal behavior deviation and the audio feature matrix, and send an intrusion report to the security center based on the intrusion probability.

[0035] In a third aspect, the present application provides a computer device comprising a memory and a processor, wherein the memory is used to store a computer program, and the processor is used to call and run the computer program from the memory, so that the computer device executes the above-mentioned customer-associated risk prediction method based on machine learning.

[0036] In a fourth aspect, the present application provides a computer-readable storage medium, which stores instructions or codes. When the instructions or codes are run on a computer, the computer implements the above-mentioned customer-associated risk prediction method based on machine learning.

[0037] The technical solutions provided by the embodiments disclosed in this application have the following beneficial effects:

[0038] By collecting surveillance video information in the intelligent security area, the behavioral information of the target person in the surveillance video information is extracted; based on the behavioral information, a behavioral topology map of the target person in the intelligent security area is constructed, abnormal behavior nodes are extracted from the behavioral topology map, and then the abnormal behavior deviation degree of the target person in the intelligent security area is determined through the behavioral deviation characteristics between each abnormal behavior node; the audio data stream of the target person in the intelligent security area is obtained, the audio data stream is converted into a security audio segment of the target person, and an audio feature matrix of the target person in the intelligent security area is constructed based on the security audio segment; the probability of the target person invading a sensitive area in the intelligent security area is determined based on the abnormal behavior deviation degree and the audio feature matrix, and an intrusion report is sent to the security center based on the intrusion probability.

[0039] It can be seen that in this application, the behavior pattern of the target person can be accurately identified, and the intrusion threat of the target person can be evaluated in combination with audio features; among them, by constructing a behavior topology map, the behavior pattern of the target person in the intelligent security area can be systematically analyzed, and its normal and abnormal behaviors can be identified. Then, by extracting abnormal behavior nodes, activities that are significantly different from normal behaviors can be timely identified, thereby realizing real-time monitoring and response, and the calculation of behavioral abnormality deviation can provide a quantitative indicator for each behavior pattern; then, by extracting the security audio segment of the target person, a more comprehensive behavioral feature can be obtained, which is helpful to identify potential abnormal behavior and intrusion behavior, thereby enhancing the monitoring capability of the security system, and combining audio and video data to realize multimodal analysis, and constructing an audio feature matrix can help identify the behavior pattern of the target person; finally, based on the behavioral abnormality deviation and the audio feature matrix, the probability of the target person intruding into the sensitive area in the intelligent security area can be more accurately judged, so that a report can be sent to the security center in time to take security measures.

[0040] In summary, the technical solution adopted in this application can accurately identify the behavior pattern of the target person and combine the audio features to evaluate the intrusion threat of the target person, so as to improve the safety and reliability of the intelligent security system. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following is a brief introduction to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only the embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0042] Figure 1 This is an exemplary flow chart of the intelligent security method based on AI analysis provided by this application;

[0043] Figure 2 is an exemplary flow chart for extracting abnormal behavior nodes provided by this application;

[0044] Figure 3 is an exemplary flow chart for determining a security audio segment according to the present application;

[0045] Figure 4 This is a module structure diagram of the intelligent security system based on AI analysis provided by this application;

[0046] Figure 5 This is a structural diagram of a computer device for implementing an intelligent security method based on AI analysis provided in this application. DETAILED DESCRIPTION

[0047] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0048] The embodiment of the present application provides an intelligent security method based on AI analysis, the core of which is to collect surveillance video information in an intelligent security area and extract the behavioral information of a target person in the surveillance video information; construct a behavioral topology map of the target person in the intelligent security area based on the behavioral information, extract abnormal behavior nodes from the behavioral topology map, and then determine the abnormal deviation of the target person's behavior in the intelligent security area through the behavioral deviation characteristics between each abnormal behavior node; obtain the audio data stream of the target person in the intelligent security area, convert the audio data stream into a security audio segment of the target person, and construct an audio feature matrix of the target person in the intelligent security area based on the security audio segment; determine the probability of the target person invading a sensitive area in the intelligent security area based on the abnormal behavioral deviation and the audio feature matrix, and send an intrusion report to the security center based on the intrusion probability. The above scheme can accurately identify the behavioral pattern of the target person and evaluate the intrusion threat of the target person in combination with the audio features to improve the safety and reliability of the intelligent security system.

[0049] Example 1

[0050] In order to better understand the above technical solution, the following will be described in detail with reference to the accompanying drawings and specific implementation methods. Figure 1 As shown in FIG, this figure is an exemplary flow chart of an intelligent security method based on AI analysis according to this embodiment of the present application, and the security method includes the following steps:

[0051] In step S1, surveillance video information in the intelligent security area is collected, and behavior information of a target person in the surveillance video information is extracted.

[0052] In specific implementation, first, high-definition surveillance cameras can be deployed in the smart security area to ensure coverage of all sensitive areas in the smart security area, such as entrances and exits, valuables areas, etc.; then, the cameras can be configured to record real-time video and transmit the data to storage devices or cloud platforms, that is, to collect surveillance video information in the smart security area.

[0053] In this embodiment, the behavior information of the target person in the surveillance video information can be extracted in the following manner, namely:

[0054] Extracting facial feature information from the surveillance video information;

[0055] Based on the comparison between the facial feature information and the candidate facial feature information, a target person entering the intelligent security area is determined;

[0056] Obtaining image information of a target person in the surveillance video information;

[0057] Behavioral features are extracted from the image information to obtain behavioral information of the target person in the monitoring video information.

[0058] In specific implementations, face detection algorithms (such as Haar cascade classifiers or MTCNN) can be used to detect and extract facial feature information from surveillance video. This facial feature information includes key points, outlines, and depth features of all faces appearing in the smart security zone. The extracted facial feature information is then compared with candidate facial feature information. A deep learning model (such as FaceNet or VGGFace) can be used to calculate the similarity between faces. Faces with a similarity below a set threshold are then identified as targets entering the smart security zone and requiring monitoring. After identifying the target person, image information of the target person is extracted from the surveillance video. This image information includes both static image frames and motion trajectories for subsequent analysis. Finally, behavioral features are extracted from the target person's image information. A pose estimation algorithm (such as OpenPose or AlphaPose) can be used to analyze the person's posture and motion, extracting key behavioral features (such as walking, running, staying, and interacting). These extracted behavioral features are then organized into a dataset, namely, the target person's behavioral information. This behavioral information records the target person's activities within the smart security zone, including the type, duration, and frequency of the behavior.

[0059] In step S2, a behavior topology map of the target person in the smart security area is constructed based on the behavior information, abnormal behavior nodes are extracted from the behavior topology map, and then the abnormal deviation degree of the target person's behavior in the smart security area is determined through the behavior deviation characteristics between each abnormal behavior node.

[0060] In this embodiment, the behavior topology map of the target person in the intelligent security area can be constructed based on the behavior information in the following manner:

[0061] Divide the intelligent security area into intervals to obtain a grid map of the intelligent security area;

[0062] The behavior information is mapped to a grid map of the intelligent security area to obtain a behavior topology map of the target person in the intelligent security area.

[0063] In the specific implementation, first, the smart security area is divided into intervals to form a grid map, that is, the grid map of the smart security area. Each grid represents a specific small area. The details and computational efficiency can be balanced by setting an appropriate grid size; then, the position coordinates of the target person in the behavior information are converted into corresponding grids in the grid map to determine the target person's activities in each grid. It should be noted that in the grid map, each grid is a node, recording the target person's behavior information in the grid. By observing the target person's movement between different grids, edge connections are established to represent the transfer relationship from one grid to another, and attributes are assigned to each edge, such as residence time, behavior type, transfer frequency, etc., so that a behavioral topology map of the target person in the smart security area can be obtained.

[0064] Preferably, in this embodiment, reference Figure 2 As shown in FIG. 1 , this figure is an exemplary flow chart of extracting abnormal behavior nodes in an embodiment of the present application. In this embodiment, the abnormal behavior node extraction from the behavior topology graph can be implemented by the following steps:

[0065] First, in step S21, the behavior characteristics corresponding to each behavior topology node in the behavior topology graph are determined;

[0066] Then, in step S22, the behavior representation characteristics of the target person are determined based on all the behavior characteristics;

[0067] Next, in step S23, for each sensitive area in the behavior topology map, the behavior risk of each behavior topology node in the sensitive area is determined based on the behavior characterization feature;

[0068] Finally, in step S24, the behavior topology node with the highest behavior risk is used as the abnormal behavior node corresponding to the sensitive area, and then the abnormal behavior nodes corresponding to each sensitive area in the behavior topology graph are obtained.

[0069] In specific implementation, first, the behavioral characteristics corresponding to each behavioral topology node in the behavioral topology graph can be analyzed, such as dwell time, movement speed, and behavior type (such as walking, staying, running, etc.); then, the behavioral representation characteristics of the target person can be determined based on all behavioral characteristics. The behavioral characteristics corresponding to the behavioral topology nodes can be aggregated to form the overall behavioral representation characteristics of the target person, that is, the behavioral representation characteristics of the target person. This can be achieved by calculating the average of all behavioral characteristics; second, the sensitive areas in the behavioral topology graph are identified. For all behavioral topology nodes in each sensitive area, the behavioral risk of each behavioral topology node in the sensitive area is determined based on the behavioral representation characteristics of the target person. The behavioral risk is used to measure the risk level brought about by the behavior performed by the target person at the behavioral topology node. The mean of all edges connected to the behavioral topology node can be calculated, and the product of the result and the behavioral representation characteristics of the target person is used as the behavioral risk of the behavioral topology node. In this way, the behavioral risk of each behavioral topology node in the sensitive area can be obtained; finally, the behavioral topology node with the largest behavioral risk is used as the abnormal behavior node corresponding to the sensitive area. In this way, the abnormal behavior nodes corresponding to each sensitive area in the behavioral topology graph can be obtained.

[0070] In this embodiment, the abnormal behavior deviation degree of the target person in the smart security area is determined by the behavior deviation features between each abnormal behavior node, and the standard deviation of all behavior deviation features is used as the abnormal behavior deviation degree of the target person in the smart security area.

[0071] It should be noted that in this application, the behavior deviation degree represents the degree of deviation between the target person's behavior in the smart security area and the normal behavior. The larger the behavior deviation degree, the more abnormal the target person's behavior in the smart security area. The behavior deviation feature is a feature used to represent the degree of difference between abnormal behavior nodes.

[0072] In the specific implementation, first, for each abnormal behavior node, the behavioral deviation characteristics between the abnormal behavior node and other abnormal behavior nodes are determined. This can be achieved by calculating the difference distance between the abnormal behavior nodes, and the difference distance is used as the behavioral deviation characteristic between the abnormal behavior nodes. The difference distance can be determined by the Euclidean distance; then, the standard deviation of all behavioral deviation characteristics can be used as the abnormal deviation degree of the target person's behavior in the intelligent security area.

[0073] It should be noted that by constructing a behavioral topology map, the behavioral patterns of target persons in the intelligent security area can be systematically analyzed, and their normal and abnormal behaviors can be identified. Then, by extracting abnormal behavior nodes, activities that are significantly different from normal behaviors can be identified in a timely manner, thereby achieving real-time monitoring and response. In addition, the calculation of behavioral abnormality deviation can provide a quantitative indicator for each behavioral pattern.

[0074] In step S3, an audio data stream of the target person in the smart security area is obtained, the audio data stream is converted into a security audio segment of the target person, and an audio feature matrix of the target person in the smart security area is constructed based on the security audio segment.

[0075] In specific implementation, first, highly sensitive audio sensors are deployed in the smart security area to ensure coverage of the entire smart security area; then, the audio sensors are configured for real-time recording, using an appropriate sampling rate (such as 44.1 kHz or 48 kHz) to ensure sound quality and capture the target person's voice details. The recorded audio data stream is then transmitted in real time to local storage or a cloud server to obtain the target person's audio data stream in the smart security area.

[0076] Preferably, in this embodiment, reference Figure 3 As shown in FIG. 1 , this figure is an exemplary flow chart of determining a security audio segment in an embodiment of the present application. In this embodiment, converting the audio data stream into a security audio segment of a target person can be implemented by the following steps:

[0077] First, in step S31, the short-time energy and zero-crossing rate of the audio data stream are determined;

[0078] Then, in step S32, the audio data stream is converted according to the short-time energy and the zero-crossing rate, thereby obtaining a security audio segment of the target person.

[0079] In specific implementation, first, the short-time energy of each frame of audio in the audio data stream can be determined, and the average of the short-time energies of all audio frames can be used as the short-time energy of the audio data stream. The short-time energy is used to reflect the characteristics of the voiced part in the audio data stream. The zero-crossing rate of each frame of audio in the audio data stream can be determined, and the average of the zero-crossing rates of all audio frames can be used as the zero-crossing rate of the audio data stream. The zero-crossing rate is used to reflect the characteristics of the unvoiced part in the audio data stream. Then, the audio data stream can be converted according to the short-time energy and the zero-crossing rate, that is, the audio data stream can be segmented using a dual-threshold endpoint detection algorithm, so that the security audio segment of the target person can be obtained. The security audio segment refers to the audio of the target person after removing noise interference, which can be achieved by using the short-time energy and the zero-crossing rate as the thresholds in the dual-threshold endpoint detection algorithm, which will not be repeated here.

[0080] In this embodiment, constructing the audio feature matrix of the target person in the smart security area based on the security audio segment is to extract features from the security audio segment, and then constructing the audio feature matrix of the target person in the smart security area based on the extracted audio features.

[0081] It should be noted that in this application, audio features include scale features, pitch features, and unvoiced sound fluctuation features. Among them, the scale feature is used to represent the overall scale characteristics of the security audio segment, the pitch feature can be used to distinguish the high-pitched part in the security audio segment, and the unvoiced sound fluctuation feature is used to represent the degree of fluctuation of the unvoiced part in the security audio segment.

[0082] In the specific implementation, first, feature extraction is performed on the security audio segment, and the scale feature of the security audio segment can be calculated by the Mel filter bank; secondly, the fundamental frequency value of each frame of audio in the security audio segment can be extracted by the cepstrum method, so as to obtain all the fundamental frequency values, which represent the basic frequency value of the audio frame, and the mean of all the fundamental frequency values ​​can be used as the fundamental pitch feature of the security audio segment; then, the zero-crossing rate variance of each frame of audio in the security audio segment can be calculated, so as to use the mean of all zero-crossing rate variances as the unvoiced sound fluctuation feature of the security audio segment; finally, the audio feature matrix of the target person in the intelligent security area is constructed by the scale feature, the fundamental pitch feature and the unvoiced sound fluctuation feature, that is, the feature matrix composed of the scale feature, the fundamental pitch feature and the unvoiced sound fluctuation feature is used as the audio feature matrix of the target person in the intelligent security area.

[0083] It should be noted that by extracting the security audio segments of the target person, more comprehensive behavioral features can be obtained, which helps to identify potential abnormal behaviors and intrusions, thereby enhancing the monitoring capabilities of the security system. In addition, by combining audio and video data, multimodal analysis can be achieved to provide more accurate behavior recognition. Audio features can supplement the deficiencies of video surveillance. For example, in the case of insufficient light or obstructed vision, audio information can still provide important clues. Constructing an audio feature matrix can help identify the behavioral patterns of the target person.

[0084] In step S4, the probability of the target person invading the sensitive area in the intelligent security area is determined according to the abnormal behavior deviation and the audio feature matrix, and an intrusion report is sent to the security center based on the intrusion probability.

[0085] In this embodiment, the probability of a target person invading a sensitive area in the intelligent security area can be determined based on the abnormal behavior deviation and the audio feature matrix in the following manner, namely:

[0086] Get pre-built Bayesian network probability models;

[0087] The abnormal behavior deviation and the audio feature matrix are input as input data into the Bayesian network probability model for prediction, so as to obtain the probability of the target person invading the sensitive area in the intelligent security area.

[0088] In the specific implementation, first, a Bayesian network probability model is pre-built. The model should contain nodes related to the target person's behavior, audio features and intrusion probability. The structure and conditional probability of the Bayesian network can be trained using historical data to ensure that the model can effectively predict the intrusion probability; then, the behavioral abnormality deviation and audio feature matrix are standardized to ensure that the input data is within a reasonable range to facilitate the Bayesian network calculation, and the input data is organized into a format suitable for the Bayesian network to ensure that each feature corresponds to the corresponding node of the Bayesian network; then, the prepared behavioral abnormality deviation and audio feature matrix are input as input data into the Bayesian network probability model, and the reasoning process is executed. Through the reasoning mechanism of the Bayesian network, the probability of the target person invading the sensitive area in the intelligent security area is obtained. The intrusion probability will reflect the risk level of the target person entering the sensitive area.

[0089] In this embodiment, an intrusion report is sent to the security center based on the intrusion probability. In specific implementation, the calculated intrusion probability is used as one of the key data of the report, and includes basic information of the target person (such as facial features, abnormal behavior deviation), audio feature analysis results (such as fundamental pitch features, voiceless sound fluctuation features) and the specific location of the sensitive area. The time of the incident is recorded to ensure that the security center can track the timeline of the incident, and thus provide corresponding response suggestions based on the level of intrusion probability, such as whether security personnel need to be mobilized immediately.

[0090] It can be seen that in this application, the behavior pattern of the target person can be accurately identified, and the intrusion threat of the target person can be evaluated in combination with audio features; among them, by constructing a behavior topology map, the behavior pattern of the target person in the intelligent security area can be systematically analyzed, and its normal and abnormal behaviors can be identified. Then, by extracting abnormal behavior nodes, activities that are significantly different from normal behaviors can be timely identified, thereby realizing real-time monitoring and response, and the calculation of behavioral abnormality deviation can provide a quantitative indicator for each behavior pattern; then, by extracting the security audio segment of the target person, a more comprehensive behavioral feature can be obtained, which is helpful to identify potential abnormal behavior and intrusion behavior, thereby enhancing the monitoring capability of the security system, and combining audio and video data to realize multimodal analysis, and constructing an audio feature matrix can help identify the behavior pattern of the target person; finally, based on the behavioral abnormality deviation and the audio feature matrix, the probability of the target person intruding into the sensitive area in the intelligent security area can be more accurately judged, so that a report can be sent to the security center in time to take security measures.

[0091] In summary, the technical solution adopted in this application can accurately identify the behavior pattern of the target person and combine the audio features to evaluate the intrusion threat of the target person, so as to improve the safety and reliability of the intelligent security system.

[0092] Example 2

[0093] This application provides an intelligent security system based on AI analysis, refer to Figure 4 As shown in FIG, this figure is a module structure diagram of the security system shown in this embodiment of the present application, and the security system includes:

[0094] The acquisition module 100 is used to collect surveillance video information in the intelligent security area and extract the behavior information of the target person in the surveillance video information;

[0095] Abnormal behavior determination module 200, configured to construct a behavior topology map of the target person in the intelligent security area based on the behavior information, extract abnormal behavior nodes from the behavior topology map, and then determine the abnormal deviation degree of the target person's behavior in the intelligent security area based on the behavior deviation characteristics between each abnormal behavior node;

[0096] An audio feature extraction module 300 is configured to obtain an audio data stream of a target person in an intelligent security zone, convert the audio data stream into a security audio segment of the target person, and construct an audio feature matrix of the target person in the intelligent security zone based on the security audio segment;

[0097] The intrusion security module 400 is used to determine the probability of a target person invading a sensitive area in the intelligent security area based on the abnormal behavior deviation and the audio feature matrix, and send an intrusion report to the security center based on the intrusion probability.

[0098] The above describes in detail the examples of the intelligent security method and system based on AI analysis provided by the embodiments of the present application. It can be understood that in order to realize the above functions, the corresponding device includes hardware structures and / or software modules corresponding to the execution of each function. It should be easily appreciated by those skilled in the art that, in combination with the units and algorithm steps of each example described in the embodiments disclosed herein, the present application can be implemented in the form of hardware or a combination of hardware and computer software. Whether a function is executed in the form of hardware or computer software driving hardware depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0099] Example 3

[0100] The present application also provides a computer device, which includes a memory and a processor, wherein the memory is used to store a computer program, and the processor is used to call and run the computer program from the memory, so that the computer device executes the above-mentioned intelligent security method based on AI analysis.

[0101] In this embodiment, reference Figure 5 , the dotted line in the figure indicates that the unit or module is optional. The figure is a structural diagram of a computer device according to an AI analysis-based intelligent security method provided in an embodiment of the present application. The above-mentioned AI analysis-based intelligent security method in the above embodiment can be achieved by Figure 5 The computer device shown in the figure is implemented, and the computer device includes at least one processor 501, a memory 502 and at least one communication unit 505. The computer device can be a terminal device, a server or a chip.

[0102] The processor 501 may be a general-purpose processor or a dedicated processor. For example, the processor 501 may be a central processing unit (CPU). The CPU may be used to control the computer device, execute software programs, and process data from the software programs. The computer device may also include a communication unit 505 to implement signal input (reception) and output (transmission).

[0103] For example, the computer device may be a chip, the communication unit 505 may be an input and / or output circuit of the chip, or the communication unit 505 may be a communication interface of the chip, and the chip may be a component of a terminal device, a network device, or other device.

[0104] For another example, the computer device may be a terminal device or a server, and the communication unit 505 may be a transceiver of the terminal device or the server, or the communication unit 505 may be a transceiver circuit of the terminal device or the server.

[0105] Computer device 500 may include one or more memories 502, on which a program 504 is stored. Program 504 can be executed by processor 501 to generate instructions 503, causing processor 501 to execute the method described in the above method embodiment according to instructions 503. Optionally, memory 502 may also store data (such as a target audit model). Optionally, processor 501 may also read data stored in memory 502. This data may be stored at the same memory address as program 504, or at a different memory address.

[0106] The processor 501 and the memory 502 may be provided separately or integrated together, for example, integrated on a system on chip (SOC) of a terminal device.

[0107] It should be understood that each step of the above method embodiment can be completed by a hardware-based logic circuit or software-based instructions in the processor 501. The processor 501 can be a central processing unit, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA), or other programmable logic devices, such as discrete gates, transistor logic devices, or discrete hardware components.

[0108] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0109] Example 4

[0110] The present application also provides a computer-readable storage medium, which stores instructions or codes. When the instructions or codes are run on a computer, the computer implements the above-mentioned intelligent security method based on AI analysis when executing the computer.

[0111] Although the preferred embodiments of the present application have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present application.

[0112] Obviously, those skilled in the art may make various changes and modifications to the present application without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present application fall within the scope of the claims and their equivalents, the present application is intended to include such modifications and variations.

Claims

1. An intelligent security method based on AI analysis, characterized in that: The security method comprises the following steps: Collecting surveillance video information in the intelligent security area and extracting behavioral information of the target person in the surveillance video information; constructing a behavior topology map of the target person in the intelligent security area based on the behavior information, extracting abnormal behavior nodes from the behavior topology map, and then determining the abnormal deviation degree of the target person's behavior in the intelligent security area based on the behavior deviation characteristics between each abnormal behavior node; Acquire an audio data stream of a target person in an intelligent security area, convert the audio data stream into a security audio segment of the target person, and construct an audio feature matrix of the target person in the intelligent security area based on the security audio segment, wherein the security audio segment refers to the audio of the target person after noise interference is removed; The probability of a target person invading a sensitive area in an intelligent security area is determined according to the abnormal behavior deviation and the audio feature matrix, and an intrusion report is sent to a security center based on the intrusion probability.

2. The intelligent security method based on AI analysis according to claim 1, characterized in that: Extracting the behavior information of the target person from the surveillance video information specifically includes: Extracting facial feature information from the surveillance video information; Based on the comparison between the facial feature information and the candidate facial feature information, a target person entering the intelligent security area is determined; Obtaining image information of a target person in the surveillance video information; Behavioral features are extracted from the image information to obtain behavioral information of the target person in the monitoring video information.

3. The intelligent security method based on AI analysis according to claim 1, characterized in that: Constructing a behavior topology map of the target person in the intelligent security area based on the behavior information specifically includes: Divide the intelligent security area into intervals to obtain a grid map of the intelligent security area; The behavior information is mapped to a grid map of the intelligent security area to obtain a behavior topology map of the target person in the intelligent security area.

4. The intelligent security method based on AI analysis according to claim 1, characterized in that: Extracting abnormal behavior nodes from the behavior topology graph specifically includes: Determine the behavior characteristics corresponding to each behavior topology node in the behavior topology graph; Determine the target person's behavioral characteristics based on all behavioral characteristics; For each sensitive area in the behavior topology map, determining the behavior risk of each behavior topology node in the sensitive area based on the behavior characterization feature; The behavior topology node with the largest behavior risk is used as the abnormal behavior node corresponding to the sensitive area, and then the abnormal behavior nodes corresponding to each sensitive area in the behavior topology graph are obtained.

5. The intelligent security method based on AI analysis according to claim 1, characterized in that: Converting the audio data stream into a security audio segment of the target person specifically includes: Determining the short-term energy and zero-crossing rate of the audio data stream; The audio data stream is converted according to the short-time energy and the zero-crossing rate to obtain a security audio segment of the target person.

6. The intelligent security method based on AI analysis according to claim 1, characterized in that: Constructing the audio feature matrix of the target person in the smart security area based on the security audio segment is to extract features from the security audio segment, and then construct the audio feature matrix of the target person in the smart security area according to the extracted audio features.

7. The intelligent security method based on AI analysis according to claim 1, characterized in that: Determining the probability of a target person intruding into a sensitive area in the intelligent security area based on the abnormal behavior deviation and the audio feature matrix specifically includes: Get pre-built Bayesian network probability models; The abnormal behavior deviation and the audio feature matrix are input as input data into the Bayesian network probability model for prediction, so as to obtain the probability of the target person invading the sensitive area in the intelligent security area.

8. An intelligent security system based on AI analysis, used to execute the intelligent security method based on AI analysis according to any one of claims 1 to 7, characterized in that: The security system includes: The acquisition module is used to collect surveillance video information in the intelligent security area and extract the behavior information of the target person in the surveillance video information; an abnormal behavior determination module, configured to construct a behavior topology map of the target person in the intelligent security area based on the behavior information, extract abnormal behavior nodes from the behavior topology map, and then determine the abnormal deviation degree of the target person's behavior in the intelligent security area based on the behavior deviation characteristics between each abnormal behavior node; An audio feature extraction module is used to obtain an audio data stream of a target person in an intelligent security area, convert the audio data stream into a security audio segment of the target person, and construct an audio feature matrix of the target person in the intelligent security area based on the security audio segment; The intrusion security module is used to determine the probability of a target person invading a sensitive area in the intelligent security area based on the abnormal behavior deviation and the audio feature matrix, and send an intrusion report to the security center based on the intrusion probability.

9. A computer device, characterized in that: The computer device includes a memory and a processor, the memory is used to store a computer program, and the processor is used to call and run the computer program from the memory, so that the computer device executes the intelligent security method based on AI analysis according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores instructions or codes, and when the instructions or codes are executed on a computer, the computer implements the intelligent security method based on AI analysis according to any one of claims 1 to 7.

Images