Intelligent positioning terminal integrating remote video recording and public network talkback functions

Abstract

The invention discloses an intelligent positioning terminal integrating remote video recording and public network talkback functions, and belongs to the technical field of recorders. The system comprises a terminal main body, a video acquisition module, a public network talkback module, a positioning module, a danger identification module, a control module and a remote communication module, multiple functions are realized through module integration, the number of devices carried by a user is greatly reduced, the equipment burden in outdoor operation, law enforcement and other scenes is reduced, meanwhile, the problem of data splitting among multiple devices is avoided, and the user experience is improved. The danger identification module extracts object, behavior and scene features, matches the features with a danger feature library after weighted fusion, integrates positioning information to generate alarm data when identifying potential dangers, realizes upgrading of safety early warning from passive recording to active prevention and control, ensures communication and positioning continuity of equipment in a complex environment through multi-mode communication and multi-source positioning, and improves the safety of the equipment. And meanwhile, the remote control capability is improved, and the environmental adaptability and the intelligent control level of the equipment are remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of recorders, in particular to an intelligent positioning terminal integrating remote video recording and public network intercom functions. BACKGROUND

[0002] In the fields of law enforcement, security, outdoor work, etc., staff often need to rely on multiple devices to achieve video recording, voice communication and location tracking functions at the same time, for example, collecting on-site videos through a law enforcement recorder, realizing remote voice interaction through a public network intercom, and feeding back real-time positions through a locator. At present, these functions are mostly realized by independent devices respectively, which has the problems of too many devices to carry and complicated operation. Staff need to frequently switch devices when working outdoors, which not only increases the burden of equipment, but also may cause key information to be unable to be synchronized and associated due to data fragmentation between devices, affecting the comprehensive judgment of the remote management platform on the on-site situation.

[0003] From the aspects of video recording and danger warning, the traditional law enforcement recorder only has the function of passively collecting and storing videos, and cannot actively identify potential dangers on site, so it needs to rely on staff to manually discover and trigger alarms, which leads to a lag in emergency response. At the same time, the traditional public network intercom device relies on stable public network signals, and is prone to communication interruption in weak signal or no signal areas such as remote mountainous areas, underground passages and sealed workshops, which cannot meet the voice interaction needs in cross-regional operations or extreme environments. The positioning device mostly uses single satellite positioning or base station positioning method, and the satellite positioning is prone to failure in signal shielding areas such as indoors and underground, and the base station positioning has a large decrease in accuracy in areas with weak base station coverage, making it difficult to realize continuous positioning in all scenarios. SUMMARY

[0004] The present application aims to provide an intelligent positioning terminal integrating remote video recording and public network intercom functions to solve the problems raised in the background.

[0005] To achieve the above-mentioned purpose, the present application provides the following technical solution: an intelligent positioning terminal integrating remote video recording and public network intercom functions, comprising a terminal main body and a functional module group integrated in the interior of the terminal main body, wherein the functional module group comprises a video acquisition module, a public network intercom module, a positioning module, a danger identification module, a control module and a remote communication module.

[0006] The video acquisition module is used to acquire real-time video data around the terminal and transmit the real-time video data to the control module.

[0007] The public network intercom module is used to establish a public network voice communication link between the terminal and external devices, realize two-way voice interaction, and synchronously transmit voice interaction data to the control module.

[0008] The positioning module is configured to acquire real-time position information of the terminal and transmit the real-time position information to the control module;

[0009] The danger identification module is configured to implement potential danger analysis and alarm triggering.

[0010] The control module is configured to receive and process data transmitted by the video acquisition module, the public network intercom module, the positioning module and the danger identification module, establish a data interaction channel with the remote management platform through the remote communication module, and realize remote transmission of data and receiving of instructions.

[0011] Further, the danger identification module is electrically connected with the control module, the video acquisition module and the positioning module, and the specific process of the danger identification module to implement potential danger analysis and alarm triggering includes:

[0012] Acquiring video data collected by the video acquisition module in real time, and extracting frames from the video data to obtain continuous video frame images;

[0013] Performing feature recognition on each video frame image to extract target object features, human behavior features and environmental scene features in the image;

[0014] Matching the extracted features with feature data in a preset danger feature library, the preset danger feature library at least including violent behavior features, dangerous instrument features and abnormal crowd gathering features;

[0015] When the matching result meets a preset danger triggering condition, it is determined that a potential danger is recognized, and real-time position information output by the positioning module is acquired at this time;

[0016] Integrating the potential danger recognition result and the real-time position information into alarm data, triggering the remote communication module through the control module, sending the alarm data to the remote management platform, and synchronously triggering a local sound and light alarm of the terminal main body.

[0017] Further, when the danger identification module performs feature recognition on the video frame image, a multi-dimensional feature fusion algorithm is adopted, specifically including:

[0018] For target object features, an edge detection algorithm is used to extract object contours, color histogram analysis is combined to analyze object color distribution, and an object feature vector is generated;

[0019] For human behavior features, a skeleton key point detection algorithm is used to acquire human joint coordinates, the change amount of joint coordinates in adjacent frames is calculated, the type of human action is determined, a behavior feature vector is generated;

[0020] For environmental scene features, a semantic segmentation algorithm is used to divide scene areas, identify danger area markers in the scene, and generate a scene feature vector.

[0021] The object feature vector, behavior feature vector, and scene feature vector are weighted and fused to obtain a comprehensive feature vector. Then, the similarity is calculated with the feature vector in the preset danger feature library. If the similarity is greater than the preset threshold, it is determined that the danger feature matching condition is met.

[0022] Furthermore, the hazard identification module also includes a hazard level classification unit, which is used to classify potential hazards into level one, level two, and level three hazards based on the degree of matching of hazard features and the urgency of the scenario.

[0023] The first-level danger is defined as the presence of violent behavior or dangerous equipment characteristics, and the presence of clear personnel conflict actions in the video frame, which is determined to be an emergency danger. At this time, the control module triggers the highest priority alarm, and the remote communication module uses preemptive transmission to send the alarm data to the remote management platform and simultaneously dials the preset emergency contact number.

[0024] The level 2 hazard is determined by matching the characteristics of abnormal crowd gathering, which is identified as a potential hazard. The abnormal crowd gathering characteristics are that the number of people gathered is ≥10 and the duration is ≥5 minutes. The control module triggers a medium-priority alarm, the remote communication module sends alarm data according to the regular data transmission queue, and starts the high-definition recording mode of the video acquisition module.

[0025] The Level 3 hazard is determined by matching suspected hazard features. The suspected hazard features include vague outlines of dangerous equipment and unclear abnormal behavior. The control module triggers a low-priority alarm, stores only the suspicious information and real-time location locally, displays the information on the terminal screen, and asks the user to confirm whether to upgrade the alarm level.

[0026] Furthermore, the public network intercom module supports multi-mode communication switching, specifically including:

[0027] When the terminal is in an area with good public network signal and the signal strength is ≥-70dBm, it will automatically establish a voice link using the 4G / 5G public network communication mode.

[0028] When the terminal is in an area with weak public network signal, and -90dBm < signal strength < -70dBm, it automatically switches to narrowband IoT communication mode.

[0029] When the terminal is in an area without public network signal and the signal strength is ≤-90dBm, the terminal’s built-in emergency intercom function is activated. It establishes a temporary communication network with similar terminals in the vicinity through short-range wireless communication, and automatically uploads the local intercom records to the remote management platform after the public network signal is restored.

[0030] Furthermore, the positioning module employs multi-source positioning fusion technology, specifically including:

[0031] Prioritize obtaining high-precision location information of the terminal through satellite positioning;

[0032] When satellite positioning signal is blocked, causing positioning failure, it automatically switches to base station positioning. By obtaining the signal parameters of at least 3 surrounding communication base stations, the terminal's location information is calculated.

[0033] When the base station positioning signal is also weak, Wi-Fi positioning assistance is activated. By scanning the MAC addresses of surrounding Wi-Fi hotspots and matching them with a preset Wi-Fi location database, the approximate location area of ​​the terminal is determined.

[0034] The control module performs weighted fusion of the results from satellite positioning, base station positioning, and Wi-Fi positioning, dynamically adjusts the weights according to the signal strength of different positioning methods, and outputs the optimal real-time location information.

[0035] Furthermore, when sending alarm data, the remote communication module will also simultaneously transmit historical video clips from 10 minutes before the alarm occurred to the moment the alarm was triggered;

[0036] The remote management platform also sends remote control commands to the terminal through the remote communication module. The remote control commands include at least starting real-time video transmission, adjusting the hazard identification sensitivity, and enabling public network intercom broadcast. After receiving the commands, the control module drives the corresponding functional modules to perform the operations and feeds back the execution results to the remote management platform.

[0037] Furthermore, the front of the terminal body is equipped with a camera that interacts with the video acquisition module, the back of the terminal body is equipped with a display screen, the sides of the terminal body are equipped with physical buttons that interact with the control module, and the sides of the terminal body are respectively equipped with a charging port and a memory card slot.

[0038] Furthermore, the intelligent positioning terminal, which integrates remote video recording and public network intercom functions, also includes a dynamic hazard assessment and optimization module. This module is used to optimize the accuracy and response efficiency of hazard identification in complex dynamic scenarios, solving the problem of misjudgment or missed judgment of hazards caused by environmental changes, the moving speed of target objects, and video data noise. Specifically, it includes the following steps:

[0039] The system acquires real-time video data from the video acquisition module, obtains the terminal's real-time location information from the positioning module, and acquires environmental sound features from the voice interaction data from the public network intercom module.

[0040] Based on real-time video data, real-time location information, and ambient sound characteristics, a dynamic hazard assessment index is calculated to quantify the degree of danger in the current scene. The formula is as follows:

[0041] DDEI=(w1×VFI+w2×MSI+w3×AFI) / (1+e^(-k×(STI-S0)))

[0042] Wherein, DDEI is the Dynamic Hazard Assessment Index, and a higher value indicates a higher potential hazard level in the current scene; VFI is the Video Frame Interference Factor, calculated by analyzing the image sharpness of video frames, the normalized motion speed of the target object, and the normalized brightness change rate, with the formula VFI = w v ×(1-C)×(V / V max )+w l ×(L / L max ), where C is the image sharpness, V is the target object's speed, and V max The preset maximum motion speed is given by L, which represents the rate of change of brightness (unit: cd / m). 2 / s), L max To preset the maximum rate of change in brightness, w v and w l Here are the video factor weight coefficients, and w v +w l =1; MSI is the moving speed influence factor, which is obtained by comparing the terminal moving speed output by the positioning module with the preset speed threshold V0. The formula is MSI = min(|V_terminal - V0| / V0, 1), where V_terminal is the real-time moving speed of the terminal; AFI is the ambient sound feature factor, which is obtained by comparing the ambient sound intensity in the voice interaction data extracted by the public network intercom module with the preset noise threshold N0. The formula is AFI = min((N-N0) / N0, 1), where N is the real-time ambient sound intensity; w1, w2 w1 and w2 are dimensionless weighting coefficients, corresponding to the influence weights of video frame interference factor, motion speed influence factor, and ambient sound feature factor, respectively, and w1 + w2 + w3 = 1. The specific values ​​are dynamically adjusted according to the scene type. STI is the scene time index, calculated based on the timestamp change rate of video frames and the time synchronization data of the positioning module. The formula is STI = ΔT / T0, where ΔT is the time difference between adjacent frames, T0 is the preset time base, S0 is the scene time index base value, k is the adjustment coefficient used to control the sensitivity of the exponential function, and e is the base of the natural logarithm.

[0043] Based on the calculated DDEI, the dynamic hazard assessment optimization module performs the following operations:

[0044] When DDEI≥0.7, it is determined to be a high-risk scenario. The control module automatically increases the feature matching threshold of the hazard identification module, triggers the first-level hazard alarm first, and uploads the high-risk alarm data and real-time video stream to the remote management platform through the remote communication module.

[0045] When 0.3≤DDEI<0.7, it is determined to be a medium-risk scenario. The control module maintains the existing hazard identification sensitivity, triggers a level-two hazard alarm, and starts the high-resolution mode of the video acquisition module.

[0046] When DDEI < 0.3, it is determined to be a low-risk scenario. The control module lowers the feature matching threshold and only records potentially dangerous data to local storage, waiting for further confirmation from the user.

[0047] Furthermore, the intelligent positioning terminal integrating remote video recording and public network intercom functions also includes an adaptive power management module. This module dynamically adjusts power distribution and power consumption modes based on the terminal's operating status, environmental conditions, and user interaction needs to extend the terminal's battery life and ensure stable operation of critical functions in low-power scenarios. Specifically, it includes the following steps:

[0048] The adaptive power management module monitors the battery level of the terminal body, the operating status of the video acquisition module, the communication strength of the public network intercom module, the positioning frequency of the positioning module, and the computing load of the hazard identification module in real time, and dynamically classifies the terminal's working state into high load mode, standard mode, and low power mode.

[0049] In high-load mode, the video acquisition module operates at the highest resolution, the public network intercom module maintains a continuous voice link, the positioning module updates location information at the highest frequency, and the hazard identification module enables all feature recognition algorithms; the adaptive power management module prioritizes allocating power resources to the video acquisition module and the hazard identification module to ensure the real-time performance of hazard identification and the integrity of video data, while limiting the background operation of non-essential functions.

[0050] In standard mode, the video acquisition module operates at medium resolution, the public network intercom module establishes voice links as needed, the positioning module updates location information at a medium frequency, and the hazard identification module only enables the core feature recognition algorithm; the adaptive power management module balances the power distribution of each functional module and optimizes power consumption by reducing the brightness of the display screen and reducing the frequency of non-emergency data transmission of the remote communication module.

[0051] In low-power mode, the video acquisition module pauses real-time recording and only starts when the hazard identification module triggers an alarm; the public network intercom module switches to narrowband communication mode or short-range wireless communication; the positioning module reduces the update frequency to the lowest threshold; and the hazard identification module only analyzes key hazard features. The adaptive power management module shuts down non-core hardware components, such as the display backlight and secondary sensors, and sends a low-battery warning to the remote management platform through the remote communication module when the battery level is below the power threshold.

[0052] The adaptive power management module also dynamically adjusts the charging strategy based on ambient temperature and terminal usage time. When the ambient temperature is higher than the first temperature threshold or lower than the second temperature threshold, it reduces the charging current to protect battery life. When it detects that the user has not operated the terminal for a long time, it automatically enters standby mode, keeping only the positioning module and remote communication module running at the lowest power consumption.

[0053] Compared with the prior art, the beneficial effects of the present invention are:

[0054] 1. The multi-module integration and hardware optimization layout of this invention achieves multi-functionality through module integration, significantly reducing the number of devices users carry and alleviating the equipment burden in outdoor operations, law enforcement, and other scenarios. It also avoids data fragmentation between multiple devices. The control module can process data from each module in a coordinated manner, enabling multi-dimensional information synchronous transmission and allowing the remote management platform to obtain more comprehensive on-site data. Full-function physical buttons cover core operations such as taking photos, recording audio, intercom, and location check-in, triggering functions without external devices and adapting to the rapid operation needs in complex scenarios. The separate charging port and memory card slot ensure convenient battery replenishment and support local data expansion storage, preventing the loss of critical data due to remote transmission delays and significantly improving the adaptability and reliability of the device in law enforcement, security, and outdoor operations scenarios.

[0055] 2. The hazard identification module of this invention extracts physical, behavioral, and scene features, weights and fuses them, and matches them with a hazard feature database. When identifying potential hazards, it integrates location information to generate alarm data, simultaneously triggering remote transmission and local audible and visual alarms. This upgrades safety warnings from passive recording to proactive prevention. Through a multi-dimensional feature fusion algorithm, it comprehensively covers diverse information in hazardous scenarios. The linkage between location information and alarm data allows the remote management platform to quickly locate the location of the hazard and promptly allocate surrounding resources. The local audible and visual alarms can alert personnel on-site and prevent the escalation of the hazard.

[0056] 3. The multi-mode communication and multi-source positioning of this invention ensure the continuity of communication and positioning of the device in complex environments, while improving remote control capabilities. The multi-mode switching of public network intercom solves the problem of communication loss of traditional devices in areas with weak or no signal. Multi-source positioning fusion covers the entire scenario, avoiding location loss caused by the failure of a single positioning method, and significantly improving the environmental adaptability and intelligent control level of the device. Attached Figure Description

[0057] Fig. 1 This is a schematic diagram of the internal system modules of the terminal body of the present invention;

[0058] Fig. 2 This is a schematic diagram of the front structure of the terminal body of the present invention;

[0059] Fig. 3This is a schematic diagram of the rear structure of the terminal body of the present invention.

[0060] In the image: 1. Terminal body; 2. Camera; 3. Display screen; 4. Physical buttons; 5. Charging port; 6. Memory card slot. Detailed Implementation

[0061] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0062] Please see Figs. 1-3 The present invention provides the following technical solutions:

[0063] The intelligent positioning terminal integrates remote video recording and public network intercom functions, including a terminal body 1 and a group of functional modules integrated inside the terminal body 1. The functional module group includes a video acquisition module, a public network intercom module, a positioning module, a hazard identification module, a control module, and a remote communication module.

[0064] The video acquisition module is used to collect real-time video data from the surrounding area of ​​the terminal and transmit the real-time video data to the control module;

[0065] The public network intercom module is used to establish a public network voice communication link between the terminal and external devices, realize two-way voice interaction, and synchronously transmit voice interaction data to the control module;

[0066] The positioning module is used to obtain the real-time location information of the terminal and transmit the real-time location information to the control module;

[0067] The hazard identification module is used to analyze potential hazards and trigger alarms;

[0068] The control module is used to receive and process data transmitted from the video acquisition module, public network intercom module, positioning module and hazard identification module. It establishes a data interaction channel with the remote management platform through the remote communication module to realize remote data transmission and command reception.

[0069] The front of the terminal body 1 is equipped with a camera 2 that interacts with the video acquisition module. The back of the terminal body 1 is equipped with a display screen 3. The sides of the terminal body 1 are equipped with physical buttons 4 that interact with the control module. The physical buttons 4 include a photo button, a record button, a video recording button, a menu button, a confirmation button, a return button, up and down control buttons, a walkie-talkie button, a location check-in button, and a power button. The two sides of the terminal body 1 are respectively equipped with a charging port 5 and a memory card slot 6.

[0070] In the above embodiments, by integrating the terminal body with the multi-functional module, the three core functions of public network intercom, video recording and location tracking are deeply integrated, which significantly improves the practicality and scenario adaptability of the device compared with traditional single-function devices.

[0071] In the above embodiments, the front camera 2 of the terminal body 1 interacts directly with the video acquisition module, ensuring the convenience and real-time nature of video acquisition. The rear display screen 3 can intuitively present data information. The physical buttons 4 on both sides cover all-scenario operation needs such as taking photos, recording audio, talking to each other, and location check-in. Core functions can be triggered without additional external devices, greatly reducing the complexity of operation. The separate design of the charging port 5 and the memory card slot 6 not only ensures the convenience of device power replenishment, but also supports local data expansion storage through the memory card slot 6, avoiding the loss of critical data due to remote transmission delays.

[0072] In the above embodiments, the control module serves as the core of data processing, enabling the coordinated processing of data from various modules. Voice data from the public network intercom and location data from the positioning module can be synchronously transmitted to the remote management platform along with video data from the video acquisition module. This allows the remote control terminal to obtain multi-dimensional information including audio, video, and location. It is suitable for scenarios with high demands for multi-dimensional data synchronization, such as law enforcement, security, and outdoor operations. It provides users with an integrated smart terminal solution, reducing the number of devices carried and lowering usage costs and operational burdens.

[0073] The hazard identification module is electrically connected to the control module, video acquisition module, and positioning module, respectively. The specific process by which the hazard identification module performs potential hazard analysis and alarm triggering includes:

[0074] The system acquires real-time video data from the video acquisition module, extracts frames from the video data, and obtains continuous video frame images.

[0075] Feature recognition is performed on each video frame to extract features of target objects, human behavior, and environmental scene.

[0076] The extracted features are matched with feature data in a pre-set hazard feature database, which includes at least violent behavior features, dangerous equipment features, and abnormal crowd gathering features.

[0077] When the matching result meets the preset danger triggering condition, the potential danger is identified, and the real-time location information output by the positioning module is obtained.

[0078] The potential hazard identification results are integrated with real-time location information into alarm data. The remote communication module is triggered by the control module to send the alarm data to the remote management platform and simultaneously trigger the local audible and visual alarm of the terminal.

[0079] When the hazard identification module performs feature recognition on video frame images, it employs a multi-dimensional feature fusion algorithm.

[0080] For the target object features, the object contour is extracted using an edge detection algorithm, and the object color distribution is analyzed by combining a color histogram to generate an object feature vector.

[0081] For human behavioral features, the human joint coordinates are obtained through the skeletal keypoint detection algorithm, the change in joint coordinates in adjacent frames is calculated, the type of human action is determined, and a behavioral feature vector is generated.

[0082] Based on the environmental scene features, the scene regions are divided using a semantic segmentation algorithm, dangerous area markers in the scene are identified, and scene feature vectors are generated.

[0083] The object feature vector, behavior feature vector, and scene feature vector are weighted and fused to obtain a comprehensive feature vector. Then, the similarity is calculated with the feature vector in the preset danger feature library. If the similarity is greater than the preset threshold, it is determined that the danger feature matching condition is met.

[0084] In the above embodiments, the hazard identification module achieves proactive identification and intelligent alarm of potential hazards through a multi-dimensional feature fusion algorithm and a hazard feature database matching mechanism, significantly improving the terminal's safety warning capability and emergency response efficiency. At the same time, it extracts three types of features: target object, human behavior, and environmental scene. It identifies the object features of dangerous instruments through edge detection and color histogram analysis, judges whether there are violent behaviors such as pushing and beating by personnel through skeletal key point detection, and identifies environmental scene features such as flammable material markings and high-altitude fall risk zones through semantic segmentation. The weighted fusion of the three types of features can comprehensively cover the diverse information of dangerous scenes and avoid misjudgment or omission due to the bias of single feature identification. For example, when only the object features of a knife are identified, combining whether the environmental scene is a kitchen or a public place and whether the human behavior is waving or carrying normally can more accurately determine whether it constitutes a hazard.

[0085] In the above embodiments, the linkage and integration of hazard identification and location information allows alarm data to not only include the type of hazard but also provide real-time location information. The remote management platform can quickly locate the location of the hazard and promptly allocate surrounding resources to carry out emergency response. This is especially suitable for scenarios such as campus security, large-scale event site management, and night patrols, providing full-process support for safety management, including identification, location, and alarm, thereby improving the initiative and timeliness of safety management.

[0086] The hazard identification module also includes a hazard level classification unit, which is used to classify potential hazards into Level 1, Level 2, and Level 3 hazards based on the degree of matching of hazard features and the urgency of the scenario.

[0087] Level 1 danger is defined as the presence of violent behavior or dangerous equipment characteristics, and clear personnel conflict actions in the video frame. This is considered an emergency danger. At this time, the control module triggers the highest priority alarm, and the remote communication module uses preemptive transmission to send the alarm data to the remote management platform and simultaneously dials the preset emergency contact number.

[0088] Level 2 hazard is determined by the matching of abnormal crowd gathering characteristics, which is identified as a potential hazard. The abnormal crowd gathering characteristics are that the number of people gathered is ≥10 and the duration is ≥5 minutes. The control module triggers a medium-priority alarm, the remote communication module sends alarm data according to the regular data transmission queue, and starts the high-definition recording mode of the video acquisition module.

[0089] Level 3 hazard is defined as a suspected hazard when a suspected hazard feature is matched. Suspected hazard features include vague outlines of dangerous equipment and unclear abnormal behavior. The control module triggers a low-priority alarm, stores only the suspicious information and real-time location locally, displays the information on the terminal screen, and requires the user to confirm whether to upgrade the alarm level.

[0090] In the above embodiments, the setting of the hazard level classification unit makes hazard identification and alarm triggering more targeted. Level 1 hazard targets emergency scenarios involving violent behavior, dangerous equipment, and clear personnel conflicts. These scenarios are often accompanied by high personal safety risks and require the highest priority response. Preemptive transmission ensures that alarm data is sent first, avoiding information delays caused by conventional transmission queues. Simultaneously dialing preset emergency contact numbers can further broaden the alarm reach and ensure rapid intervention by emergency forces. Level 2 hazard targets potential risk scenarios involving abnormal crowd gatherings. Although these scenarios do not directly involve danger, there are potential risks such as escalation of conflict and disorder. Medium-priority alarms do not consume too many communication resources and can retain on-site video evidence through high-definition recording mode, providing data support for subsequent control or event tracing. Level 3 hazard targets suspicious scenarios with ambiguous hazard characteristics. These scenarios have a high possibility of misjudgment. Low-priority alarms only store information locally and prompt users for confirmation. This can avoid interference from false alarms to the remote management platform and supplement the shortcomings of machine recognition through user manual confirmation. Users can determine whether to upgrade the alarm level through on-site observation.

[0091] In the above embodiments, the graded alarm mechanism can rationally allocate communication resources and response forces according to the urgency of the danger. Level 1 danger occupies the emergency communication channel and the nearest emergency personnel are dispatched. Level 2 danger is handled according to the conventional process and personnel are arranged for remote monitoring. Level 3 danger relies on user confirmation and reduces invalid responses. This differentiated handling method not only improves the efficiency of handling emergency dangers, but also reduces the resource consumption in non-emergency scenarios, making safety management more efficient and accurate, and is applicable to scenarios with different safety risk levels.

[0092] The public network intercom module supports multi-mode communication switching, specifically including:

[0093] When the terminal is in an area with good public network signal and the signal strength is ≥-70dBm, it will automatically use the 4G / 5G public network communication mode to establish a voice link and realize high-definition voice intercom.

[0094] When the terminal is in an area with weak public network signal, and -90dBm < signal strength < -70dBm, it automatically switches to narrowband IoT communication mode to ensure basic voice communication.

[0095] When the terminal is in an area without public network signal and the signal strength is ≤-90dBm, the terminal's built-in emergency intercom function is activated. It establishes a temporary communication network with similar terminals in the vicinity through short-range wireless communication to realize local intercom between terminals. Short-range wireless communication includes Bluetooth and ZigBee. After the public network signal is restored, the local intercom records are automatically uploaded to the remote management platform.

[0096] In the above embodiments, the public network intercom module accurately classifies the signal environment by signal strength threshold, corresponding to three modes: 4G / 5G, NB-IoT, and short-range wireless communication. Each mode optimizes communication performance for different signal conditions. The 4G / 5G mode provides high-definition voice intercom in areas with good signal, meeting the clear communication needs in daily scenarios. The NB-IoT mode ensures basic voice communication in areas with weak signal. This mode has the characteristics of low power consumption and wide coverage, and can stably transmit voice data in weak signal environments, avoiding communication loss due to signal interruption. The short-range wireless communication establishes a temporary network between terminals in areas without public network signal to ensure data integrity.

[0097] In the above embodiments, multi-mode switching does not require manual operation by the user. The terminal can automatically adapt to the optimal communication mode according to the signal environment, reducing the complexity of user operation. The terminal can automatically switch communication modes to ensure uninterrupted communication between the driver and the dispatch center or surrounding vehicles. The temporary network function of short-range wireless communication can also realize local collaborative communication between rescue personnel in disaster scenarios, providing key communication support for emergency rescue and improving the practicality of the terminal in extreme environments.

[0098] The positioning module employs multi-source positioning fusion technology, specifically including:

[0099] Prioritize obtaining high-precision location information of the terminal through satellite positioning, with a positioning accuracy error of ≤5 meters;

[0100] When satellite positioning signal is blocked, causing positioning failure, it automatically switches to base station positioning. By obtaining the signal parameters of at least 3 surrounding communication base stations, the terminal's location information is calculated, and the positioning accuracy error is ≤100 meters.

[0101] When the base station positioning signal is also weak, Wi-Fi positioning assistance is activated. By scanning the MAC addresses of surrounding Wi-Fi hotspots and matching them with the preset Wi-Fi location database, the approximate location area of ​​the terminal is determined, with a positioning accuracy error of ≤50 meters.

[0102] The control module performs weighted fusion of the results from satellite positioning, base station positioning, and Wi-Fi positioning, dynamically adjusts the weights according to the signal strength of different positioning methods, and outputs the optimal real-time location information.

[0103] In the above embodiments, the multi-source positioning fusion of the positioning module achieves accurate positioning in all scenarios. Satellite positioning is the preferred choice, and its high accuracy can meet the location requirements in most scenarios, such as law enforcement personnel positioning and vehicle dispatching. When satellite signals are blocked, it automatically switches to base station positioning, which calculates the location using signal parameters from more than three base stations. Although the accuracy of ≤100 meters is lower than that of satellite positioning, it can still ensure location output in environments without satellite signals, such as the positioning of security personnel in shopping malls and the tracking of vehicle locations in underground parking lots. Wi-Fi positioning serves as a supplement to base station positioning, determining the approximate area by matching the MAC address of Wi-Fi hotspots. The accuracy of ≤50 meters is between that of satellite positioning and base station positioning, and it is suitable for indoor scenarios where base station signals are also weak.

[0104] In the above embodiments, the control module dynamically adjusts the weight of each positioning method by signal strength, integrates the advantages of each positioning method, outputs the optimal location result, avoids the deviation of a single positioning method, covers the positioning needs of all scenarios including outdoor, indoor and underground, and can maintain continuous positioning without user intervention throughout the entire process to provide full-scenario guarantee for location services.

[0105] When sending alarm data, the remote communication module also simultaneously transmits historical video clips from 10 minutes before the alarm occurred to the moment the alarm was triggered;

[0106] The remote management platform also sends remote control commands to the terminal through the remote communication module. The remote control commands include at least starting real-time video transmission, adjusting the hazard identification sensitivity, and enabling public network intercom broadcast. After receiving the commands, the control module drives the corresponding functional modules to perform the operations and feeds back the execution results to the remote management platform.

[0107] In the above embodiments, the bidirectional data interaction between the remote communication module and the remote management platform enables the transmission of alarm data and feedback of results, significantly improving the remote control capabilities and application flexibility of the terminal. It can provide the remote management platform with background information before the occurrence of dangerous events, and the synchronous transmission of historical videos can reduce the operation of subsequent data retransmission, improving the efficiency of event tracing. The remote management platform can send real-time video feedback, adjust the sensitivity of hazard identification, and broadcast public network intercom commands to realize remote control of the terminal. After the control module executes the remote commands, it provides feedback on the results, allowing the remote management platform to confirm the execution status of the commands and avoid control failures caused by commands not being delivered or not being executed.

[0108] A smart positioning terminal integrating remote video recording and public speaking functions is characterized by further including a dynamic hazard assessment and optimization module. This module is used to optimize the accuracy and response efficiency of hazard identification in complex dynamic scenarios, solving the problem of misjudgment or missed judgment of hazards caused by environmental changes, target object movement speed, and video data noise. Specifically, it includes the following steps:

[0109] Real-time video data is acquired from the video acquisition module, real-time location information of the terminal is obtained from the positioning module, and environmental sound features from voice interaction data are obtained from the public network intercom module. Based on this data, a dynamic hazard assessment index is calculated to quantify the degree of danger in the current scene. The formula is as follows:

[0110] DDEI=(w1×VFI+w2×MSI+w3×AFI) / (1+e^(-k×(STI-S0)))

[0111] in:

[0112] DDEI stands for Dynamic Hazard Assessment Index. It is dimensionless and ranges from 0 to 1. The higher the value, the higher the potential hazard level of the current scenario.

[0113] VFI is the video frame interference factor, dimensionless, calculated by analyzing the image sharpness, normalized motion velocity of the target object, and normalized brightness change rate of the video frame. The formula is VFI = w v ×(1-C)×(V / V max )+w l ×(L / L max ), where C is the image sharpness (0 to 1), V is the target object's speed (unit: m / s), V max The preset maximum motion speed is L (in m / s), and the brightness change rate is Cd / m². 2 / s), L max Preset maximum brightness change rate (unit: cd / m²) 2 / s), w v and wl Here are the video factor weight coefficients, and w v +w l =1;

[0114] MSI is the mobile speed influence factor, which is dimensionless. It is based on the comparison between the terminal mobile speed (unit: m / s) output by the positioning module and the preset speed threshold V0 (unit: m / s). The formula is MSI = min(|V_terminal - V0| / V0, 1), where V_terminal is the real-time mobile speed of the terminal.

[0115] AFI is the ambient sound feature factor, which is dimensionless. It is calculated by comparing the ambient sound intensity (in dB) in the voice interaction data extracted by the public network intercom module with the preset noise threshold N0 (in dB). The formula is AFI = min((N-N0) / N0, 1), where N is the real-time ambient sound intensity.

[0116] w1, w2, and w3 are weight coefficients, which are dimensionless and correspond to the influence weights of video frame interference factor, motion speed influence factor, and ambient sound feature factor, respectively. w1+w2+w3=1, and the specific values ​​are dynamically adjusted according to the scene type.

[0117] STI stands for Scene Time Index, which is dimensionless and based on the rate of change of timestamps of video frames (in seconds). -1 The time synchronization data between the positioning module and the time synchronization data is calculated using the formula STI = ΔT / T0, where ΔT is the time difference between adjacent frames (unit: s) and T0 is the preset time base (unit: s).

[0118] S0 is the baseline value of the scene time index, which is dimensionless and is preset to 1.

[0119] k is an adjustment coefficient, dimensionless, used to control the sensitivity of the exponential function, and is preset to 2;

[0120] e is the base of the natural logarithm, approximately 2.718.

[0121] Based on the calculated DDEI, the dynamic hazard assessment optimization module performs the following operations:

[0122] When DDEI≥0.7, it is determined to be a high-risk scenario. The control module automatically increases the feature matching threshold of the hazard identification module, triggers the first-level hazard alarm first, and uploads the high-risk alarm data and real-time video stream to the remote management platform through the remote communication module.

[0123] When 0.3≤DDEI<0.7, it is determined to be a medium-risk scenario. The control module maintains the existing hazard identification sensitivity, triggers a level-two hazard alarm, and starts the high-resolution mode of the video acquisition module.

[0124] When DDEI < 0.3, it is determined to be a low-risk scenario. The control module lowers the feature matching threshold and only records potentially dangerous data to local storage, waiting for further confirmation from the user.

[0125] In the above embodiments, the dynamic hazard assessment optimization module optimizes the accuracy and response efficiency of hazard identification in complex dynamic scenarios by fusing multi-source data, solving the problem of misjudgment or missed judgment caused by environmental changes, target object movement speed, and video data noise. This module acquires real-time video data from the video acquisition module, extracts video frame images, and performs image sharpness analysis (based on image entropy calculation, with entropy values ​​ranging from 0 to 8, normalized to C = entropy / 8), target object movement speed (calculated using an optical flow algorithm for pixel displacement, in m / s, normalized to V / Vmax, where Vmax is preset to 10 m / s), and brightness change rate (based on inter-frame brightness difference, in cd / m²). 2 / s, normalized to L / Lmax, where Lmax is preset to 1000cd / m 2The video frame interference factor VFI is calculated using the formula VFI=wv×(1-C)×(V / Vmax)+wl×(L / Lmax), where the weights wv and wl (preset wv=0.6, wl=0.4, satisfying wv+wl=1) are dynamically adjusted based on scene complexity. The real-time terminal movement speed Vterminal (unit m / s, calculated via satellite positioning or base station positioning differential) is obtained from the positioning module and compared with the preset speed threshold V0 (preset to 2 m / s) to calculate the movement speed influence factor MSI=min(|Vterminal-V0| / V0,1). The ambient sound intensity N (unit dB, extracted via Fourier transform after microphone acquisition) is extracted from the public network intercom module and compared with the preset noise threshold N0 (preset to 60 dB) to calculate the ambient sound feature factor AFI=min((N-N0) / N0,1). The Scene Time Index (STI) is calculated using the video frame timestamp difference ΔT (in seconds, calculated based on a frame rate of 30fps, ΔT = 1 / 30) and a preset time base T0 (preset to 0.033s), with the formula STI = ΔT / T0. The Dynamic Risk Assessment Index (DDEI) is calculated using the formula DDEI = (w1 × VFI + w2 × MSI + w3 × AFI) / (1 + e^(-k × (STI - S0))), where the weights w1, w2, and w3 (preset w1 = 0.5, w2 = 0.3, w3 = 0.2, satisfying w1 + w2 + w3 = 1) are dynamically adjusted according to the scene type, k = 2, S0 = 1, and e ≈ 2.718. The DDEI ranges from 0 to 1, with higher values ​​indicating a higher level of danger. Based on the DDEI value, the module dynamically adjusts the feature matching threshold of the hazard identification module (preset to 0.8, ranging from 0.6 to 0.9). In high-risk scenarios (DDEI≥0.7), the threshold is increased to 0.9 to trigger a level 1 alarm. In medium-risk scenarios (0.3≤DDEI<0.7), the threshold is kept at 0.8 to trigger a level 2 alarm. In low-risk scenarios (DDEI<0.3), the threshold is decreased to 0.6 for local recording only, thus optimizing identification sensitivity and resource allocation.

[0126] To better illustrate this embodiment, the following specific implementation examples are provided:

[0127] In a campus security scenario, the dynamic hazard assessment and optimization module acquires a 1080p real-time video stream from the video acquisition module, analyzes the frame image clarity (entropy value 6.4, C = 6.4 / 8 = 0.8), and detects the target object (person) movement speed V = 3 m / s (V / Vmax = 3 / 10 = 0.3) and brightness change rate L = 200 cd / m² using an optical flow algorithm. 2 / s(L / Lmax=200 / 1000=0.2), calculate VFI=0.6×(1-0.8)×0.3+0.4×0.2=0.036+0.08=0.116. The positioning module obtains the terminal's moving speed Vterminal=1.5m / s through satellite positioning, compares it with V0=2m / s, and MSI=min(|1.5-2| / 2,1)=0.25. The public network intercom module extracts the ambient sound intensity N=80dB, compares it with N0=60dB, and AFI=min((80-60) / 60,1)=0.333. The video frame timestamp difference ΔT=0.033s, T0=0.033s, STI=0.033 / 0.033=1. Using weights w1=0.5, w2=0.3, w3=0.2, k=2, and S0=1, the DDEI is calculated as (0.5×0.116+0.3×0.25+0.2×0.333) / (1+e^(-2×(1-1)))=0.1996 / 2=0.0998≈0.100. DDEI < 0.3, indicating a low-risk scenario, the control module lowers the feature matching threshold to 0.6, recording only potentially dangerous data (such as suspected fast-moving objects) to the SD card in storage slot 6. The display screen 3 prompts the user to confirm whether to escalate the alarm. In another scenario (such as nighttime patrol), if V=8m / s, N=90dB, and DDEI=0.75 are detected, indicating a high-risk scenario, the module raises the threshold to 0.9, triggering a level one alarm. The remote communication module uploads alarm data and real-time video stream to the management platform via the 5G link, simultaneously triggering an audible and visual alarm.

[0128] In summary, the Dynamic Hazard Assessment Optimization module, by integrating video, location, and ambient sound data, accurately quantifies the hazard level of a scene and dynamically adjusts the hazard identification sensitivity, effectively reducing the risk of misjudgment or missed judgment caused by environmental noise, rapid object movement, or changes in lighting. The exponential adjustment mechanism of the DDEI formula ensures rapid response in high-risk scenarios and saves computing resources in low-risk scenarios, making it suitable for dynamic scenarios such as campus security and nighttime patrols. The module's automated processing reduces user intervention and improves emergency response efficiency, while the combination of local storage and remote uploads ensures data integrity and traceability.

[0129] A smart positioning terminal integrating remote video recording and public speaking functions is characterized by further including an adaptive power management module. This module dynamically adjusts power distribution and power consumption mode according to the terminal's operating status, environmental conditions, and user interaction needs to extend the terminal's battery life and ensure stable operation of critical functions in low-power scenarios. Specifically, it includes the following steps:

[0130] The adaptive power management module monitors the battery level of the terminal body, the operating status of the video acquisition module, the communication strength of the public network intercom module, the positioning frequency of the positioning module, and the computing load of the hazard identification module in real time, and dynamically classifies the terminal's working state into high load mode, standard mode, and low power mode.

[0131] In high-load mode, the video acquisition module operates at the highest resolution, the public network intercom module maintains a continuous voice link, the positioning module updates location information at the highest frequency, and the hazard identification module enables all feature recognition algorithms; the adaptive power management module prioritizes allocating power resources to the video acquisition module and the hazard identification module to ensure the real-time performance of hazard identification and the integrity of video data, while limiting the background operation of non-essential functions.

[0132] In standard mode, the video acquisition module operates at medium resolution, the public network intercom module establishes voice links as needed, the positioning module updates location information at a medium frequency, and the hazard identification module only enables the core feature recognition algorithm; the adaptive power management module balances the power distribution of each functional module and optimizes power consumption by reducing the brightness of the display screen and reducing the frequency of non-emergency data transmission of the remote communication module.

[0133] In low-power mode, the video acquisition module pauses real-time recording and only starts when the hazard identification module triggers an alarm; the public network intercom module switches to narrowband communication mode or short-range wireless communication; the positioning module reduces the update frequency to the lowest threshold; and the hazard identification module only analyzes key hazard features. The adaptive power management module shuts down non-core hardware components, such as the display backlight and secondary sensors, and sends a low-battery warning to the remote management platform through the remote communication module when the battery level is below the power threshold.

[0134] The adaptive power management module also dynamically adjusts the charging strategy based on ambient temperature and terminal usage time. When the ambient temperature is higher than the first temperature threshold or lower than the second temperature threshold, it reduces the charging current to protect battery life. When it detects that the user has not operated the terminal for a long time, it automatically enters standby mode, keeping only the positioning module and remote communication module running at the lowest power consumption.

[0135] In the above embodiments, the adaptive power management module dynamically adjusts power distribution and power consumption modes by monitoring the terminal's operating status and environmental conditions in real time, extending battery life and ensuring stable operation of critical functions in low-power scenarios. The module obtains battery power (in mAh, accuracy ±1%) from the battery management chip, determines operating status through the operating parameters of the video acquisition module (e.g., resolution 1080p or 720p, frame rate 30fps), obtains communication strength (in dBm, measured by the baseband chip RSSI value, range -110dBm to -50dBm) from the public network intercom module, obtains the positioning frequency (in Hz, range 0.1Hz to 1Hz) from the positioning module, and obtains the computing load (by CPU / GPU utilization, in %, range 0 to 100%) from the hazard identification module. Based on these parameters, the module classifies the terminal's operating state into three modes: high-load mode (battery > 50%, video resolution 1080p, continuous intercom connection, positioning frequency 1Hz, computational load > 80%), standard mode (battery 30%-50%, video resolution 720p, on-demand intercom connection, positioning frequency 0.5Hz, computational load 40%-80%), and low-power mode (battery < 30%, video recording paused, narrowband or short-range intercom communication, positioning frequency 0.1Hz, computational load < 40%). In high-load mode, the module prioritizes power allocation to the video acquisition module (power consumption approximately 2W) and the hazard identification module (power consumption approximately 1.5W), and closes unnecessary background processes (such as Wi-Fi scanning, power consumption approximately 0.3W) via the I2C interface. In standard mode, the display brightness is reduced (from 300cd / m²) via a PWM signal. 2 Reduced to 150 cd / m 2 The system reduces the frequency of non-emergency data transmission from the remote communication module (from 1 time / s to 0.2 times / s, saving approximately 0.2W). In low-power mode, the display backlight is turned off (saving approximately 0.7W) and secondary sensors (such as the gyroscope, saving approximately 0.1W). The video acquisition module only starts when an alarm is triggered (power consumption approximately 0.5W / alarm). The module obtains the ambient temperature (in °C, range -20°C to 60°C) through a temperature sensor. When the temperature is higher than the first threshold (45°C) or lower than the second threshold (0°C), the charging current is reduced (from 1A to 0.5A) through the charging management IC. If the system detects that the user has not operated for more than 5 minutes (counted by the interrupt signal of physical button 4), it enters standby mode, retaining only the positioning module (power consumption approximately 0.05W) and the remote communication module (power consumption approximately 0.1W). When the low battery threshold (preset 10%) is triggered, a warning is sent through the remote communication module (data packet size approximately 1KB).

[0136] To better illustrate this embodiment, the following specific implementation examples are provided:

[0137] In an outdoor operation scenario, the adaptive power management module detected a battery level of 60% (read from the battery management chip, accuracy ±1%). The video acquisition module operated at 1080p resolution (power consumption 2W), the public network intercom module maintained a continuous voice link via 5G (signal strength -65dBm, power consumption 1W), the positioning module updated at a frequency of 1Hz (power consumption 0.2W), and the hazard identification module's CPU utilization was 85% (power consumption 1.5W), indicating a high-load mode. The module disabled Wi-Fi scanning via the I2C interface (saving 0.3W) and prioritized power supply to the video and hazard identification modules to ensure stable operation of real-time video recording and hazard analysis. The ambient temperature was 30℃ (read from a temperature sensor), and the charging current remained at 1A. After 2 hours of operation, the battery level dropped to 40%. The module switched to standard mode, the video resolution decreased to 720p (power consumption 1.5W), the intercom connection was on demand (power consumption decreased to 0.6W), the positioning frequency decreased to 0.5Hz (power consumption 0.1W), and the display brightness decreased to 150cd / m². 2 (Saving 0.5W), CPU utilization drops to 50% (power consumption 1W). When the battery level drops further to 15%, it switches to low-power mode, video capture pauses (power consumption 0W), intercom switches to ZigBee communication (power consumption 0.3W), positioning frequency drops to 0.1Hz (power consumption 0.05W), hazard identification only analyzes violent behavior characteristics (power consumption 0.5W), and the display backlight is turned off (saving 0.7W). When the temperature rises to 50℃, the charging current drops to 0.5A to protect the battery. After detecting 5 minutes of no button operation, it enters standby mode, retaining only the positioning and communication modules (total power consumption 0.15W). When the battery level drops to 10%, a low-battery warning is sent to the remote management platform via 5G (data packet 1KB). This dynamic adjustment ensures that the terminal maintains critical functions during 12 hours of operation, extending battery life by approximately 30%.

[0138] The adaptive power management module optimizes power distribution across modules through fine-grained monitoring and dynamic adjustment, significantly extending the terminal's battery life. Especially in low-battery scenarios, it prioritizes video capture and hazard identification functions, ensuring uninterrupted safety monitoring. Dynamic charging and standby strategies based on ambient temperature and user operation effectively protect battery life and reduce the risk of damage from overheating or low temperatures. Automated module management reduces user intervention, improving device reliability and efficiency.

[0139] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. An intelligent positioning terminal integrating remote video recording and public network intercom functions, characterized in that: It includes a terminal body and a group of functional modules integrated inside the terminal body. The group of functional modules includes a video acquisition module, a public network intercom module, a positioning module, a hazard identification module, a control module, and a remote communication module. The video acquisition module is used to acquire real-time video data around the terminal and transmit the real-time video data to the control module; The public network intercom module is used to establish a public network voice communication link between the terminal and external devices, realize two-way voice interaction, and synchronously transmit voice interaction data to the control module; The positioning module is used to obtain the real-time location information of the terminal and transmit the real-time location information to the control module; The hazard identification module is used to perform potential hazard analysis and alarm triggering; The control module is used to receive and process data transmitted from the video acquisition module, public network intercom module, positioning module and hazard identification module, and establish a data interaction channel with the remote management platform through the remote communication module to realize remote data transmission and command reception.

2. The intelligent positioning terminal integrating remote video recording and public network intercom functions as described in claim 1, characterized in that, The hazard identification module is electrically connected to the control module, video acquisition module, and positioning module, respectively. The specific process by which the hazard identification module performs potential hazard analysis and alarm triggering includes: The system acquires real-time video data collected by the video acquisition module, performs frame extraction on the real-time video data, and obtains continuous video frame images. Feature recognition is performed on each video frame to extract features of target objects, human behavior, and environmental scene. The extracted features are matched with feature data in a preset danger feature database, which includes at least violent behavior features, dangerous equipment features, and abnormal crowd gathering features. When the matching result meets the preset danger triggering condition, the potential danger is identified, and the real-time location information output by the positioning module is obtained. The identification results of potential hazards are integrated with real-time location information to form alarm data. The remote communication module is triggered by the control module to send the alarm data to the remote management platform and simultaneously trigger the local audible and visual alarm of the terminal.

3. The intelligent positioning terminal integrating remote video recording and public network intercom functions as described in claim 2, characterized in that, When the hazard identification module performs feature recognition on video frame images, it employs a multi-dimensional feature fusion algorithm, specifically including: For the target object features, the object contour is extracted using an edge detection algorithm, and the object color distribution is analyzed by combining a color histogram to generate an object feature vector. For human behavioral features, the human joint coordinates are obtained through the skeletal keypoint detection algorithm, the change in joint coordinates in adjacent frames is calculated, the type of human action is determined, and a behavioral feature vector is generated. Based on the environmental scene features, the scene regions are divided using a semantic segmentation algorithm, dangerous area markers in the scene are identified, and scene feature vectors are generated. The object feature vector, behavior feature vector, and scene feature vector are weighted and fused to obtain a comprehensive feature vector. Then, the similarity is calculated with the feature vector in the preset danger feature library. If the similarity is greater than the preset threshold, it is determined that the danger feature matching condition is met.

4. The intelligent positioning terminal integrating remote video recording and public network intercom functions as described in claim 2, characterized in that, The hazard identification module also includes a hazard level classification unit, which is used to classify potential hazards into level one, level two, and level three hazards based on the degree of matching of hazard features and the urgency of the scenario. The first-level danger is defined as the presence of violent behavior or dangerous equipment characteristics, and the presence of clear personnel conflict actions in the video frame, which is determined to be an emergency danger. At this time, the control module triggers the highest priority alarm, and the remote communication module uses preemptive transmission to send the alarm data to the remote management platform and simultaneously dials the preset emergency contact number. The level 2 hazard is determined by matching the characteristics of abnormal crowd gathering, which is identified as a potential hazard. The abnormal crowd gathering characteristics are that the number of people gathered is ≥10 and the duration is ≥5 minutes. The control module triggers a medium-priority alarm, the remote communication module sends alarm data according to the regular data transmission queue, and starts the high-definition recording mode of the video acquisition module. The third level of danger is determined by matching suspected danger features. The suspected danger features include vague outlines of dangerous equipment and unclear abnormal behavior. The control module triggers a low-priority alarm, stores the suspicious information and real-time location locally, displays the information on the terminal screen, and asks the user to confirm whether to upgrade the alarm level.

5. The intelligent positioning terminal integrating remote video recording and public network intercom functions as described in claim 1, characterized in that, The public network intercom module supports multi-mode communication switching, specifically including: When the terminal is in an area with good public network signal and the signal strength is ≥-70dBm, it will automatically establish a voice link using the 4G / 5G public network communication mode. When the terminal is in an area with weak public network signal, and -90dBm < signal strength < -70dBm, it automatically switches to narrowband IoT communication mode. When the terminal is in an area without public network signal and the signal strength is ≤-90dBm, the terminal’s built-in emergency intercom function is activated. It establishes a temporary communication network with similar terminals in the vicinity through short-range wireless communication, and automatically uploads the local intercom records to the remote management platform after the public network signal is restored.

6. The intelligent positioning terminal integrating remote video recording and public network intercom functions as described in claim 1, characterized in that, The positioning module employs multi-source positioning fusion technology, specifically including: Prioritize obtaining high-precision location information of the terminal through satellite positioning; When satellite positioning signal is blocked, causing positioning failure, it automatically switches to base station positioning. By obtaining the signal parameters of at least 3 surrounding communication base stations, the terminal's location information is calculated. When the base station positioning signal is also weak, Wi-Fi positioning assistance is activated. By scanning the MAC addresses of surrounding Wi-Fi hotspots and matching them with a preset Wi-Fi location database, the approximate location area of ​​the terminal is determined. The control module performs weighted fusion of the results from satellite positioning, base station positioning, and Wi-Fi positioning, dynamically adjusts the weights according to the signal strength of different positioning methods, and outputs the optimal real-time location information.

7. The intelligent positioning terminal integrating remote video recording and public network intercom functions as described in claim 1, characterized in that, When sending alarm data, the remote communication module also simultaneously transmits historical video clips from 10 minutes before the alarm occurred to the moment the alarm was triggered. The remote management platform also sends remote control commands to the terminal through the remote communication module. The remote control commands include at least starting real-time video transmission, adjusting the hazard identification sensitivity, and enabling public network intercom broadcast. After receiving the commands, the control module drives the corresponding functional modules to perform the operations and feeds back the execution results to the remote management platform.

8. The intelligent positioning terminal integrating remote video recording and public network intercom functions as described in claim 1, characterized in that, The front of the terminal body is equipped with a camera that interacts with the video acquisition module, the back of the terminal body is equipped with a display screen, the sides of the terminal body are equipped with physical buttons that interact with the control module, and the sides of the terminal body are respectively equipped with a charging port and a memory card slot.

9. The intelligent positioning terminal integrating remote video recording and public network intercom functions as described in claim 1, characterized in that, It also includes a dynamic hazard assessment and optimization module, which is used to optimize the accuracy and response efficiency of hazard identification in complex dynamic scenes, and solve the problem of hazard misjudgment or missed judgment caused by environmental changes, target object movement speed and video data noise. Specifically, it includes the following steps: The system acquires real-time video data from the video acquisition module, obtains the terminal's real-time location information from the positioning module, and acquires environmental sound features from the voice interaction data from the public network intercom module. Based on real-time video data, real-time location information, and ambient sound characteristics, a dynamic hazard assessment index is calculated to quantify the degree of danger in the current scene. The formula is as follows: DDEI=(w1×VFI+w2×MSI+w3×AFI) / (1+e^(-k×(STI-S0))) Wherein, DDEI is the Dynamic Hazard Assessment Index, and a higher value indicates a higher potential hazard level in the current scene; VFI is the Video Frame Interference Factor, calculated by analyzing the sharpness of the video frame image, the normalized motion speed of the target object, and the normalized brightness change rate, with the formula VFI = w v ×(1-C)×(V / V max )+w l ×(L / L max ), where C is the image sharpness, V is the target object's speed, and V max The preset maximum motion speed is given by L, which represents the rate of change of brightness (unit: cd / m). 2 / s), L max To preset the maximum rate of change in brightness, w v and w l Here are the video factor weight coefficients, and w v +w l =1; MSI is the moving speed influence factor, which is obtained by comparing the terminal moving speed output by the positioning module with the preset speed threshold V0. The formula is MSI = min(|V_terminal-V0| / V0,1), where V_terminal is the real-time moving speed of the terminal; AFI is the ambient sound feature factor, which is obtained by comparing the ambient sound intensity in the voice interaction data extracted by the public network intercom module with the preset noise threshold N0. The formula is AFI = min((N-N0) / N0,1), where N is the real-time ambient sound intensity; w1, w2, and w3 are weight coefficients, which correspond to the influence weights of the video frame interference factor, moving speed influence factor, and ambient sound feature factor, respectively, and w1 + w2 + w3 = 1; STI is the scene time index, which is calculated based on the timestamp change rate of the video frame and the time synchronization data of the positioning module. The formula is STI = ΔT / T0, where ΔT is the time difference between adjacent frames, T0 is the preset time base; S0 is the scene time index base value; k is the adjustment coefficient, used to control the sensitivity of the exponential function; e is the base of the natural logarithm; Based on the calculated DDEI, the dynamic hazard assessment optimization module performs the following operations: When DDEI≥0.7, it is determined to be a high-risk scenario. The control module automatically increases the feature matching threshold of the hazard identification module, triggers the first-level hazard alarm first, and uploads the high-risk alarm data and real-time video stream to the remote management platform through the remote communication module. When 0.3≤DDEI<0.7, it is determined to be a medium-risk scenario. The control module maintains the existing hazard identification sensitivity, triggers a level-two hazard alarm, and starts the high-resolution mode of the video acquisition module. When DDEI < 0.3, it is determined to be a low-risk scenario. The control module lowers the feature matching threshold and only records potentially dangerous data to local storage, waiting for further confirmation from the user.

10. The intelligent positioning terminal integrating remote video recording and public network intercom functions as described in claim 1, characterized in that, It also includes an adaptive power management module, which dynamically adjusts power distribution and power consumption mode according to the terminal's working status, environmental conditions, and user interaction needs to extend the terminal's battery life and ensure the stable operation of critical functions in low-power scenarios. Specifically, it includes the following steps: The adaptive power management module monitors the battery level of the terminal body, the operating status of the video acquisition module, the communication strength of the public network intercom module, the positioning frequency of the positioning module, and the computing load of the hazard identification module in real time, and dynamically classifies the terminal's working state into high load mode, standard mode, and low power mode. In high-load mode, the video acquisition module operates at the highest resolution, the public network intercom module maintains a continuous voice link, the positioning module updates location information at the highest frequency, and the hazard identification module enables all feature recognition algorithms; the adaptive power management module prioritizes allocating power resources to the video acquisition module and the hazard identification module to ensure the real-time performance of hazard identification and the integrity of video data, while limiting the background operation of non-essential functions. In standard mode, the video acquisition module operates at medium resolution, the public network intercom module establishes voice links as needed, the positioning module updates location information at a medium frequency, and the hazard identification module only enables the core feature recognition algorithm; the adaptive power management module balances the power distribution of each functional module and optimizes power consumption by reducing the brightness of the display screen and reducing the frequency of non-emergency data transmission of the remote communication module. In low-power mode, the video acquisition module pauses real-time recording and only starts when the hazard identification module triggers an alarm; the public network intercom module switches to narrowband communication mode or short-range wireless communication; the positioning module reduces the update frequency to the lowest threshold; and the hazard identification module only analyzes key hazard features. The adaptive power management module shuts down non-core hardware components, such as the display backlight and secondary sensors, and sends a low-battery warning to the remote management platform through the remote communication module when the battery level is below the power threshold. The adaptive power management module also dynamically adjusts the charging strategy based on ambient temperature and terminal usage time. When the ambient temperature is higher than the first temperature threshold or lower than the second temperature threshold, it reduces the charging current to protect battery life. When it detects that the user has not operated the terminal for a long time, it automatically enters standby mode, keeping only the positioning module and remote communication module running at the lowest power consumption.

Images