Public area face collector, system and method

Abstract

The invention relates to the technical field of security and protection monitoring, in particular to a public area face collector, system and method. Comprising a multispectral imaging module, a dynamic light supplementing module, a living body detection module, a load adjusting module, at least two edge computing layers, a privacy processing module, a block chain storage layer, an encryption transmission unit, a dynamic secret key updating module and a protection assembly, the encryption transmission unit is connected with the multispectral imaging module, and the living body detection module is connected with the encryption transmission unit. The load adjusting module is connected with the living body detection module, and the two edge calculation layers are both connected with the load adjusting module, so that the problems that in the prior art, acquisition equipment only depends on a single spectral band to acquire images, so that the quality of obtained face images is poor; and in data transmission and storage links, a security guarantee mechanism is lacked, face data is faced with a serious leakage risk, and user privacy security is dangerous.

Description

Technical Field

[0001] This invention relates to the field of security monitoring technology, and in particular to a face capture device, system and method for public areas. Background Technology

[0002] In today's digital age, facial recognition technology in public areas is developing rapidly, with extremely wide applications. In security monitoring scenarios, facial recognition technology plays an irreplaceable and crucial role. Traditional security monitoring methods often rely on manual review of surveillance footage, which is not only inefficient but also prone to oversights. The introduction of facial recognition technology has completely changed this situation. Using advanced facial capture equipment, it can instantly capture the facial information of people within the monitored area and quickly compare it with a pre-entered database of suspicious individuals. Once a matching suspicious person is detected, the system immediately issues a real-time alert and accurately tracks their movements, providing strong support for security personnel to take timely measures, thus building an impenetrable defense for public safety.

[0003] However, existing acquisition equipment relies on a single spectral band to acquire images, resulting in poor quality of the acquired facial images; moreover, there is a lack of security mechanisms in the data transmission and storage process, which poses a serious risk of facial data leakage and endangers user privacy. Summary of the Invention

[0004] The purpose of this invention is to provide a public area face capture device, system, and method, aiming to solve the technical problems of existing capture devices relying on only a single spectral band to capture images, resulting in poor quality of the acquired face images; and the lack of security mechanisms in the data transmission and storage stages, which poses a serious risk of face data leakage and endangers user privacy.

[0005] To achieve the above objectives, the present invention employs a public area face capture device, comprising a multispectral imaging module, a dynamic illumination module, a liveness detection module, a load adjustment module, at least two edge computing layers, a privacy processing module, a blockchain storage layer, an encrypted transmission unit, a dynamic key update module, and protection components. The encrypted transmission unit is connected to the multispectral imaging module, the liveness detection module is connected to the encrypted transmission unit, the load adjustment module is connected to the liveness detection module, both edge computing layers are connected to the load adjustment module, the privacy processing module is connected to both edge computing layers, the blockchain storage layer is connected to the privacy processing module, all protection components are connected to the multispectral imaging module, the dynamic illumination module is also connected to the multispectral imaging module, and the dynamic key update module is connected to the encrypted transmission unit.

[0006] The multispectral imaging module is used to acquire facial image data in multiple spectral bands;

[0007] The encrypted transmission unit is used to encrypt and transmit the data collected by the multispectral imaging module to the liveness detection module;

[0008] The liveness detection module is used to determine the liveness of the acquired face images, so that only live data is allowed to enter the edge computing layer.

[0009] The load adjustment module is used to dynamically adjust the data flow and computing resource allocation based on the processing results of the liveness detection module and the real-time load of the two edge computing layers.

[0010] The edge computing layer deploys a lightweight model for real-time extraction of facial feature vectors;

[0011] The privacy processing module is used to encrypt and desensitize the extracted facial feature values ​​and erase the original image data;

[0012] The blockchain storage layer adopts the Hyperledger Fabric consortium blockchain architecture, which encrypts the facial feature vector and stores it in a distributed ledger. Each node saves a complete copy of the data, ensuring that a single point of failure does not affect the system availability.

[0013] The dynamic fill light module uses a spectral adaptive controller, which predicts illumination compensation parameters based on a convolutional neural network and outputs a PWM dimming signal with an adjustable duty cycle to the fill light for fill light.

[0014] The dynamic key update module is used to periodically and dynamically update the encryption key used by the encrypted transmission unit.

[0015] The encrypted transmission unit includes a data encryption module, a data decryption module, a high-efficiency transmission layer, a key manager, and a key generator. The high-efficiency transmission layer is connected to both the multispectral imaging module and the liveness detection module. The data encryption module and the data decryption module are respectively connected to the multispectral imaging module and the liveness detection module. The key manager is connected to both the data encryption module, the data decryption module, and the key generator.

[0016] The protective component includes a housing and a piezoelectric sensor array. The housing covers the multispectral imaging module, and the piezoelectric sensor array monitors the deformation of the housing in real time. When the deformation exceeds a threshold, the power is cut off and data destruction is initiated.

[0017] The present invention also provides a public area face capture system, including the public area face capture device as described above.

[0018] The public area face capture system also includes an energy management module, an intelligent analysis layer, and an anti-obstruction mechanism. The energy management module and the anti-obstruction mechanism are both connected to the multispectral imaging module. The intelligent analysis layer is connected to the blockchain storage layer. The intelligent analysis layer is based on the Spark big data framework to build a behavior analysis engine, which supports real-time trajectory tracking, frequency statistics, and abnormal behavior warnings.

[0019] The energy management module includes a dual-mode power supply interface and a dynamic power consumption adjustment module, both of which are connected to the multispectral imaging module.

[0020] The anti-occlusion mechanism includes a multi-sensor fusion detector and a hierarchical response mechanism. The multi-sensor fusion detector integrates millimeter-wave radar, infrared thermal imaging and image semantic analysis to identify three types of occlusion scenarios: rain and snow, paint stains and physical occlusion.

[0021] The hierarchical response mechanism is configured as follows:

[0022] The waterproof coating activates its electrostatic repulsion function when rain or snow is detected.

[0023] After identifying the contaminant, spray a fine mist of organic solvent.

[0024] When physically obstructed, a 110dB audible and visual alarm is triggered and location information is uploaded.

[0025] The public area face capture system also includes an environment perception module, which is connected to the dynamic lighting module. The environment perception module acquires and analyzes data in the environment in real time and triggers the dynamic lighting module to adjust the lighting parameters.

[0026] This invention also provides a method for facial recognition in public areas, applicable to the facial recognition system for public areas as described above.

[0027] The multispectral imaging module simultaneously acquires face images in at least two spectral bands;

[0028] The encrypted transmission unit dynamically generates a session key and uses the SM4 algorithm to encrypt the image data before transmitting it to the liveness detection module. The liveness detection module verifies liveness through micro-expression, skin texture, and blood flow information analysis, and only allows verified data to enter the load adjustment module.

[0029] The load adjustment module dynamically allocates data streams based on the real-time resource utilization of the edge computing layer, and each edge computing layer deploys a lightweight model to extract facial feature vectors and transmits them to the privacy processing module.

[0030] The privacy processing module uses homomorphic encryption to de-identify the feature vector and erases the original image. The blockchain storage layer writes the de-identified data into the distributed ledger through the PBFT consensus algorithm, and each node saves a complete copy.

[0031] The intelligent analysis layer tracks trajectories in real time, counts frequencies, and issues warnings for abnormal behaviors based on the Spark framework. The anti-obstruction mechanism triggers an audible and visual alarm, switches to backup power, and uploads location information when it detects physical obstruction.

[0032] The dynamic fill light module outputs a PWM signal based on environmental perception data to drive the fill light for spectral adaptive compensation.

[0033] The protective component monitors equipment deformation in real time and cuts off power and destroys data in case of abnormality.

[0034] This invention discloses a public area face capture device, system, and method. In practical use, firstly, addressing the problem of poor image quality under complex lighting conditions caused by traditional single-spectral acquisition, this solution introduces a multispectral imaging module. This module can acquire face image data across multiple spectral bands, significantly improving image clarity and accuracy, and providing a high-quality data foundation for subsequent liveness detection and feature extraction. Secondly, to solve security issues during data transmission and storage, this invention includes an encrypted transmission unit, a privacy processing module, and a blockchain storage layer. The encrypted transmission unit ensures data security during transmission. The privacy processing module encrypts and desensitizes facial feature values ​​and erases the original image data, further protecting user privacy. The blockchain storage layer adopts a Hyperledger Fabric consortium blockchain architecture, storing the encrypted facial feature vectors in a distributed ledger, ensuring data immutability and traceability, and effectively preventing data leakage. Finally, to address the issues of device load balancing and computing resource allocation, this solution incorporates a load adjustment module. This module dynamically adjusts data flow and computing resource allocation based on the processing results of the liveness detection module and the real-time load of the two edge computing layers. Simultaneously, the edge computing layer deploys a lightweight model to achieve real-time extraction of facial feature vectors, improving the overall processing efficiency and resource utilization of the system. This approach solves the technical problems of existing technologies where acquisition devices rely solely on a single spectral band to acquire images, resulting in poor quality facial images; and the lack of security mechanisms in data transmission and storage, leading to serious risks of facial data leakage and jeopardizing user privacy. Attached Figure Description

[0035] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0036] Figure 1 This is a schematic diagram of the public area face capture device of the present invention.

[0037] Figure 2 This is a schematic diagram of the public area face capture system of the present invention.

[0038] Figure 3 This is a flowchart of the steps of the public area face collection method of the present invention.

[0039] 101-Multispectral Imaging Module, 102-Dynamic Illumination Module, 103-Liveness Detection Module, 104-Load Adjustment Module, 105-Edge Computing Layer, 106-Privacy Processing Module, 107-Blockchain Storage Layer, 108-Encrypted Transmission Unit, 109-Dynamic Key Update Module, 110-Protection Component, 111-Self-Optimization Module, 112-Status Indication Module, 113-Face Tracking Module, 114-Data Encryption Module, 115-Data Decryption Module, 116-High-Efficiency Transmission Layer, 117-Key Manager, 118-Key Generator, 119-Shell, 120-Piezoelectric Sensor Array, 121-Tamper-proof Electronic Lock, 201-Energy Management Module, 202-Intelligent Analysis Layer, 203-Anti-Obstruction Mechanism, 204-Environmental Sensing Module, 205-Dual-Mode Power Supply Interface, 206-Dynamic Power Consumption Adjustment Module, 207-Multi-Sensor Fusion Detector, 208-Graded Response Mechanism. Detailed Implementation

[0040] The embodiments of the present invention are described in detail below. Examples of the embodiments are shown in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, but should not be construed as limiting the present invention.

[0041] Please see Figure 1 , Figure 1 This is a schematic diagram of the public area face capture device of the present invention.

[0042] This invention provides a public area face capture device, comprising a multispectral imaging module 101, a dynamic illumination module 102, a liveness detection module 103, a load adjustment module 104, at least two edge computing layers 105, a privacy processing module 106, a blockchain storage layer 107, an encrypted transmission unit 108, a dynamic key update module 109, a protection component 110, a self-optimization module 111, a status indication module 112, and a face tracking module 113. The encrypted transmission unit 108 includes a data encryption module 114, a data decryption module 115, a high-efficiency transmission layer 116, a key manager 117, and a key generator 118. The protection component 110 includes a housing 119, a piezoelectric sensor array 120, and an tamper-proof electronic lock 121. The aforementioned technical solution solves the technical problems of existing capture devices relying only on a single spectral band to capture images, resulting in poor quality of the acquired face images; and the lack of security mechanisms in the data transmission and storage stages, leading to serious risks of face data leakage and jeopardizing user privacy.

[0043] In this specific embodiment, the multispectral imaging module 101 is used to acquire face image data in multiple spectral bands;

[0044] The encrypted transmission unit 108 is used to encrypt and transmit the data collected by the multispectral imaging module 101 to the liveness detection module 103;

[0045] The liveness detection module 103 is used to determine the liveness of the acquired face images, so that only liveness data is allowed to enter the edge computing layer 105.

[0046] The load adjustment module 104 is used to dynamically adjust the data flow and computing resource allocation based on the processing results of the liveness detection module 103 and the real-time load of the two edge computing layers 105.

[0047] The edge computing layer 105 deploys a lightweight model for real-time extraction of facial feature vectors;

[0048] The privacy processing module 106 is used to encrypt and desensitize the extracted facial feature values ​​and erase the original image data;

[0049] The blockchain storage layer 107 adopts the Hyperledger Fabric consortium blockchain architecture, which encrypts the facial feature vector and stores it in a distributed ledger. Each node saves a complete data copy to ensure that a single point of failure does not affect the system availability.

[0050] The dynamic fill light module 102 adopts a spectrum adaptive controller, predicts the illumination compensation parameters based on a convolutional neural network, and outputs a PWM dimming signal with an adjustable duty cycle to the fill light for fill light.

[0051] The dynamic key update module 109 is used to periodically and dynamically update the encryption key used by the encrypted transmission unit 108.

[0052] The encrypted transmission unit 108 is connected to the multispectral imaging module 101, the liveness detection module 103 is connected to the encrypted transmission unit 108, the load adjustment module 104 is connected to the liveness detection module 103, both edge computing layers 105 are connected to the load adjustment module 104, the privacy processing module 106 is connected to both edge computing layers 105, the blockchain storage layer 107 is connected to the privacy processing module 106, the protection components 110 are all connected to the multispectral imaging module 101, the dynamic supplementary lighting module 102 is also connected to the multispectral imaging module 101, and the dynamic key update module 109 is connected to the encrypted transmission unit 108. In practical use, the multispectral imaging module 101 initiates data acquisition to obtain multispectral image information of the target, the dynamic supplementary lighting module 102 adjusts the light to assist imaging, and the protection components 110 ensure the stability of the imaging module. The acquired data is encrypted by the encrypted transmission unit 108, and the dynamic key update module 109 synchronously updates the key to prevent data leakage. After the liveness detection module 103 verifies that the object is a real living entity, the load adjustment module 104 distributes the data to the two edge computing layers 105 for initial processing according to the system load. Then, the privacy processing module 106 performs privacy protection operations such as data anonymization on the data. Finally, the blockchain storage layer 107 securely stores the processed data on the blockchain.

[0053] Secondly, the high-efficiency transmission layer 116 is connected to both the multispectral imaging module 101 and the liveness detection module 103. The data encryption module 114 and the data decryption module 115 are respectively connected to the multispectral imaging module 101 and the liveness detection module 103. The key manager 117 is connected to both the data encryption module 114, the data decryption module 115, and the key generator 118. When the data encryption module 114 detects that data is ready to be transmitted, it immediately requests an encryption key from the key manager 117. The key manager 117 obtains a new key from the key generator 118 and distributes it to the data encryption module 114. The data encryption module 114 uses this key to encrypt the aggregated data. The encrypted data is securely sent to the liveness detection module 103 via the high-efficiency transmission layer 116. When the key manager 117 provides the corresponding decryption key, the data decryption module 115 decrypts the received encrypted data to restore the original data for subsequent analysis and processing.

[0054] Meanwhile, the outer shell 119 is placed over the multispectral imaging module 101. The outer shell 119 is made of carbon fiber reinforced polycarbonate material. The piezoelectric sensor array 120 monitors the deformation of the outer shell in real time. When the deformation exceeds the threshold, the power is cut off and data destruction is initiated.

[0055] Furthermore, the self-optimization module 111 is connected to the dynamic fill light module 102. The self-optimization module 111 can use advanced optimization algorithms such as gradient descent (e.g., Adam optimization algorithm), Newton's method (e.g., quasi-Newton method), parameter adaptive adjustment strategy, and chaotic mapping strategy to continuously optimize the fill light model of the dynamic fill light module 102. Appropriately, the dynamic fill light module 102 can accurately calculate the most suitable fill light parameters, including fill light intensity, angle, and spectral distribution, based on the characteristics of the target object, ambient lighting conditions, and shooting requirements in different scenarios.

[0056] Furthermore, the status indication module 112 is connected to the multispectral imaging module 101. The status indication module 112 acquires real-time operating status information of the multispectral imaging module 101, such as operating temperature, signal transmission stability, and response status of each spectral channel. Once the multispectral imaging module 101 malfunctions, the status indication module 112 will immediately issue an alarm signal to inform the operator in a visually intuitive way.

[0057] By setting the anti-tamper electronic lock 121, flame-retardant gas is injected into the equipment cavity to fill it in case of illegal disassembly.

[0058] Finally, the face tracking module 113 is connected to the multispectral imaging module 101. By setting the face tracking module 113, when a face is detected entering the acquisition range, the face tracking module 113 can quickly lock the face target and continuously track its movement trajectory.

[0059] In practical use, the public area face capture device of this invention addresses the problem of poor image quality under complex lighting conditions caused by traditional single-spectral acquisition. Firstly, it introduces the multispectral imaging module 101, which can acquire face image data across multiple spectral bands, significantly improving image clarity and accuracy, and providing a high-quality data foundation for subsequent liveness detection and feature extraction. Secondly, to solve security issues during data transmission and storage, this invention includes an encrypted transmission unit 108, a privacy processing module 106, and a blockchain storage layer 107. The encrypted transmission unit 108 ensures data security during transmission. The privacy processing module 106 encrypts and desensitizes facial feature values ​​and erases the original image data, further protecting user privacy. The blockchain storage layer 107 adopts a Hyperledger Fabric consortium blockchain architecture, storing the encrypted facial feature vectors in a distributed ledger, ensuring data immutability and traceability, and effectively preventing data leakage. Finally, to address the issues of device load balancing and computing resource allocation, this solution incorporates the load adjustment module 104. This module dynamically adjusts the data flow and computing resource allocation based on the processing results of the liveness detection module 103 and the real-time load of the two edge computing layers 105. Simultaneously, the edge computing layer 105 deploys a lightweight model to achieve real-time extraction of facial feature vectors, improving the overall processing efficiency and resource utilization of the system. This approach solves the technical problems of existing technologies where acquisition devices rely solely on a single spectral band to acquire images, resulting in poor quality facial images; and the lack of security mechanisms in data transmission and storage, leading to serious risks of facial data leakage and jeopardizing user privacy.

[0060] Please see Figure 2 , Figure 2 This is a schematic diagram of the public area face capture system of the present invention.

[0061] The present invention also provides a public area face capture system, including the public area face capture device as described above, and further including an energy management module 201, an intelligent analysis layer 202, an anti-occlusion mechanism 203 and an environmental perception module 204. The energy management module 201 includes a dual-mode power supply interface 205 and a dynamic power consumption adjustment module 206. The anti-occlusion mechanism 203 includes a multi-sensor fusion detector 207 and a hierarchical response mechanism 208.

[0062] In this specific embodiment, both the energy management module 201 and the anti-obstruction mechanism 203 are connected to the multispectral imaging module 101, and the intelligent analysis layer 202 is connected to the blockchain storage layer 107. The intelligent analysis layer 202 is built on the Spark big data framework to construct a behavior analysis engine, which supports real-time trajectory tracking, frequency statistics, and abnormal behavior warnings. The intelligent analysis layer 202 is built on the Spark big data framework to construct a powerful behavior analysis engine, which can process and analyze massive amounts of multispectral imaging data in real time, and quickly realize functions such as real-time trajectory tracking, frequency statistics, and abnormal behavior warnings.

[0063] The dual-mode power supply interface 205 and the dynamic power consumption adjustment module 206 are both connected to the multispectral imaging module 101. The synergy between the dual-mode power supply interface 205 and the dynamic power consumption adjustment module creates a comprehensive energy security system for the multispectral imaging module 101. The dual-mode power supply interface 205 supports multiple power supply methods, easily adapting to both stable mains power and flexible battery power, ensuring a continuous power supply for the multispectral imaging module 101 in different scenarios, greatly enhancing the applicability and stability of the device. The dynamic power consumption adjustment module 206 can adjust the power consumption in real time according to the actual working status of the multispectral imaging module 101.

[0064] Secondly, the multi-sensor fusion detector 207 integrates millimeter-wave radar, infrared thermal imaging and image semantic analysis to identify three types of occlusion scenarios: rain and snow, paint stains and physical covering.

[0065] The hierarchical response mechanism 208 is configured as follows:

[0066] The waterproof coating activates its electrostatic repulsion function when rain or snow is detected.

[0067] After identifying the contaminant, spray a fine mist of organic solvent.

[0068] When physically obstructed, a 110dB audible and visual alarm is triggered and location information is uploaded.

[0069] Furthermore, the environmental perception module 204 is connected to the dynamic supplementary lighting module 102. The environmental perception module 204 acquires and analyzes data in the environment in real time and triggers the dynamic supplementary lighting module 102 to adjust the supplementary lighting parameters.

[0070] Using the public area face capture system of this embodiment, the intelligent analysis layer 202 is built on the Spark big data framework to construct a behavior analysis engine, which supports real-time trajectory tracking, frequency statistics and abnormal behavior warning. The intelligent analysis layer 202 is built on the Spark big data framework to construct a powerful behavior analysis engine, which can process and analyze massive amounts of multispectral imaging data in real time, and quickly realize functions such as real-time trajectory tracking, frequency statistics and abnormal behavior warning.

[0071] Please see Figure 3 , Figure 3 This is a flowchart of the steps of the public area face collection method of the present invention.

[0072] This invention also provides a method for facial recognition in public areas, applicable to the facial recognition system for public areas as described above.

[0073] S1. The multispectral imaging module 101 simultaneously acquires face images in at least two spectral bands;

[0074] S2. The encrypted transmission unit 108 dynamically generates a session key and uses the SM4 algorithm to encrypt the image data and transmits it to the liveness detection module 103. The liveness detection module 103 verifies the liveness by analyzing micro-expressions, skin texture and blood flow information, and only allows verified data to enter the load adjustment module 104.

[0075] S3. The load adjustment module 104 dynamically allocates data streams according to the real-time resource occupancy rate of the edge computing layer 105. Each edge computing layer 105 deploys a lightweight model to extract facial feature vectors and transmits them to the privacy processing module 106.

[0076] S4. The privacy processing module 106 uses homomorphic encryption to de-identify the feature vector and erases the original image. The blockchain storage layer 107 writes the de-identified data into the distributed ledger through the PBFT consensus algorithm, and each node saves a complete copy.

[0077] S5. The intelligent analysis layer 202 tracks the trajectory in real time, counts the frequency, and issues warnings for abnormal behavior based on the Spark framework. The anti-obstruction mechanism 203 triggers an audible and visual alarm, switches to backup power, and uploads location information when it detects physical obstruction.

[0078] The dynamic fill light module 102 outputs a PWM signal based on environmental perception data to drive the fill light for spectral adaptive compensation.

[0079] The protective component 110 monitors equipment deformation in real time and cuts off power and destroys data when an abnormality occurs.

[0080] The above description discloses only one preferred embodiment of the present invention, and should not be construed as limiting the scope of the present invention. Those skilled in the art will understand that all or part of the processes of the above embodiments can be implemented, and equivalent changes made in accordance with the claims of the present invention are still within the scope of the invention.

Claims

1. A face capture device for public areas, characterized in that, The system includes a multispectral imaging module, a dynamic illumination module, a liveness detection module, a load adjustment module, at least two edge computing layers, a privacy processing module, a blockchain storage layer, an encrypted transmission unit, a dynamic key update module, and protection components. The encrypted transmission unit is connected to the multispectral imaging module, the liveness detection module is connected to the encrypted transmission unit, the load adjustment module is connected to the liveness detection module, both edge computing layers are connected to the load adjustment module, the privacy processing module is connected to both edge computing layers, the blockchain storage layer is connected to the privacy processing module, all protection components are connected to the multispectral imaging module, the dynamic illumination module is also connected to the multispectral imaging module, and the dynamic key update module is connected to the encrypted transmission unit. The multispectral imaging module is used to acquire facial image data in multiple spectral bands; The encrypted transmission unit is used to encrypt and transmit the data collected by the multispectral imaging module to the liveness detection module; The liveness detection module is used to determine the liveness of the acquired face images, so that only live data is allowed to enter the edge computing layer. The load adjustment module is used to dynamically adjust the data flow and computing resource allocation based on the processing results of the liveness detection module and the real-time load of the two edge computing layers. The edge computing layer deploys a lightweight model for real-time extraction of facial feature vectors; The privacy processing module is used to encrypt and desensitize the extracted facial feature values ​​and erase the original image data; The blockchain storage layer adopts the Hyperledger Fabric consortium blockchain architecture, which encrypts the facial feature vector and stores it in a distributed ledger. Each node saves a complete copy of the data, ensuring that a single point of failure does not affect the system availability. The dynamic fill light module uses a spectral adaptive controller, which predicts illumination compensation parameters based on a convolutional neural network and outputs a PWM dimming signal with an adjustable duty cycle to the fill light for fill light. The dynamic key update module is used to periodically and dynamically update the encryption key used by the encrypted transmission unit.

2. The public area face capture device as described in claim 1, characterized in that, The encrypted transmission unit includes a data encryption module, a data decryption module, a high-efficiency transmission layer, a key manager, and a key generator. The high-efficiency transmission layer is connected to both the multispectral imaging module and the liveness detection module. The data encryption module and the data decryption module are respectively connected to the multispectral imaging module and the liveness detection module. The key manager is connected to both the data encryption module, the data decryption module, and the key generator.

3. The public area face capture device as described in claim 2, characterized in that, The protective component includes a housing and a piezoelectric sensor array. The housing covers the multispectral imaging module, and the piezoelectric sensor array monitors the deformation of the housing in real time. When the deformation exceeds a threshold, the power is cut off and data destruction is initiated.

4. A public area face capture system, comprising the public area face capture device as described in claim 3, characterized in that, The public area face capture system also includes an energy management module, an intelligent analysis layer, and an anti-obstruction mechanism. The energy management module and the anti-obstruction mechanism are both connected to the multispectral imaging module. The intelligent analysis layer is connected to the blockchain storage layer. The intelligent analysis layer is based on the Spark big data framework to build a behavior analysis engine, which supports real-time trajectory tracking, frequency statistics, and abnormal behavior warnings.

5. The public area face capture system as described in claim 4, characterized in that, The energy management module includes a dual-mode power supply interface and a dynamic power consumption adjustment module, both of which are connected to the multispectral imaging module.

6. The public area face capture system as described in claim 5, characterized in that, The anti-occlusion mechanism includes a multi-sensor fusion detector and a hierarchical response mechanism. The multi-sensor fusion detector integrates millimeter-wave radar, infrared thermal imaging and image semantic analysis to identify three types of occlusion scenarios: rain and snow, paint and dirt, and physical occlusion. The hierarchical response mechanism is configured as follows: The waterproof coating activates its electrostatic repulsion function when rain or snow is detected. After identifying the contaminant, spray a fine mist of organic solvent. When physically obstructed, a 110dB audible and visual alarm is triggered and location information is uploaded.

7. The public area face capture system as described in claim 6, characterized in that, The public area face capture system also includes an environment perception module, which is connected to the dynamic fill light module. The environment perception module acquires and analyzes data in the environment in real time and triggers the dynamic fill light module to adjust the fill light parameters.

8. A method for capturing faces in public areas, applied to the public area face capture system as described in claim 7, characterized in that, The multispectral imaging module simultaneously acquires face images in at least two spectral bands; The encrypted transmission unit dynamically generates a session key and uses the SM4 algorithm to encrypt the image data before transmitting it to the liveness detection module. The liveness detection module verifies liveness through micro-expression, skin texture, and blood flow information analysis, and only allows verified data to enter the load adjustment module. The load adjustment module dynamically allocates data streams based on the real-time resource utilization of the edge computing layer, and each edge computing layer deploys a lightweight model to extract facial feature vectors and transmits them to the privacy processing module. The privacy processing module uses homomorphic encryption to de-identify the feature vector and erases the original image. The blockchain storage layer writes the de-identified data into the distributed ledger through the PBFT consensus algorithm, and each node saves a complete copy. The intelligent analysis layer tracks trajectories in real time, counts frequencies, and issues warnings for abnormal behaviors based on the Spark framework. The anti-obstruction mechanism triggers an audible and visual alarm, switches to backup power, and uploads location information when it detects physical obstruction.

9. The public area face capture system as described in claim 8, characterized in that, The dynamic fill light module outputs a PWM signal based on environmental perception data to drive the fill light for spectral adaptive compensation.

10. The public area face capture system as described in claim 9, characterized in that, The protective component monitors equipment deformation in real time and cuts off power and destroys data in case of abnormality.

Images