Security and disinfection integrated fire early-stage detection and early-warning system integrating infrared temperature measurement and multi-intelligent algorithm

Abstract

A security and disinfection integrated fire early-stage detection and early-warning system fusing infrared temperature measurement and a multi-intelligent algorithm comprises a thermal imaging acquisition module, an intelligent processing module, a linkage control module and an alarm module. The thermal imaging acquisition module realizes temperature field distribution acquisition and real-time image shooting of a monitoring area; the intelligent processing module is used for carrying out noise reduction, enhancement and feature extraction on the acquired thermal imaging image, and accurately identifying flame and high-temperature point fire features; the linkage control module is connected with the security and protection monitoring system, the alarm module receives a fire early warning signal of the intelligent processing module, gives an alarm to related personnel in time in a sound-light alarm and short message notification mode, and stores early warning information and related data locally at the same time; all the modules work cooperatively to achieve accurate monitoring and quick response to a fire disaster. The system has the characteristics of dual-light fusion, high-precision temperature measurement and fire early warning, a multi-pseudo-color display mode, an intelligent alarm linkage mechanism, high-definition visible light imaging capability, high environmental adaptability, expansibility and integration.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of fire monitoring, in particular to a fire early detection and early warning system integrating infrared temperature measurement and multi-intelligent algorithm. BACKGROUND

[0002] In wind power plants, factories and other industrial sites, fire hazard monitoring is crucial. Traditional fire monitoring equipment often has single function. Although thermal imaging equipment can realize temperature monitoring, its temperature measurement accuracy, image analysis capability and collaborative work capability with other security systems need to be improved in complex environments. With the development of technology, various thermal imaging related patent technologies have emerged, such as infrared thermal imaging temperature measurement core and thermal imaging image processing method. However, these technologies are mostly used independently and have not been effectively integrated, resulting in limited comprehensive monitoring performance.

[0003] Existing thermal imaging fire monitoring devices are mostly based on single thermal imaging temperature measurement technology, such as using the infrared thermal imaging temperature measurement core of Zhejiang Dahua Technology Co., Ltd. to realize temperature collection. Although it can complete the basic temperature measurement function, it lacks advanced image processing algorithms and linkage mechanism with security systems. Some devices have introduced simple image processing, but have not combined optimization algorithms such as thermal imaging image processing method, which has poor effect in image noise reduction and target recognition, and cannot realize rapid linkage with video monitoring system when fire occurs, affecting fire disposal efficiency. SUMMARY

[0004] In order to overcome the above technical problems, the purpose of the present application is to provide a fire early detection and early warning system integrating infrared temperature measurement and multi-intelligent algorithm, which has the characteristics of dual light fusion design, high-precision temperature measurement and fire warning, multi-pseudo-color display mode, intelligent alarm linkage mechanism, high-definition visible light imaging capability, strong environmental adaptability, and expansion and integration.

[0005] The technical solution adopted by the present application is: The fire early detection and early warning system integrating infrared temperature measurement and multi-intelligent algorithm includes a thermal imaging acquisition module, an intelligent processing module, a linkage control module and an alarm module, which work together to realize accurate monitoring and rapid response to fire; The thermal imaging acquisition module realizes temperature field distribution acquisition and real-time image shooting of the monitoring area, and obtains thermal imaging; The intelligent processing module has a multi-stage intelligent image algorithm based on the improved thermal imaging image processing method, which can reduce noise, enhance and extract features of the collected thermal imaging image, and accurately identify flame, high-temperature point fire features; The linkage control module is connected with the security and protection monitoring system, when the intelligent processing module judges that there is a fire hazard, the video monitoring equipment is rapidly linked through the preset communication protocol, the real-time high-definition picture of the corresponding area is called, and the temperature data and image information are synchronously transmitted to the monitoring center.

[0006] The thermal imaging acquisition module includes an infrared thermal imaging temperature measurement core as a core temperature measurement component, integrates a high-definition optical lens, realizes temperature field distribution acquisition and real-time image shooting of the monitoring area, and simultaneously introduces a thermal imaging device to optimize the heat dissipation performance and anti-interference capability of the device, and ensure stable operation in harsh environments.

[0007] The thermal imaging acquisition module includes a non-refrigerated vanadium oxide (VOx) micro bolometer focal plane detector, a visible light imaging module, an optical system, an image processing unit, an audio and video module, an alarm and IO interface, a network communication module, a storage module and a light supplement system. The non-refrigerated vanadium oxide (VOx) micro bolometer focal plane detector is connected to the FPGA end of the image processing unit through an LVDS high-speed data interface; the visible light imaging module is connected to the FPGA end of the image processing unit through an MIPI-CSI image interface; the thermal imaging lens of the optical system is installed at the front end of the VOx detector, and the visible light lens is installed at the front end of the CMOS sensor of the visible light imaging module; the image processing unit is connected to the light supplement system, the alarm and IO interface through a GPIO interface, to the storage module through an SDIO interface, and to the network communication module through an RMII interface; the microphone of the audio and video module is connected to the DSP end of the image processing unit through an I2S audio interface, and the speaker is connected to the DSP end through a PWM audio drive interface; the network communication module, the storage module and the alarm and IO interface realize bidirectional data interaction with the image processing unit through corresponding data interfaces.

[0008] The non-refrigerated vanadium oxide (VOx) micro bolometer focal plane detector is a "core sensing element" for thermal imaging acquisition, responsible for converting the infrared radiation signal of the monitoring area into a digital radiation value (DN value), providing original data for subsequent temperature calculation; supports 8-14μm long-wave infrared band (can penetrate smoke and dark environment), resolution can be selected as 384×288 or 640×512, NETD≤50mK (can distinguish 0.05℃ small temperature difference), ensures accurate capture of early high temperature points (such as overheating of electrical equipment).

[0009] The non-refrigerated vanadium oxide (VOx) micro bolometer focal plane detector is directly connected to the FPGA end of the image processing unit through an LVDS high-speed data interface: DN value matrix (each frame contains 384×288 / 640×512 pixel radiation data) is transmitted to the FPGA end in real time; Receive the parameter configuration instruction (such as integration time, gain adjustment) sent by the FPGA end, and adapt the infrared signal collection sensitivity in different environments.

[0010] The visible light imaging module (1 / 2.7 inch CMOS, 4 million pixels) is used to supplement the "detail recognition ability" of the thermal imaging image, collect the color visible light image of the monitoring area, and is used for fire visualization confirmation (such as identifying smoke profile and flame shape); support WDR (wide dynamic range), low-illumination full-color imaging (0.005 lux@F1.6), even in night or backlight environment, the target details can be clearly presented, and "target misjudgment" caused by relying on thermal imaging only (such as distinguishing "high-temperature equipment" from "open fire") is avoided.

[0011] The visible light imaging module is connected to the FPGA end of the image processing unit through the MIPI-CSI image interface: transmits RGB color image data (frame rate 25 / 30 fps, resolution 2560x1440) to the FPGA; receives the control signal (such as exposure parameter, white balance adjustment, WDR mode switching) sent by the FPGA, and ensures the image quality in different light environments.

[0012] The optical system is a double-lens, including a thermal imaging lens and a visible light lens, which provides a "precise light path channel" for infrared radiation and visible light signals, ensures efficient focusing of light to corresponding detectors / sensors, and reduces light loss and interference: The thermal imaging lens adopts germanium glass material (8-14 μm waveband transmittance ≥92%), which adapts to the infrared sensing needs of VOx detector; The visible light lens adopts multi-layer coated glass (glare resistance ≤1%), which reduces the interference of environmental light reflection on CMOS imaging and adapts to low-illumination full-color needs.

[0013] The thermal imaging lens is directly installed in front of the uncooled VOx detector through a mechanical support, and the lens optical axis is accurately aligned with the photosensitive surface of the detector (error ≤0.1 mm); The visible light lens is installed in front of the CMOS sensor of the visible light imaging module through a support of the same specification, and the field angle of the lens is matched with the thermal imaging lens (to ensure that the two light images can be accurately registered subsequently).

[0014] The image processing unit includes DSP and FPGA architecture, which is the "data processing center" of the thermal imaging acquisition module, and undertakes the core tasks of "raw data preprocessing→algorithm operation→control instruction output"; Specifically, it includes: The FPGA is responsible for high-speed data preprocessing and dual-light image pixel-level registration (error <1.5 pixels); The preprocessing includes thermal imaging DN value denoising, bad point correction, visible light image distortion correction; The DSP is responsible for complex algorithm operation (temperature solution based on Planck formula, 12 kinds of pseudo-color mapping, ROI region temperature analysis, fire point feature recognition), and simultaneously generates linkage control instructions.

[0015] The FPGA end of the image processing unit is connected with the VOx detector through the LVDS interface, connected with the visible light imaging module through the MIPI-CSI interface, and receives original data; The bidirectional data transmission between the FPGA and the DSP is realized through the AXI high-speed internal bus (the FPGA sends the preprocessed data to the DSP, and the DSP feeds back configuration instructions to the FPGA); The DSP end of the image processing unit is connected with the light supplementing system through the GPIO interface, connected with the alarm and IO interface, connected with the storage module through the SDIO interface, connected with the network communication module through the RMII interface, and outputs the processing result and control instructions.

[0016] The audio and video module includes a built-in microphone and a high-loudness loudspeaker, and realizes "on-site interaction" and "local early warning": The microphone supports two-way voice intercom (such as the monitoring center issuing evacuation instructions to on-site personnel), supports a built-in adaptive spectral subtraction noise reduction algorithm, dynamically adjusts the voice signal gain by analyzing the environmental noise spectrum in real time, suppresses the background noise (such as fan noise and echo), and extracts clear voice; The loudspeaker plays a pre-recorded alarm prompt sound (such as "detecting high-temperature hidden danger, please check in time"), and the loudness is ≥70dB@1m (A-weighted), which ensures that on-site personnel can quickly perceive the warning.

[0017] The microphone is connected to the DSP end of the image processing unit through the I2S audio interface, and transmits the collected voice signal to the DSP for coding (supports G.711 format); The loudspeaker is connected to the DSP end through the PWM audio drive interface, and receives the audio playing instructions (alarm sound / intercom voice) sent by the DSP.

[0018] The alarm and IO interface includes 1-way alarm input and 1-way relay output, and is used to expand the "external device linkage capability" of the system, and realizes the bidirectional signal interaction between the "system-external device": Alarm input: receives the dry contact point signal (DC5-24V) of the external security device (such as a smoke detector and a door magnetic switch), and triggers the targeted monitoring of the system (such as when the smoke alarm sounds, the thermal imaging automatically focuses on the smoke area); Relay output: outputs a dry contact point control signal (rated load AC 250V / 5A, DC 30V / 5A), and links with external execution devices (such as a police light flashing, an exhaust fan starting, and a fire host alarm).

[0019] Alarm and IO interface is connected to the DSP end of image processing unit through GPIO level interface: transmit alarm input signal of external device to DSP; Receive the relay closing / opening instruction sent by DSP to control the start / stop of external device.

[0020] The network communication module (10 / 100 Mbps adaptive network interface) realizes local and remote data transmission and power supply integration: The network communication module supports RTSP / ONVIF / GB28181 protocol, and pushes real-time thermal imaging / visible light video stream, alarm information (temperature data, device ID, ROI coordinates) to NVR / monitoring center; The network communication module is powered by POE: in line with IEEE802.3af standard, transmits data and power (maximum power supply power 15.4W) through network cable, no need to deploy power line additionally, simplifies installation.

[0021] The network communication module is connected to the DSP end of image processing unit through RMII Ethernet interface: Receive the video stream (H.265 encoding) and alarm data (JSON format) packaged by DSP, and transmit to remote platform; Receive the control instruction (such as adjusting temperature threshold, turning on light compensation) sent by remote platform, and forward to DSP for execution.

[0022] The storage module (MicroSD card slot) realizes local data retention, which is convenient for offline query and evidence tracing: supports maximum 256GB MicroSD card, compatible with FAT32 / exFAT file system; Support two storage modes: loop recording (automatic cover old data according to code rate 1-8Mbps), event recording (20 seconds video before and after alarm + double light snapshot picture is separately marked and stored, picture naming format: DeviceID_YYYYMMDD_HHMMSS.jpg).

[0023] The storage module is connected to the DSP end of image processing unit through SDIO data interface: receive the encoded video stream, double light snapshot picture and structured log (alarm time, temperature value, ROI area) sent by DSP; respond to the "data reading instruction" of DSP (such as remote platform calling local historical recording).

[0024] The light supplement system is built-in high-efficiency infrared lamp array, adopts annular uniform arrangement mode, 8 940nm infrared lamps are distributed along the circumference of the visible light lens, the light supplement angle is accurately matched with the field of view angle (such as 60°) of the visible light lens (error ±5°), and the maximum light supplement distance is 25 meters; the problem of "visible light imaging blur in low-illumination environment" is solved, and it is ensured that the visible light module can still clearly capture target details in night or dim environment (such as underground warehouse), and the fire is confirmed by auxiliary thermal imaging.

[0025] The light supplement system is connected to the DSP end of the image processing unit through the GPIO control interface: receiving the light supplement start-stop instruction sent by the DSP (triggering logic: the DSP automatically judges through the brightness detection value of the visible light module, and automatically starts when the average brightness of the visible light image is detected to be lower than 5lux (low-illumination threshold), and automatically stops when it is higher than 15lux; or receiving the manual start instruction of the remote platform); The light supplement working state (such as "has started" "fault") is fed back to the DSP.

[0026] 1. Thermal imaging core: non-refrigeration vanadium oxide (VOx) microbolometer focal plane detector, the resolution is according to the model (such as 384x288 or 640x512), and the NETD is less than or equal to 50mK.

[0027] 2. Visible light imaging module: 1 / 2.7 inch CMOS, 4 million pixels, supporting WDR, low-illumination full-color imaging.

[0028] 3. Optical system: double-lens design: thermal imaging lens (germanium glass material) + visible light lens (multilayer coated glass).

[0029] 4. Image processing unit: DSP+FPGA architecture, realizing double-light image fusion, temperature calculation, pseudo-color mapping, alarm analysis and other algorithms.

[0030] 5. Audio and video module: built-in microphone, high-loudness loudspeaker (≥70dB), supporting two-way voice intercom and alarm prompt sound playing.

[0031] 6. Alarm and IO interface: 1-way alarm input, 1-way relay output, which can be connected with external devices such as smoke sensor, door magnet, warning light and the like.

[0032] 7. Network communication module: supporting 10 / 100Mbps adaptive network port, supporting POE power supply, supporting RTSP / ONVIF protocol access platform.

[0033] 8. Storage module: MicroSD card slot, supporting 256GB at most, supporting cycle recording and event recording.

[0034] 9. Light supplement system: built-in high-efficiency infrared lamp array, maximum light supplement distance 25 meters.

[0035] 10. Structure shell: metal shell, IP67 protection level, working temperature -30℃~+70℃, suitable for long-term outdoor operation.

[0036] The intelligent processing module is through wavelet threshold denoising→ Laplace edge enhancement→ convolutional neural network (CNN) feature extraction three-level processing flow; The Planck formula is the core basis for temperature calculation, providing accurate temperature data support for the three-level processing flow; the actual temperature measurement model is used to correct the temperature data to ensure the accuracy of the temperature information input into the three-level processing flow; the image features processed by the three-level processing flow combined with the temperature data further improve the fire point feature recognition accuracy; Accurate identification of the dynamic profile of the flame (such as irregular edges + flicker frequency), temperature gradient change of the high temperature point (such as ≥5℃ / cm), fire features, while excluding non-fire interference sources such as sunlight reflection, welding sparks, and temporary high temperature equipment, false alarm rate reduced to ≤0.1 times / month; Thermal imaging equipment calculates the surface temperature of the target object by detecting the infrared energy it radiates, and its basic physical basis is Planck's radiation law and the Stefan-Boltzmann law; 1. Relationship between radiation intensity and temperature (Planck formula): L: spectral radiance λ: wavelength (usually 8-14μm long-wave infrared) T: object absolute temperature (unit: K) h: Planck's constant c: speed of light k: Boltzmann constant The device collects the radiation values of each pixel point in the scene through the sensor, and combines the emissivity (ε), environmental temperature, relative humidity, distance, etc. Parameters for compensation; 2. Actual temperature measurement model (simplified): Combined with a temperature calibration algorithm in an infrared thermal imaging temperature measurement method, the temperature measurement accuracy is improved, and the temperature of the monitored area is analyzed and judged in real time.

[0037] The thermal imaging analysis and judgment operation steps (taking fire warning as an example): Step 1: Device installation and configuration Fixed installation at the high point of the monitoring area, avoiding high temperature interference sources (such as air conditioner outdoor unit, sunlight direct surface); connect the power (POE or DC12V), after network online, login Web management interface or NVR platform; Step 2: Parameter basic setting Set time, time zone, video stream (recommended main stream H.265@1080P~4MP); turn on thermal imaging channel, select pseudo-color mode (recommended "rainbow" or "iron red" to enhance temperature difference identification); Calibration of environmental parameters: Input ambient temperature (use external sensor); set relative humidity, observation distance; adjust emissivity (usually set to 0.95, metal surface needs to be adjusted) Step 3: Configure temperature measurement area and alarm rules: Enter the "temperature analysis" page and draw one or more temperature measurement ROI areas (such as electrical boxes, warehouse stacking areas). Set alarm conditions: Enable "high temperature alarm": trigger when the maximum temperature is greater than 70°C; Enable "temperature rise rate alarm": such as rising more than 10°C in 1 minute; Enable "cold and hot spot automatic tracking": the system automatically identifies the hottest / coldest point in the picture and alarms; Set alarm delay: avoid false alarms (recommended 2~5 seconds); Step 4: Alarm linkage configuration; Bind alarm output: trigger relay to close, link alarm light, exhaust fan or fire main. Enable audible and visual alarm: local speaker emits high-decibel alarm sound (≥70dB). Configure upload platform: alarm information is pushed to the central management platform or mobile APP notification; Step 5: Real-time monitoring and event response Observe the color change of the thermal imaging picture in real time, and focus on the red / white area. When an alarm occurs: check the visible light picture to confirm whether it is a real fire (such as smoke, flame); play back the temperature change trend of the historical video; manually start recording or capture and save evidence; if it is confirmed as a real fire, immediately notify the on-site personnel to handle or link the automatic fire extinguishing system; Step 6: Regular maintenance and calibration Clean the lens every quarter (use special air blowing ball and lens paper). Regularly check the temperature measurement accuracy (use black body or standard heat source to verify). Update firmware to get the latest features and security patches.

[0038] The linkage control module is the core functional unit of "integrated security and fire protection", responsible for automatically triggering the video monitoring system response when the thermal imaging detects abnormalities, and uploading multi-dimensional data (temperature + image) to the monitoring center; when the thermal imaging detects fire hazards, the system will automatically retrieve the high-definition visible light picture of the corresponding area and display it with the thermal imaging image, achieving dual-mode evidence preservation and remote confirmation.

[0039] The specific method steps are: Step 1: Trigger condition judgment The intelligent processing module detects any of the following conditions: the maximum temperature of the area is greater than or equal to a threshold value (such as 70°C) and the temperature rise rate exceeds a set value (such as 10°C / min); automatically identifies the fire point feature (high temperature point + rapid expansion); the system generates a "fire warning" event and enters the linkage process.

[0040] Step 2: Local video stream retrieval Retrieval method: built-in dual-channel video acquisition system (thermal imaging + visible light) in the device, no external request required. Retrieval action: the main control chip caches the last 10 seconds of video frames (H.265 encoding) in real time; saves a video clip of 20 seconds (5 seconds before the trigger and 15 seconds after the trigger) immediately after triggering; synchronously captures a high-resolution picture (including pseudo-color thermal image and visible light image).

[0041] Step 3: Image and data packaging and transmission Data packet content: current thermal image (including temperature annotation); synchronous visible light image (with timestamp) Highest / lowest temperature value, alarm type, device ID, geographic location; video clip (H.265 encoding, low bit rate); Transmission method: Push to NVR / VMS platform; use RTSP or RTMP protocol to actively push real-time stream; support SIP registration (GB / T28181) to access national standard platform. HTTP / HTTPS event notification; send JSON format alarm message to preset URL, including picture link and video path. FTP upload media file; automatically upload screenshots and videos to the specified server directory. Mobile push; push alarm information and thumbnails through platform APP.

[0042] Step 4: Monitoring center reception and display After the center platform receives the alarm: automatically pops up the corresponding channel video window; displays dual-picture comparison (thermal imaging vs. visible light); plays the event video and annotates the temperature change curve; supports manual review and one-key confirmation of fire alarm; linkage control logic in the thermal imaging integrated device (state machine description).

[0043] This module learns from the linkage control logic in the thermal imaging integrated device to ensure the rapidity and accuracy of the linkage response.

[0044] Alarm module: receives the fire warning signal from the intelligent processing module, timely alarms relevant personnel through sound and light alarm, SMS notification, etc., and stores the warning information and related data locally for subsequent query and analysis.

[0045] The beneficial effects of the present application are: Improve fire monitoring accuracy: By integrating high-precision temperature measurement algorithms and intelligent image processing technology, it can accurately identify fire hazards, effectively reduce false positive and false negative rates, and improve the reliability of early fire warning.

[0046] Speed up emergency response: Rapid linkage with security monitoring systems enables staff to obtain real-time images and temperature information of the fire scene in time, providing support for rapid development of disposal plans and shortening emergency response time.

[0047] Enhance environmental adaptability: Optimized device structure design and anti-interference capability enable stable operation in high-temperature, dust, vibration and other complex industrial environments, expanding the application range of the device.

[0048] Improve system integration efficiency: Integrating multiple functions into one reduces the connection and debugging between devices, reducing system construction cost and maintenance difficulty. BRIEF DESCRIPTION OF DRAWINGS

[0049] Figure 1 The linkage control logic flowchart inside the device.

[0050] Figure 2 The overall workflow example (take fire occurrence as an example).

[0051] Figure 3 The internal structure diagram of the integrated security and fire temperature sensing and imaging device.

[0052] Figure 4 System module connection and function distribution diagram. DETAILED DESCRIPTION

[0053] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor are within the scope of protection of the present application.

[0054] As shown in Figure 2 The integrated security and fire early detection and warning system combining infrared temperature measurement and multiple intelligent algorithms includes a thermal imaging acquisition module, an intelligent processing module, a linkage control module and an alarm module, which work together to achieve accurate monitoring and rapid response to fire; The thermal imaging acquisition module realizes temperature field distribution acquisition and real-time image shooting of the monitoring area; The intelligent processing module "integrates a multi-stage intelligent image algorithm based on the improved thermal imaging image processing method. Through a three-stage processing flow of 'wavelet threshold denoising → Laplace edge enhancement → convolutional neural network (CNN) feature extraction', the dynamic profile of the flame (such as irregular edge + flicker frequency), the temperature gradient change of the high-temperature point (such as ≥5℃ / cm) and other fire characteristics are accurately identified, while non-fire interference sources such as sunlight reflection, welding sparks and temporary high-temperature equipment are excluded, and the false alarm rate is reduced to ≤0.1 times / month". The collected thermal imaging images are denoised, enhanced and feature extracted to accurately identify the flame and high-temperature point fire characteristics; The linkage control module is connected with the security monitoring system. When the intelligent processing module determines that there is a fire hazard, the video monitoring equipment is quickly linked through the preset communication protocol, the real-time high-definition picture of the corresponding area is retrieved, and the temperature data and image information are synchronously transmitted to the monitoring center.

[0055] Thermal imaging acquisition module: adopts the infrared thermal imaging temperature measurement core of Zhejiang Dahua Technology Co., Ltd. as the core temperature measurement component, integrates a high-definition optical lens, realizes temperature field distribution acquisition and real-time image shooting of the monitoring area, and introduces the structural design concept of the thermal imaging equipment, optimizes the heat dissipation performance and anti-interference ability of the equipment, and ensures stable work in harsh environment.

[0056] The thermal imaging acquisition module includes a non-cooled vanadium oxide (VOx) microbolometer focal plane detector, a visible light imaging module, an optical system, an image processing unit, an audio and video module, an alarm and IO interface, a network communication module, a storage module, and a light supplementing system; 1. Non-cooled vanadium oxide (VOx) microbolometer focal plane detector Core role The "core sensing element" of thermal imaging acquisition is responsible for converting the infrared radiation signal of the monitoring area into digital radiation value (DN value) to provide original data for subsequent temperature calculation; supports 8-14μm long-wave infrared band (can penetrate smoke and dark environment), resolution can be selected as 384×288 or 640×512, NETD≤50mK (can distinguish 0.05℃ small temperature difference), and ensures accurate capture of early high-temperature points (such as overheated electrical equipment).

[0057] Connection relationship with other components Connected to the FPGA end of the image processing unit through the LVDS high-speed data interface: Real-time transmission of DN value matrix (each frame contains 384×288 / 640×512 pixel radiation data) to FPGA; Receive parameter configuration instructions (such as integration time and gain adjustment) sent by FPGA to adapt the infrared signal acquisition sensitivity in different environments.

[0058] 2. Visible light imaging module (1 / 2.7-inch CMOS, 4 million pixels) Core role Supplement the "detail recognition ability" of thermal imaging images, collect color visible light images of the monitoring area, and use them for fire visualization confirmation (such as identifying smoke contours and flame patterns); support WDR (wide dynamic range) and low-illumination full-color imaging (0.005 lux @ F1.6), which can clearly present target details even in night or backlight environments, avoiding "target misjudgment" caused by relying solely on thermal imaging (such as distinguishing between "high-temperature equipment" and "open fire").

[0059] Connection relationship with other components Connected to the FPGA end of the image processing unit through the MIPI-CSI image interface: Transmit RGB color image data to the FPGA (frame rate 25 / 30 fps, resolution 2560x1440); Receive control signals sent by the FPGA (such as exposure parameters, white balance adjustment, and WDR mode switching) to ensure image quality in different lighting environments.

[0060] 3. Optical system (double-lens design: thermal imaging lens + visible light lens) Core role Provide "precise light path channels" for infrared radiation and visible light signals to ensure efficient focusing of light onto corresponding detectors / sensors and reduce light loss and interference: Thermal imaging lens: made of germanium glass (8-14 μm waveband transmittance ≥92%) to meet the infrared sensing needs of VOx detectors; Visible light lens: made of multi-layer coated glass (glare resistance ≤1%) to reduce the interference of environmental light reflection on CMOS imaging and meet the low-illumination full-color demand.

[0061] Connection relationship with other components Thermal imaging lens: directly installed in front of the uncooled VOx detector through a mechanical support, with the lens optical axis precisely aligned with the detector photosurface (error ≤0.1 mm); Visible light lens: installed in front of the CMOS sensor of the visible light imaging module through a support of the same specification, with the lens field of view matched with the thermal imaging lens (to ensure accurate registration of the two light images).

[0062] 4. Image processing unit (DSP + FPGA architecture) Core role The "data processing center" of the thermal imaging acquisition module, responsible for the core tasks of "raw data preprocessing → algorithm operation → control instruction output", including: FPGA: responsible for high-speed data preprocessing (thermal imaging DN value denoising, bad point correction; visible light image distortion correction), dual light image pixel-level registration (error <1.5 pixels); DSP: responsible for complex algorithm operation (temperature solution based on Planck formula, 12 kinds of pseudo-color mapping, ROI region temperature analysis, fire point feature identification), while generating linkage control instructions.

[0063] Connection relationship with other components Input connection: connect VOx detector through LVDS interface, connect visible light imaging module through MIPI-CSI interface, receive raw data; Internal interaction: realize bidirectional data transmission between FPGA and DSP through AXI high-speed internal bus (FPGA sends preprocessed data to DSP, DSP feeds back configuration instructions to FPGA); Output connection: connect light supplementing system through GPIO interface, connect alarm and IO interface, connect storage module through SDIO interface, connect network communication module through RMII interface, output processing results and control instructions.

[0064] 5. Audio and video module (built-in microphone + high-loudness loudspeaker) Core role Realize "on-site interaction" and "local early warning": Microphone: support two-way voice intercom (such as the monitoring center issuing evacuation instructions to on-site personnel), built-in noise reduction algorithm (reduce environmental noise interference); Loudspeaker: play pre-recorded alarm prompt sound (such as "detecting high temperature hidden danger, please investigate in time"), loudness ≥ 70dB@1m (A-weighted), to ensure that on-site personnel quickly perceive the warning.

[0065] Connection relationship with other components Microphone: connected to the DSP end of the image processing unit through the I2S audio interface, transmits the collected voice signal to the DSP for encoding (supports G.711 format); Loudspeaker: connected to the DSP end through the PWM audio drive interface, receives the audio playback instructions (alarm sound / intercom voice) sent by the DSP.

[0066] 6. Alarm and IO interface (1-way alarm input + 1-way relay output) Core role Expand the "external device linkage capability" of the system, realize the bidirectional signal interaction between "system-external device": Alarm input: receive the dry contact signal (DC5-24V) of external security equipment (such as smoke detector, door magnetic switch), trigger the system to monitor (such as smoke alarm, thermal imaging automatically focuses on the smoke area); Relay output: output dry contact control signal (rated load AC 250V / 5A, DC 30V / 5A), linkage external execution equipment (such as warning light flickering, exhaust fan starting, fire-fighting host alarm).

[0067] Connection relationship with other components Directly connected to the DSP end of the image processing unit through the GPIO level interface: Transmit the alarm input signal of the external device to the DSP; Receive the relay closing / opening instructions sent by the DSP to control the start / stop of the external device.

[0068] 7. Network communication module (10 / 100Mbps adaptive network interface) Core role Realize the integration of "local-remote" data transmission and power supply: Data transmission: support RTSP / ONVIF / GB28181 protocol, push real-time thermal imaging / visible light video stream, alarm information (temperature data, device ID, ROI coordinate) to NVR / monitoring center; POE power supply: comply with IEEE802.3af standard, transmit data and power (maximum power supply power 15.4W) through network cable at the same time, no need to deploy power line additionally, simplify installation.

[0069] Connection relationship with other components Connected to the DSP end of the image processing unit through the RMII Ethernet interface: Receive the video stream (H.265 encoding) and alarm data (JSON format) packaged by the DSP, and transmit them to the remote platform; Receive the control instructions sent by the remote platform (such as adjusting the temperature measurement threshold, turning on the fill light), and forward them to the DSP for execution.

[0070] 8. Storage module (MicroSD card slot) Core role Realize "local data retention" for offline query and evidence tracing: Support maximum 256GB MicroSD card, compatible with FAT32 / exFAT file system; Support two storage modes: loop recording (old data is automatically overwritten according to code rate 1-8Mbps), event recording (20 seconds of video before and after alarm + double light snapshot picture is separately marked and stored, picture naming format: DeviceID_YYYYMMDD_HHMMSS.jpg).

[0071] Connection relationship with other components Connect to the DSP end of the image processing unit through the SDIO data interface: Receive the encoded video stream, dual-light snapshot, and structured log (alarm time, temperature value, ROI area) sent by the DSP. Respond to the "data read instruction" of the DSP (such as the local history video call of the remote platform).

[0072] 9. Light supplement system (built-in high-efficiency infrared lamp array) Core role Solve the problem of "blur in low-light environment": built-in 8 940nm infrared lamps, maximum light supplement distance 25 meters, light supplement angle matching the visible light lens field of view angle (error ±5°), ensuring that the visible light module can still clearly capture target details in night or dim environment (such as underground warehouse), assisting thermal imaging to confirm fire.

[0073] Connection relationship with other components Connect to the DSP end of the image processing unit through the GPIO control interface: Receive the light-on / off instruction sent by the DSP (trigger logic: the DSP automatically determines whether to turn on or off through the brightness detection value of the visible light module; or receive the manual on instruction from the remote platform). Feedback the light working status to the DSP (such as "turned on" and "fault").

[0074] 1. Thermal imaging core: non-refrigerated vanadium oxide (VOx) microbolometer focal plane detector, resolution according to model (such as 384x288 or 640x512), NETD≤50mK.

[0075] 2. Visible light imaging module: 1 / 2.7 inch CMOS, 4 million pixels, supporting WDR, low-light full-color imaging.

[0076] 3. Optical system: dual-lens design: thermal imaging lens (germanium glass material) + visible light lens (multi-layer coated glass).

[0077] 4. Image processing unit: DSP+FPGA architecture, realizing dual-light image fusion, temperature calculation, pseudo-color mapping, alarm analysis, etc.

[0078] 5. Audio and video module: built-in microphone, high-loudness speaker (≥70dB), supporting two-way voice intercom and alarm prompt sound playback.

[0079] 6. Alarm and IO interface: 1-way alarm input, 1-way relay output, can connect external devices such as smoke sensor, door magnet, and warning light.

[0080] 7. Network communication module: support 10 / 100Mbps adaptive network interface, support POE power supply, support RTSP / ONVIF protocol access platform.

[0081] 8. Storage module: MicroSD card slot, maximum support 256GB, support cycle recording and event recording.

[0082] 9. Light supplement system: built-in high-efficiency infrared lamp array, maximum light supplement distance 25 meters.

[0083] 10. Structural shell: metal shell, IP67 protection level, working temperature -30℃~+70℃, suitable for outdoor long-term operation.

[0084] The thermal imaging temperature measurement algorithm principle of the intelligent processing module (simplified formula description): The thermal imaging equipment calculates the surface temperature of the target object by detecting the infrared energy radiated by the target object, and its basic physical basis is Planck's radiation law and Stefan-Boltzmann law. 1. Relationship between radiation intensity and temperature (Planck formula): L: spectral radiation luminance λ: wavelength (usually 8-14μm long-wave infrared) T: absolute temperature of the object (unit: K) h: Planck constant c: speed of light k: Boltzmann constant The device collects the radiation values of each pixel point in the scene through the sensor, and combines the emissivity (ε), environmental temperature, relative humidity, distance and other parameters for compensation. 2. Actual temperature measurement model (simplified): Combined with a temperature calibration algorithm in an infrared thermal imaging temperature measurement method, the temperature measurement accuracy is improved, and the temperature of the monitoring area is analyzed and judged in real time.

[0085] Thermal imaging analysis and judgment operation steps (take fire early warning as an example) Step 1: Equipment installation and configuration Fixedly installed at the high point of the monitoring area, avoiding high temperature interference sources (such as air conditioner outdoor unit, sunlight direct surface); connect the power (POE or DC12V), and after network online, log in Web management interface or NVR platform.

[0086] Step 2: Parameter basic setting Set time, time zone, video stream (it is recommended that the main stream H.265@1080P~4MP); turn on the thermal imaging channel, and select the pseudo-color mode (it is recommended to use "rainbow" or "iron red" to enhance the temperature difference identification).

[0087] Calibrate environmental parameters: Input ambient temperature (use external sensor); set relative humidity, observation distance; adjust emissivity (usually set to 0.95, metal surface needs to be adjusted) Step 3: Configure temperature measurement area and alarm rules Enter the "Temperature Analysis" page and draw one or more temperature measurement ROI areas (such as electrical boxes, warehouse stacking areas). Set alarm conditions: Enable "High Temperature Alarm": trigger when the maximum temperature is > 70°C; Enable "Temperature Rise Rate Alarm": such as rising more than 10°C in 1 minute; Enable "Cold / Hot Spot Automatic Tracking": the system automatically identifies the hottest / coldest point in the picture and alarms; Set alarm delay: avoid false alarms (recommended 2-5 seconds); Step 4: Alarm linkage configuration Bind alarm output: trigger relay to close, link alarm light, exhaust fan or fire main. Enable audible and visual alarm: local speaker emits high-decibel alarm sound (≥ 70 dB). Configure upload platform: alarm information is pushed to the central management platform or mobile APP notification.

[0088] Step 5: Real-time monitoring and event response Observe the color change of the thermal imaging picture in real time, and focus on the red / white area. When an alarm occurs: check the visible light picture to confirm whether it is a real fire (such as smoke, flame); play back the temperature change trend of the historical video; manually start recording or capture and save evidence; if it is confirmed as a real fire, immediately notify the on-site personnel to handle or link the automatic fire extinguishing system.

[0089] Step 6: Regular maintenance and calibration Clean the lens every quarter (use special air blowing ball and lens paper). Regularly check the temperature measurement accuracy (use black body or standard heat source to verify). Update firmware to get the latest features and security patches.

[0090] The linkage control module is the core functional unit of "integrated security and fire protection", responsible for automatically triggering the video monitoring system response when the thermal imaging detects abnormalities, and uploading multi-dimensional data (temperature + image) to the monitoring center; when the thermal imaging detects fire hazards, the system will automatically retrieve the high-definition visible light picture of the corresponding area and display it with the thermal imaging image, achieving dual-mode evidence preservation and remote confirmation.

[0091] The specific method steps are: Step 1: Trigger condition judgment The intelligent processing module detects any of the following conditions: the maximum temperature in the area is greater than or equal to a threshold value (e.g. 70°C) and the temperature rise rate exceeds a set value (e.g. 10°C / min); automatically identifies fire point features (high temperature point + rapid expansion); the system generates a "fire warning" event and enters the linkage process.

[0092] Step 2: Local video stream retrieval Retrieval method: built-in dual-channel video capture system (thermal imaging + visible light) in the device, no external request required. Retrieval action: the main control chip caches the last 10 seconds of video frames (H.265 encoding) in real time; immediately saves a 20-second video clip consisting of the 5 seconds before and 15 seconds after the trigger after triggering; synchronously captures a high-resolution picture (including pseudo-color thermal image and visible light image).

[0093] Step 3: Image and data packaging and transmission Data packet content: current thermal image (with temperature annotations); synchronous visible light image (with timestamp) Highest / lowest temperature value, alarm type, device ID, geographic location; video clip (H.265 encoding, low bit rate); Transmission method: Push to NVR / VMS platform; use RTSP or RTMP protocol to actively push real-time stream; support SIP registration (GB / T28181) to access national standard platform. HTTP / HTTPS event notification; send JSON format alarm message to preset URL, including picture link and video path. FTP upload media files; automatically upload screenshots and videos to the specified server directory. Mobile push; push alarm information and thumbnails through platform APP.

[0094] Step 4: Monitoring center reception and display After the center platform receives the alarm: automatically pops up the corresponding channel video window; displays dual-picture comparison (thermal imaging vs. visible light); plays the event video and labels the temperature change curve; supports manual review and one-key confirmation of fire alarm; linkage control logic in thermal imaging integrated device (state machine description); as shown below, the following is a pseudo-code form of the internal linkage control logic flowchart of the device: Figure 1 This module draws on the linkage control logic in the thermal imaging integrated device to ensure the rapidity and accuracy of the linkage response.

[0095] Alarm module: receives the fire warning signal from the intelligent processing module, timely alarms relevant personnel through sound and light alarm, SMS notification, etc., and stores the warning information and related data locally for subsequent query and analysis.

[0096] ​The present application organically integrates infrared thermal imaging temperature measurement, image processing, device structure design and other technologies, realizes complementary advantages, and improves the comprehensive performance of the device.

[0097] The present application has high-precision temperature measurement and intelligent identification: combined with advanced temperature calibration algorithms and image processing algorithms, it improves the temperature measurement accuracy and the accuracy of fire feature identification, and reduces false positives and omissions.

[0098] Fast linkage response: build an efficient linkage control mechanism, and quickly link with the security monitoring system in fire warning, to provide timely and effective on-site information for fire disposal.

[0099] Adapt to complex environment: optimize the design of the device structure, enhance the heat dissipation and anti-interference ability, and ensure stable and reliable work in the complex environment of industrial places.

[0100] In the thermal imaging acquisition module, other infrared thermal imaging temperature measurement components with the same performance can be used to replace the infrared thermal imaging temperature core of Zhejiang Dahua Technology Co., Ltd., as long as they meet the temperature measurement accuracy and image acquisition requirements, and the temperature and image information acquisition function can be realized.

[0101] In addition to improving the thermal imaging image processing method, other efficient thermal imaging image processing algorithms such as deep learning-based image recognition algorithm can also be used in the image processing algorithm of the intelligent processing module, as long as the accurate identification of fire features can be achieved.

[0102] The communication protocol of the linkage control module can be adjusted according to the type of the actual security monitoring system, and different communication methods such as other protocols of wireless communication or wired communication can be used, as long as the effective linkage with the security monitoring system can be realized, and the purpose of the present application can be achieved.

[0103] The overall working principle of the present application is as follows: Step 1: Synchronous acquisition of dual optical signals The device acquires scene information through two independent optical channels at the same time: 1, visible light channel Use 1 / 2.7 inch, 4 million pixel CMOS image sensor; support low-illumination full-color imaging (0.005 lux@F1.6); provide real color picture for target identification and daily monitoring 2, thermal imaging channel Use non-refrigerated vanadium oxide (VOx) microbolometer focal plane detector; working waveband: 8–14μm (long-wave infrared); sensitivity: NETD≤50mK (can distinguish 0.05℃ small temperature difference); output is a digital radiation value matrix (DN value), each pixel represents a temperature point; features: can penetrate smoke, dust, dark environment, and is not affected by light.

[0104] Second step: Image alignment and fusion processing Due to the different positions of the two lenses, pixel-level image registration and geometric correction are required: use the built-in calibration parameters (focal length, field of view, relative pose) for spatial mapping; achieve accurate superposition of thermal imaging and visible light images (error <1.5 pixels); support multiple fusion modes: Picture-in-picture: small window thermal image superimposed on the visible light main screen; Transparent fusion: thermal image semi-transparent overlay, preserving details; Split-screen comparison: left and right or top and bottom dual-screen display.

[0105] Third step: Temperature calculation and pseudo-color mapping 1. Temperature calculation process Based on the infrared physical model, the device compensates the original radiation data with multiple parameters: The device uses a pre-calibrated LUT (Look-Up Table) to convert DN values to real temperature values, with an accuracy of ±2°C or ±2% (whichever is greater).

[0106] 2. Pseudo-color display mode Support 12 pseudo-color palettes to highlight temperature difference features: white-hot, black-hot; iron red, rainbow, north pole, lava, warning blue, etc. Users can choose the best contrast mode according to the scene (e.g. "rainbow" for fire, "iron red" for power temperature measurement).

[0107] Fourth step: Intelligent analysis and judgment logic The system continuously scans the set area (ROI) and performs the following judgments: 1. Real-time temperature measurement analysis Calculate the highest temperature, lowest temperature, average temperature, and temperature difference (ΔT) within the specified ROI. 2. Alarm triggering conditions (any one of the following conditions will trigger an alarm) Alarm types: Triggering conditions; high temperature out-of-limit alarm: the highest temperature in the ROI is greater than or equal to the user-set threshold (e.g. 70°C); temperature rise rate alarm: detect a temperature rise of ≥10°C / min for 3 consecutive frames; cold and hot spot automatic alarm: automatically identify the hottest and coldest points in the image and mark them, and alarm if they exceed the set range; fire point detection alarm: identify the characteristics of rapid expansion of high-temperature points (area growth >30% / s), suspected of being an open flame; the system supports independent configuration of multiple areas, and each ROI can set different alarm rules.

[0108] 3. Anti-misreporting mechanism Time delay filtering (2-5 second confirmation); area threshold filtering (excluding small-area instantaneous hotspots, such as birds flying or welding sparks); dynamic environmental learning (automatically adapting to day-night temperature changes). Step 5: Linkage Control and Alarm Output Once an anomaly is confirmed, the device will automatically perform the following actions: 1. Local response Activate the audible and visual alarm: The speaker plays a pre-recorded voice message (e.g., "Attention! High temperature detected!"), with a loudness ≥70dB (A-weighted); the red LED flashes (frequency 1Hz); and the infrared fill light is turned on to enhance nighttime imaging quality. 2. Video and Image Saving Automatically saves 10 seconds of video before and after the alarm (H.265 encoded); captures high-definition dual-light images (including thermal pseudo-color annotation); image naming format: DeviceID_YYYYMMDD_HHMMSS.jpg; 3. External Collaboration Relay output: Closes dry contacts to trigger external devices (such as warning lights, fans, and fire extinguishing systems); Alarm input monitoring: Receives external signals (such as smoke alarms) and triggers reverse-linking thermal imaging for focused viewing. 4. Network Upload and Notification RTSP Streaming: Proactively pushes real-time streams to NVR or VMS platforms; ONVIF / GB28181 Registration: Integrates with national standard platforms for unified management; HTTP Event Notification: Sends JSON-formatted alarm messages to specified URLs; FTP Upload: Automatically uploads screenshots and video clips to the server; APP Push: Sends reminders to administrators' mobile phones via the platform; 5. Log Recording Generate structured logs, including: alarm time, device ID, temperature value, ROI area coordinates, and image storage path; support exporting to CSV format for auditing and backtracking; Step 6: Monitoring Center Handling After the alarm information reaches the central platform: the alarm channel video window will automatically pop up; a dual-view comparison (thermal imaging vs. visible light) will be displayed; the event recording will be played and the temperature change curve will be superimposed; the on-duty personnel can remotely review and confirm whether the fire is real; if it is real, the emergency plan will be activated (such as evacuation, power outage, and fire extinguishing).

[0109] like Figure 3 The diagram shown is an internal structural block diagram of the integrated safety and fire protection thermal imaging device. 1. Structural layering (from top to bottom) Structural layering Components included Core functions Optical acquisition layer Thermal imaging lens (Ge germanium glass), VOx detector Receive 8-14 pm infrared radiation, convert to DN value (one temperature point per pixel) Visible light lens (multi-layer coated glass), 4 million CMOS Receive visible light signals, output color images (0.005 lux low light full color) Data processing layer FPGA (field programmable gate array) High speed pre-processing: thermal imaging DN value denoising, bad pixel correction; dual light image registration (error <1.5 pixels) DSP (digital signal processor) Complex algorithms: temperature solution (based on Planck's formula), 12 kinds of pseudo-color mapping, intelligent analysis (fire point identification) Intelligent analysis algorithm module / like Figure 4 The system module connection and functional distribution block diagram shown is as follows. 1. Module connection relationships and functional details The figure shows the module interaction of the system "perception-decision-execution" whole link, highlights the interface type and function parameters, and the key points are as follows: Linkage control module (core hub): Input interface: 1-way dry contact alarm input (connects smoke / door magnet); Output interface: 1-way relay output (trigger alarm light / fire extinguishing system), sound and light alarm drive (control loudspeaker / LED); Dispatching function: video stream scheduling (switch thermal imaging / visible light channel), event log management (generate CSV format log).

[0110] Network and storage module: Protocol support: ONVIF / GB28181 / RTSP (access national standard security platform), HTTP / FTP (alarm information / media file upload); Storage configuration: MicroSD card (maximum 256GB, supports loop recording), RJ45 network port (10 / 100Mbps, supports POE power supply).

[0111] Audio and light supplement module: Audio: built-in MIC (noise reduction), loudspeaker (≥70dB@1m, A-weighted); Light supplement: infrared lamp array (maximum 25 meters light supplement distance, light sensitive automatic opening).

[0112] Shell and protection: Material: metal shell (aluminum alloy, corrosion resistant); Protection level: IP67 (dust + short time immersion); Environmental adaptability: working temperature -30℃~+70℃, installation method supports wall hanging / hanging / universal support.

Claims

1. A fire early detection and warning system integrating infrared thermometry and multiple intelligent algorithms, characterized in that: It includes a thermal imaging acquisition module, an intelligent processing module, a linkage control module, and an alarm module; The thermal imaging acquisition module acquires thermal images by collecting temperature field distribution data and capturing real-time images of the monitored area. The intelligent processing module performs noise reduction, enhancement, and feature extraction on the acquired thermal imaging images to identify flame and high-temperature fire characteristics. The linkage control module is connected to the security monitoring system. When the intelligent processing module determines that there is a fire hazard, the linkage control module quickly links the video monitoring equipment through the preset communication protocol, retrieves the real-time high-definition image of the corresponding area, and synchronously transmits the temperature data and image information to the monitoring center. The alarm module receives fire warning signals from the intelligent processing module and promptly issues alerts to relevant personnel through audible and visual alarms and SMS notifications. At the same time, it stores warning information and related data locally. The modules work together to achieve accurate fire monitoring and rapid response.

2. The fire early detection and warning system integrating infrared thermometry and multiple intelligent algorithms as described in claim 1, characterized in that, The thermal imaging acquisition module includes an uncooled vanadium oxide VOx microbolometer focal plane detector, a visible light imaging module, an optical system, an image processing unit, an audio / video module, an alarm and I / O interface, a network communication module, a storage module, and a supplementary lighting system. The uncooled vanadium oxide VOx microbolometer focal plane detector is connected to the FPGA of the image processing unit via an LVDS high-speed data interface; the visible light imaging module is connected to the FPGA of the image processing unit via a MIPI-CSI image interface; the thermal imaging lens of the optical system is mounted at the front end of the VOx detector, and the visible light lens is mounted at the front end of the CMOS sensor of the visible light imaging module; the image processing unit is connected to the supplementary lighting system, alarm and IO interface via GPIO interface, to the storage module via SDIO interface, and to the network communication module via RMII interface; the microphone of the audio and video module is connected to the DSP of the image processing unit via an I2S audio interface, and the speaker is connected to the DSP via a PWM audio driver interface; the network communication module, storage module, alarm and IO interface all achieve bidirectional data interaction with the image processing unit through their corresponding data interfaces.

3. The integrated fire early detection and warning system combining infrared thermometry and multiple intelligent algorithms as described in claim 2, characterized in that, The uncooled vanadium oxide VOx microbolometer focal plane detector is responsible for converting the infrared radiation signal of the monitored area into digital radiation values, providing raw data for subsequent temperature calculations. The uncooled vanadium oxide VOx microbolometer focal plane detector is directly connected to the FPGA of the image processing unit via the LVDS high-speed data interface: it transmits the DN value matrix to the FPGA in real time and receives parameter configuration commands sent by the FPGA. The visible light imaging module acquires color visible light images of the monitoring area for visual confirmation of the fire situation; the visible light imaging module is connected to the FPGA of the image processing unit through the MIPI-CSI image interface: it transmits RGB color image data to the FPGA; and it receives control signals sent by the FPGA to ensure image quality under different lighting conditions. The optical system is a dual-lens system, including a thermal imaging lens and a visible light lens, providing a precise optical path for infrared radiation and visible light signals. The thermal imaging lens is directly mounted on the front end of the uncooled VOx detector via a mechanical bracket, with the lens optical axis precisely aligned with the photosensitive surface of the detector. The visible light lens is mounted on the front end of the CMOS sensor of the visible light imaging module via a bracket of the same specification, and the field of view of the lens is matched with that of the thermal imaging lens. The image processing unit includes a DSP and an FPGA architecture, and the FPGA is responsible for high-speed data preprocessing and pixel-level registration of dual-light images. The preprocessing includes thermal imaging DN value noise reduction, bad pixel correction, and visible light image distortion correction; The DSP is responsible for complex algorithm calculations and generates linkage control commands. The FPGA end of the image processing unit is connected to the VOx detector via the LVDS interface and the visible light imaging module via the MIPI-CSI interface to receive raw data; the image processing unit realizes bidirectional data transmission between the FPGA and the DSP via the AXI high-speed internal bus; the DSP of the image processing unit is connected to the supplementary lighting system, alarm and IO interface via the GPIO interface, to the storage module via the SDIO interface, and to the network communication module via the RMII interface to output processing results and control commands. The audio and video module includes a built-in microphone and a high-volume speaker, enabling on-site interaction and local early warning: The microphone supports two-way voice communication, directly capturing and transmitting voice signals; The speaker plays a pre-recorded alarm tone to ensure that on-site personnel can quickly detect the warning. The microphone is connected to the DSP of the image processing unit via an I2S audio interface, and the acquired voice signal is transmitted to the DSP for encoding. The speaker is connected to the DSP via a PWM audio driver interface to receive audio playback commands sent by the DSP. The alarm and I / O interface includes one alarm input and one relay output. The alarm and I / O interface is used to expand the system's external device linkage capabilities, enabling bidirectional signal interaction between the system and peripherals. The alarm receives dry contact signals from external security devices, triggering targeted monitoring by the system; the relay outputs dry contact control signals, linking with external actuators. The alarm and I / O interface is directly connected to the DSP of the image processing unit via the GPIO level interface: it transmits alarm input signals from external devices to the DSP; it receives relay closing / opening commands sent by the DSP to control the start and stop of external devices; The network communication module integrates local and remote data transmission and power supply: the network communication module is powered via PoE; the network communication module is connected to the DSP end of the image processing unit via an RMII Ethernet interface: it receives video streams and alarm data packaged by the DSP and transmits them to the remote platform. Receive control commands sent from a remote platform and forward them to the DSP for execution; The storage module enables local data retention, facilitating offline querying and evidence tracing. The storage module connects to the DSP of the image processing unit via an SDIO data interface, receiving encoded video streams, dual-light capture images, and structured logs from the DSP; and responding to data read commands from the DSP. The supplementary lighting system has a built-in high-efficiency infrared lamp array; it adopts a circular uniform arrangement, with 8 940nm infrared lamps distributed around the visible light lens, and the supplementary lighting angle is matched with the field of view of the visible light lens (error ±5°) to ensure that there are no blind spots in the supplementary lighting coverage; The supplementary lighting system is connected to the DSP of the image processing unit via a GPIO control interface: it receives supplementary lighting start / stop commands sent by the DSP, and the trigger logic is: the DSP automatically determines the brightness based on the brightness detection value of the visible light module, and automatically turns on when the average brightness of the visible light image is below 5 lux and automatically turns off when it is above 15 lux; or it receives a manual start command from a remote platform.

4. The integrated fire early detection and warning system combining infrared thermometry and multiple intelligent algorithms as described in claim 3, characterized in that, The intelligent processing module employs a three-stage processing flow: wavelet threshold denoising → Laplacian edge enhancement → convolutional neural network (CNN) feature extraction. Planck's formula provides precise temperature data support for the three-stage processing flow; The actual temperature measurement model is used to correct the temperature data and ensure that the temperature information input into the three-level processing flow is accurate. The image features processed by the three-level processing flow are combined with temperature data to further improve the accuracy of fire point feature recognition. Thermal imaging equipment calculates the surface temperature of a target object by detecting the infrared energy radiated by the target object. Its basic physical basis is Planck's radiation law and Stefan-Boltzmann's law. Planck's formula is the foundation of temperature calculation in DSP, and the actual temperature measurement model is used to optimize the temperature measurement results. Together, they provide reliable temperature data input for 12 pseudo-color mappings, ROI area temperature analysis, and fire point feature recognition.

1. Relationship between radiation intensity and temperature (Planck's formula): L: Spectral radiance λ: wavelength T: Absolute temperature of an object (unit: K) h: Planck's constant c: speed of light k: Boltzmann constant The device collects the radiation value of each pixel in the scene through sensors and compensates for it by combining parameters such as emissivity (ε), ambient temperature, relative humidity, and distance.

2. Actual temperature measurement model: By combining the temperature calibration algorithm in infrared thermal imaging temperature measurement methods, the temperature of the monitored area is analyzed in real time and anomalies are identified.

5. The integrated fire early detection and warning system combining infrared thermometry and multiple intelligent algorithms as described in claim 4, characterized in that, Specific steps of the DSP algorithm:

1. Temperature calculation steps: Obtain the DN value matrix of the VOx detector, substitute it into Planck's formula to calculate the spectral radiance L; input compensation parameters such as emissivity and ambient temperature, and calculate the actual surface temperature of the object through the actual temperature measurement model, with an accuracy of ±2℃ or ±2% (take the larger value). 2.12 pseudo-color mapping steps: Divide the temperature data into intervals and assign corresponding pseudo-color colors to each interval; match the pseudo-color with thermal imaging pixels to generate a pseudo-color thermal image that highlights temperature differences; 3. ROI area temperature analysis steps: Extract pixel temperature data of ROI area, calculate the highest and lowest temperature parameters, and compare them with the preset alarm threshold in real time; 4. Fire point feature identification steps: Extract the dynamic contour and temperature gradient of suspected fire points, combine the temperature analysis results, eliminate interference through the false alarm prevention mechanism, and confirm the real fire points.

6. The integrated fire early detection and warning system combining infrared thermometry and multiple intelligent algorithms as described in claim 5, characterized in that, The thermal imaging analysis and judgment operation steps are as follows: Step 1: Securely install the device at a high point in the monitoring area, away from high-temperature interference sources; connect the power supply and log in to the Web management interface or NVR platform after the network is online; Step 2: Set the time, time zone, and video bitrate in the system; enable the thermal imaging channel and select pseudo-color mode; Calibration environment parameters: Input ambient temperature; set relative humidity and observation distance; adjust emissivity; Step 3: Enter the "Temperature Analysis" page, draw one or more temperature measurement ROI regions; set alarm conditions: Enable "High Temperature Alarm": Triggered when the highest temperature exceeds 70°C; Enable "Temperature Rise Rate Alarm": Temperature rises by more than 10°C within 1 minute; Enable "Automatic Hot / Cold Spot Tracking": The system automatically identifies the hottest / coldest spot in the image and issues an alarm; Set alarm delay: to avoid momentary false alarms; Step 4: Bind alarm output: Trigger the relay to close, link the warning light, exhaust fan or fire alarm control panel, and enable the audible and visual alarm: The local speaker emits a high-decibel alarm sound; Configure upload platform: Alarm information is pushed to the central management platform or mobile APP notification; Step 5: Observe the color changes of the thermal imaging screen in real time, focusing on the red / white areas; when an alarm occurs: check the visible light screen to confirm whether it is a real fire; play back historical video to review the temperature change trend; manually start recording or take pictures to save evidence; if it is confirmed to be a real fire, immediately notify on-site personnel to handle it or activate the automatic fire extinguishing system. Step 6: Clean the thermal imaging lens and visible light lens quarterly. Use a dedicated blower to remove dust and then gently wipe with lens paper. Regularly check the accuracy of temperature measurement. Update the firmware to get the latest features and security patches.

7. The integrated fire early detection and warning system combining infrared thermometry and multiple intelligent algorithms as described in claim 6, characterized in that, The linkage control module is responsible for automatically triggering the video surveillance system to respond when thermal imaging detects an anomaly and uploading multi-dimensional data to the monitoring center; when thermal imaging detects a fire hazard, the system will automatically retrieve the high-definition visible light image of the corresponding area and overlay it with the thermal imaging image to achieve dual-mode evidence preservation and remote confirmation.

8. The integrated fire early detection and warning system combining infrared thermometry and multiple intelligent algorithms as described in claim 7, characterized in that, The specific steps are as follows: Step 1: The intelligent processing module detects any of the following situations: the highest temperature in the area is greater than or equal to the threshold temperature rise rate exceeds the set value; fire point characteristics are automatically identified; the system generates a "fire warning" event and enters the linkage process; Step 2: The device has a built-in dual-channel video acquisition system that does not require external requests. The main control chip caches the most recent 10 seconds of video frames in real time; after triggering, it immediately saves a 20-second video clip, including the 5 seconds before and 15 seconds after the trigger; and simultaneously captures high-resolution images. Step 3: Data packet content: Current thermal image; Synchronized visible light image; Maximum / minimum temperature value, alarm type, device ID, geographical location; Video clip; Push streaming to NVR / VMS platforms; actively push real-time streams using RTSP or RTMP protocols; support SIP registration and access to national standard platforms. HTTP / HTTPS event notifications; send JSON-formatted alarm messages to preset URLs, including image links and video paths; Upload media files via FTP; automatically upload screenshots and videos to the specified server directory and push them to mobile devices; push alarm information and thumbnails via the platform APP. Step 4: After receiving the alarm, the central platform automatically pops up the corresponding channel video window; displays a dual-view comparison; plays the event recording and marks the temperature change curve.

Images