Outdoor camera with fireproof function

Abstract

The invention discloses an outdoor camera with a fireproof function. The outdoor camera is characterized in that temperature characteristics of objects made of different materials in a detection area are calibrated, and compensation parameters are generated; carrying out adaptive fusion on the two types of image input data according to the compensation parameters so as to stabilize the image contrast under the illumination change; positioning the position of a suspected fire source in the fused image by using the adjusted fire initial feature extraction rule; and the moving heat source is monitored in real time and false positive false alarms are eliminated based on dynamic parameter adjustment of the time sequence abnormal temperature rise early warning model. The system has the advantages that the problem of temperature sensing errors of objects made of different materials in a detection area is solved, the fire monitoring sensitivity is improved, and the problem of camera communication delay is solved.

Description

Technical Field

[0001] The invention relates to the technical field of fire monitoring, in particular to an outdoor camera with a fire prevention function. Background Art

[0002] In modern society, with the acceleration of urbanization and the continuous expansion of various industrial activities, fire hazards are increasing, and the loss of life and property caused by fire accidents is becoming more and more serious. Traditional fire monitoring methods, such as manual patrols and smoke alarms, have problems such as low efficiency, poor real-time performance, and high false alarm rate when facing the monitoring needs of large areas and complex environments.

[0003] In this context, fire monitoring technology based on the linkage of infrared and visible light cameras came into being. This technology combines the advantages of infrared cameras that can capture high-temperature heat sources and smoke information, and visible light cameras that provide clear visual images, to achieve all-weather, multi-angle real-time fire monitoring. Through the fusion analysis of the two types of image data, it can more accurately identify the fire situation, greatly improving the ability and scope of fire monitoring.

[0004] However, this technology still faces many challenges in practical application: Problems with the linkage mechanism between infrared and visible light cameras: Objects of different materials have very different reflection and absorption characteristics for infrared and visible light, which leads to errors in the temperature perception of the camera. For example, in complex industrial scenes, multiple materials such as metal, plastic, and wood coexist. The existing linkage mechanism is difficult to accurately compensate for this temperature perception difference, which can easily cause misreading and affect the accuracy of fire monitoring.

[0005] Image fusion algorithm problems: The lighting conditions in the actual environment are complex and changeable, such as strong direct sunlight during the day, light interference at night, and lighting changes in different weather conditions. The existing image fusion algorithm cannot dynamically adapt to these changes in the lighting environment. When the light intensity changes suddenly, the image contrast is unstable, key information is easily lost, and the accuracy of fire detection is reduced.

[0006] Problems with early fire feature extraction rules: In the early stages of a fire, the characteristics of flames and smoke are not obvious and are easily confused with other similar phenomena. The current rules for early fire feature identification are not perfect. In non-fire conditions, the system is prone to false alarms, frequently disrupting normal work order; at the same time, due to the limitations of the rules, it may lead to a slow response to real fires and delay the best time to put out fires.

[0007] Problems with abnormal temperature rise warning models: In some scenarios, heat sources may be in a mobile state, such as vehicle fires, spreading fire sources in forest fires, etc. The existing abnormal temperature rise warning models based on time series cannot quickly and accurately adjust parameters when facing mobile heat sources, and there is a problem of monitoring delay, which cannot detect potential fire risks in a timely manner.

[0008] Communication protocol issues: In a multi-camera system, due to the different models and brands of cameras and the complexity of the actual network environment, it is crucial to select the appropriate communication protocol and optimize the configuration. At present, the existing communication protocol has large synchronization errors during image transmission, which is prone to data loss and delay. It cannot guarantee the efficiency and stability of the collaborative work between multiple cameras, affecting the overall performance of the fire monitoring system. Summary of the invention

[0009] The object of the present invention is to provide an outdoor fire monitoring method to solve the problems raised in the above background technology.

[0010] To achieve the above object, the present invention provides the following technical solutions: In a first aspect, the present invention provides an outdoor fire monitoring method, comprising the following steps: By calibrating the temperature characteristics of objects of different materials in the detection area and generating compensation parameters; Adaptively fusing the two image input data according to the compensation parameters to stabilize the image contrast under illumination changes; The adjusted early-stage fire feature extraction rules are used to locate the suspected fire source in the fused image; Based on the dynamic parameter adjustment of the time series abnormal temperature rise warning model, real-time monitoring of mobile heat sources can be achieved and false positive alarms can be eliminated.

[0011] Preferably, the step of adjusting the parameters of the time series abnormal temperature rise warning model to achieve real-time monitoring and eliminate false alarms further includes: Time delay correction is performed based on the heat source movement speed and historical data: the predicted movement distance D is calculated by introducing the time difference information Δt of the historical image frame and the relative position deviation Δx of the current detection frame p ,formula , where V represents the moving speed of the heat source; A multi-level filter is constructed using machine learning algorithms to initiate a more fine-grained dynamic monitoring mechanism after the initial identification of a suspected fire source to eliminate false positives caused by non-target objects; Real-time optimization of temperature change thresholds: When a significant temperature rise occurs at the same location in N consecutive images, a fire is confirmed; Fusion of data streams captured synchronously by multiple cameras, ensuring consistency and timeliness in data transmission.

[0012] Preferably, the method of applying the adjusted early fire feature extraction rules to locate the suspected fire source is further concretized into four steps as follows: Select appropriate morphological operators to enhance the unique features of the fire source area and suppress the influence of noise and other non-important features; Use color models combined with texture analysis techniques to help identify possible early signs of fire; Filter out a set of potential target objects based on specific conditions. Right now , where T r is the minimum temperature difference reference value, T u is the upper limit of the maximum temperature difference, A min Indicates the lowest limit of the target area, A r , A u are the upper and lower ranges of the aspect ratio, respectively. These values ​​are pre-set by experiments or obtained by adaptive adjustment.

[0013] Preferably, the adaptive fusion step is improved to solve the contrast distortion problem caused by illumination, and includes the following process: Construct a series of parameters for evaluating the intensity of light sources, such as illumination L(x,y), color shift ΔRGB, and contrast sensitivity function CSF, to evaluate the degree of light fluctuation and the resulting image quality issues; When illuminance L(x,y)>L max Or ΔRGB>ΔR max When the image is processed, a special image enhancement processing path is activated to ensure stable and reliable image quality required by subsequent steps.

[0014] Preferably, the following key steps are added by calibrating the temperature characteristics of objects of different materials in the detection area and generating compensation parameters: The system automatically learns typical reflection coefficient curves of various types and builds a database based on them to guide the classification and labeling of new material samples; Implement a high-resolution, multi-angle shooting plan to generate enough samples for subsequent analysis and mining, especially for subtle color differences between different types of building materials that are difficult to distinguish.

[0015] In view of the fact that the surface emission characteristics show different degrees of change with the angle, the view angle adaptive algorithm is used to compensate for the result deviation caused by the tilt of the field of view; A comprehensive characterization system including but not limited to conductive materials and thermal insulation coatings has been established to ensure comprehensive coverage of all treated materials and guarantee high credibility and accuracy of measurement results.

[0016] In a second aspect, the present invention provides an outdoor fire monitoring device for implementing the method described in any one of the above embodiments, the device comprising: A detection module, which calibrates the temperature characteristics of objects of different materials in the detection area and generates compensation parameters; An input module, wherein the input module adaptively fuses two types of image input data according to the compensation parameter to stabilize image contrast under illumination changes; A positioning module, wherein the positioning module uses the adjusted early-stage fire feature extraction rule to locate the suspected fire source in the fused image; The early warning module realizes real-time monitoring of mobile heat sources and eliminates false positive alarms based on dynamic parameter adjustment of a time series abnormal temperature rise early warning model.

[0017] In a third aspect, the present invention provides an electronic device comprising at least one processor, wherein the processor is communicatively connected to at least one memory, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute any of the methods described in the above embodiments.

[0018] In a fourth aspect, the present invention provides an outdoor camera with a fire prevention function, wherein the outdoor camera is provided with an electronic device as described in the above embodiment.

[0019] Compared with the prior art, the beneficial effects of the present invention are as follows: the present invention calibrates the temperature characteristics of objects of different materials in the detection area and generates compensation parameters; adaptively fuses the two image input data according to the compensation parameters to stabilize the image contrast under illumination changes; uses the adjusted early fire feature extraction rules to locate the suspected fire source in the fused image; and realizes real-time monitoring of mobile heat sources and eliminates false positive alarms based on dynamic parameter adjustment of the time series abnormal temperature rise warning model. The solutions of the embodiments of the present disclosure can solve the following problems: 1. How to optimize the linkage mechanism between infrared and visible light cameras to solve the problem of temperature sensing error of objects of different materials in the detection area.

[0020] 2. How to regulate the fusion algorithm of two image input data to solve the problem of unstable image contrast caused by lighting changes.

[0021] 3. How to adjust the rules for extracting features in the early stages of a fire to solve the problem of frequent false positive alarms.

[0022] 4. How to adjust the parameters of the abnormal temperature rise warning model based on time series to solve the monitoring delay problem caused by the movement of heat sources.

[0023] 5. How to regulate the selection and configuration of communication protocols between cameras to solve the problem of large image transmission synchronization errors in multi-camera systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 It is a schematic diagram of the method flow of the present invention; Figure 2 The present invention is based on the step of adjusting the parameters of the time series abnormal temperature rise warning model to achieve real-time monitoring and eliminate false alarms; Figure 3 The steps of extracting the characteristics of the initial stage of fire adjusted for the application of the present invention to locate the suspected fire source are embodied in a four-step process; Figure 4 An improved process for the adaptive fusion step of the present invention is provided to solve the contrast distortion problem caused by illumination; Figure 5 A module diagram of an outdoor fire monitoring device provided by the present invention; Figure 6 A module diagram of an electronic device provided by the present invention; Figure 7 This is a schematic diagram of the network connection of the outdoor camera of the present invention. DETAILED DESCRIPTION

[0025] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0026] In one embodiment of the present invention, referring to the attached Figure 1 This embodiment provides a fire monitoring method. It includes: S101, calibrating the temperature characteristics of objects of different materials in the detection area and generating compensation parameters; S102, adaptively fusing two types of image input data according to the compensation parameter to stabilize image contrast under illumination changes; S103, using the adjusted early stage fire feature extraction rule to locate the suspected fire source in the fused image; S104. Based on the dynamic parameter adjustment of the time series abnormal temperature rise warning model, the mobile heat source is monitored in real time and false positive alarms are eliminated.

[0027] By calibrating the temperature characteristics of objects of different materials in the detection area and generating compensation parameters, the linkage mechanism between infrared and visible light cameras is optimized, solving the temperature sensing error caused by objects of different materials in the detection area. First, before implementation, a detailed investigation of the monitoring area is required, and the temperature curves of different types of materials and their temperature curves under standard environments must be recorded.

[0028] Then, according to the compensation parameter adjustment method obtained above, adaptive image processing fusion technology is implemented for the two image input sources, namely the infrared imaging view and the optical visual shooting picture, to deal with the problem of uneven image quality caused by the drastic changes in light under outdoor conditions. Specifically, after receiving the original material transmitted from the front end, the RGB color mode is converted into the brightness Y component, the chromaticity U component, and the V component, which are processed separately and different filter operations are applied to each component before being recombined and output as the final version. For example, in a square monitoring environment with strong sunlight in the afternoon, if the visible light camera is relied on alone, the key contour around the flame may be overexposed and lost, and it is difficult to accurately determine the shape of smoke and other complex scenes by relying solely on the thermal sensitive information of a single dimension. At this time, after a specific formula is used to calculate and mix the advantages of the two materials, a composite format with bright and prominent fire position and clear background detail levels can be formed, which provides a solid and reliable basis for the next step of identification, reduces the misreading rate caused by environmental fluctuations, and ensures the overall style of painting is coordinated and unified.

[0029] In addition, in order to adjust the rules for early fire morphological recognition, a machine training method is used to feed the computer system with a large number of teaching resource libraries containing real cases in advance, including flame image samples at various angles and distances, and add data labels that are manually annotated to distinguish true and false targets, so as to build a deep neural network to complete the initial model creation, and then continuously use the latest feedback results to adjust the threshold setting parameters to adapt to the changing trend of actual application scenarios, reduce the number of unnecessary alarms, and improve work efficiency. For example, in a warehouse cargo storage area, people often bring heating appliances to work or sparks come out of the exhaust pipe of the engine of a vehicle, which can easily trigger the wrong judgment of the traditional fixed indicator system. The new solution can flexibly configure the judgment conditions in combination with environmental information, allowing a certain error tolerance range. Only when it meets the vast majority of characteristic factor combinations will it be determined that there is a real risk of fire and deflagration, and then issue an alarm command to notify relevant personnel to deal with it urgently, avoiding frequent interruptions to the normal work rhythm, and also enhancing the accuracy of the early warning.

[0030] Next, regarding the establishment of an abnormal temperature rise pre-alarm scheme based on the statistical laws of time series, considering that in some occasions there are fast-moving objects passing through the lens field of view, it is necessary to ensure real-time dynamic capture and tracking functions. For this purpose, a multi-dimensional feature point extraction and tracking algorithm is introduced to regularly capture key frame images and compare the similarity between adjacent interval frames to determine whether there are signs of sudden increase in hot spots. At the same time, considering the delay and lag that may be caused by the transfer of heat sources along the path, a buffer time slice window function is specially set up to warn the relevant person in charge in advance to pay attention to potential dangers and promptly intervene in the investigation and verification of the on-site conditions to prevent missing the best time to put out the fire. For example, if a suspected smoke point signal is found near a forest fire lookout tower but it moves quickly, it is likely that a wild beast is running by and radiating heat from its body, causing a short-term local hot spot. At this time, it is necessary to use historical trajectory modeling to predict its movement direction trend and combine geographic location analysis to determine the final confirmation plan to effectively eliminate non-hazardous sources and reduce the spread of panic.

[0031] Finally, through the work of selecting and matching the communication protocols between cameras, given that modern large-scale distributed networking architectures often have multiple models and specifications of products with different compatible docking requirements, in order to ensure the coordinated operation of multiple sets of devices, it is necessary to select stable and efficient communication means and scientifically and reasonably set various initialization configuration parameters such as IP address, port number, authentication key and other elements to ensure smooth information transmission. In a large cross-regional enterprise park, hundreds or even thousands of monitoring equipment produced by different manufacturers are installed. If a set of unified and standardized technical guidelines is not formulated, it is very likely that there will be mutual interference and conflicts, or even offline, resulting in serious consequences such as video loss and fragmentation. The priority recommendation of the secure connection mechanism under the TCP / IP protocol family is adopted, and the performance indicators of various links are centrally regulated through a dedicated management background. Regular maintenance and update of firmware software repair patches ensure that the system is always in the best working state to meet the 7×24-hour continuous operation requirements, fundamentally overcome the challenges caused by hardware and software differences, and achieve the ideal effect of seamless docking and interconnection, thereby greatly improving the efficiency of fire emergency response and laying a more solid intelligent technology foundation for protecting people's lives and property.

[0032] In one embodiment, a monitoring system consisting of at least one infrared camera and one visible light camera is set as the basic architecture, and the collaborative operation logic between the two cameras is set by programming. It is assumed that there is a threshold formula T=I*R+B to determine the triggering alarm mechanism. In this formula, I represents the effective fire probability extracted from the infrared image; R represents the weighted proportional factor, which is a real number between 0 and 1, and good recognition accuracy can be obtained when 0.8 is optimized; B is the background error coefficient, which ranges from 0 to increasing integers, and the preferred value is usually 5.

[0033] Then the operation process and application scenarios are refined. Specifically, the information transmission method and timing control between each hardware unit will be described. For example, the method mentioned in the above example will also include how to process image data and perform pattern matching judgment, that is, when the temperature is detected to exceed the preset limit or a specific color change trend is found, the corresponding alarm information will be generated immediately to ensure a rapid response to abnormal situations, reduce false alarm rates and improve the safety of the entire system.

[0034] In one embodiment of the present invention, referring to the attached Figure 2 , the present invention adjusts the parameters of the time series abnormal temperature rise warning model to achieve real-time monitoring and eliminate false alarms.

[0035] S201, perform time delay correction based on the heat source moving speed and historical data. In this operation, the algorithm introduces the time difference information Δt between historical image frames and the relative position deviation Δx between adjacent frames during the detection process to calculate the heat source prediction displacement D p . formula Indicates that the predicted distance is equal to the heat source speed multiplied by the time difference. Here, Δt can define the range according to the system capture frequency; the optimal value depends on the object movement speed and the camera refresh rate in the specific scene; V represents the expected heat source movement speed observed in real time or obtained from the previous learning process. In practical applications, if a heat source shows a trend of moving from the upper left corner to the lower right corner on both the visible light and infrared cameras, after determining the movement rate, the specific area where the object will reach after a period of time can be predicted.

[0036] S202. Use machine learning to build a multi-level filter to initiate a more sophisticated and continuous monitoring review process to identify the authenticity of potential alarms and exclude false alarm events caused by targets that do not involve fire elements. For example, when the initial round of screening identifies a high-temperature area that may be the location of the fire, the next level of verification mechanism is triggered, which can analyze other data characteristics that may cause errors, such as smoke diffusion and light intensity, and adjust the threshold or take additional verification actions accordingly until it is ensured that the detected target is indeed the fire point.

[0037] S203, dynamically fine-tune the temperature change threshold so that the same pixel area in N consecutive images shows a significant temperature increase ( ) can be considered as a fire. Here, N refers to the number of images captured continuously to confirm the fire. Its value should be sufficient to filter out transient heat fluctuations but not increase the probability of delayed reaction too much. Among them, ΔT represents the temperature change, T th is the threshold value of temperature change, P(ΔT>T c |Data from fire)≥P cThe data are from fire conditions, ΔT is greater than T c The probability of c is another temperature reference value, P C is the set probability threshold, and N is the number of consecutive images used to determine the fire. For example, a fixed interval is set to extract a number of image frames for evaluation, and at least three consecutive time points are required to reach a predetermined temperature increment, and the probability of a fire under the conditions estimated by Bayesian theory is greater than a given credible interval, then an abnormality is formally reported.

[0038] S204. Fusion of information flows provided by various video sensor units ensures that data interaction is stable, reliable and timely, which is an indispensable part of achieving overall monitoring efficiency. Specifically, for a network system with multiple complementary viewing angle cameras, the synchronously acquired data must be processed to keep the records between different sources coordinated and unified, that is, to eliminate the time skew problem to ensure that no important fragment is lost, thereby maintaining effective full-process tracking capabilities. Assuming that there is a set of deployed panoramic monitoring stations responsible for covering every corner of the building, the video stream transmission is guaranteed to be free of lag or replay by precisely adjusting the transmission rate of each camera and protocol adaptation.

[0039] In one embodiment of the present invention, referring to the attached Figure 3 ,The steps of applying the adjusted early fire feature extraction rules to locate the suspected fire source are embodied in four steps as follows.

[0040] S301. Select appropriate morphological operators to enhance the unique features of the fire source area and suppress noise. Specifically, in the images captured by infrared and visible light cameras, there are many interferences or non-flame elements. The use of morphological operators is to highlight the target area through opening operations, closing operations or morphological gradients, remove small but not global analysis results in the image, and smooth the edges to make the boundaries of the suspected fire area clear, which is convenient for subsequent accurate processing and identification. For example, when detecting fire hazards in forests, the closing operation of rectangular kernel structure elements can be used to merge the shadow parts caused by the shooting angle or the small blocks separated by dust coverage.

[0041] S302. Combine color model and texture analysis technology to identify potential early signs of flames. Utilize the data contrast differences and spectral characteristics captured by cameras of different wavelengths to comprehensively judge whether the suspected area has typical color attributes of flames from three aspects: hue, brightness and saturation; simultaneously consider texture characteristics such as the strength of granularity and its arrangement rules to judge the possibility of dynamic flame propagation. In one embodiment, the infrared camera can sense the temperature gradient change to form a thermal image, while the conventional camera obtains the visual picture. By combining the advantages of the two, a dual threshold algorithm is established to evaluate whether the local high temperature point has the appearance of a flame and whether its surface fiber texture flickers and fluctuates regularly.

[0042] S303: Screening a set of potential target objects as key inspection objects according to specific conditions Here T r and T u They correspond to the lowest and highest relative temperature difference standards in the setting (such as T r = 5K, T u = 50K), used to distinguish normal temperature changes and abnormal heating in cold and warm environments; A min Indicates the minimum physical plane space measurement that each detection unit must cover (e.g. 2cm 2 ), to prevent the misjudgment of small smoke sources or other initial fires that are difficult to detect but may spread and expand; A r and A u It is the length-to-width ratio limit of the shape, that is, 1:3 to 3:1 is more appropriate. Most of the objects within this range are strip-shaped objects or concentrated burning objects. The purpose of this link is to eliminate irrelevant interference, focus on the key parts that may really become the fire point, and improve the accuracy and timeliness of the alarm.

[0043] S304. Reduce background complexity based on the background modeling principle. By building a stable background template and continuously updating and iterating, the difference between the previous and next frames is quickly compared and analyzed after each new image is acquired, so as to achieve a subtraction effect and highlight the newly added abnormal moving highlights. Specifically for forest field scene monitoring projects, common conditions such as the interlaced light and shadow effects caused by uneven light intensity with the change of seasons and the swaying of vegetation during the alternation of day and night will be filtered out, while hot particles that break in from outside, small volcanic bodies that are not hidden, etc. will appear on the final generated difference image, thereby effectively reducing the probability of false triggering and improving the effectiveness and reliability of the early warning mechanism.

[0044] In one embodiment of the present invention, referring to the attached Figure 4 The adaptive fusion step of the present invention is improved to solve the contrast distortion problem caused by illumination, and includes the following process.

[0045] S401. Construct a series of parameters for evaluating the strength of the light source, including illumination L(x, y), color offset ΔRGB and contrast sensitivity function CSF, to determine the degree of negative impact of the lighting state in the current environment on the image. Specifically, L(x, y) represents the spatial brightness distribution value measured at the point (x, y); the color offset ΔRGB is used to indicate the degree of difference between the three primary colors of red, green and blue. When the value is too large, it implies that there is an abnormal color tone caused by complex lighting conditions such as strong light or backlight; and the contrast sensitivity function CSF reflects the different perception characteristics of the human visual system for contrast, and takes values ​​between 0 and 1. This series of evaluation parameters provides a theoretical basis for adjustment. For example, in one embodiment, assuming that a fire scene has a strong sense of contrast due to uneven external lighting or natural light directly entering the range of the surveillance camera, then applying the above parameters to this situation can accurately identify the source of the problem.

[0046] S402: When the illuminance L(x, y) is detected to exceed the preset maximum value L max When ΔRGB exceeds the critical upper limit ΔR max , a special image enhancement program path is enabled to ensure a stable and reliable supply of raw data required for subsequent links. This limit is set because the image quality will be significantly reduced after exceeding the limit level, which is not conducive to analysis and interpretation, and it is also beneficial to protect sensitive electronic components from damage due to excessive light. Specifically, when setting reasonable thresholds, the sensitivity of the equipment and the requirements of the actual application scenarios are taken into consideration, and these key values ​​are not selected arbitrarily. To illustrate, in a case where there is smoke obstruction or bright flames that interfere with the coordinated operation of the infrared and visible light camera components, once the trigger conditions are met, a specific algorithm is immediately called to improve the stability of the quality of the information content obtained by the acquisition port.

[0047] S403. A novel method combining local gradient estimation and global histogram equalization technology is applied to carry out image enhancement. This solution helps to solve the problem of unclear light and dark levels caused by inconsistent lighting. The so-called calculation of slope changes based on local features can highlight the edge information of the object contour and reduce the visual confusion caused by the shadow part; as for the latter, the dynamic range is wider by rearranging the overall grayscale spectrum, so as to better reflect the details of the target's true appearance. For example, the temperature field distribution in the surrounding environment of the high heat radiation area emitted by a burning object is accurately reproduced, which not only improves the clarity but also enhances the interpretability and practicality.

[0048] S404. The optimized new image is softer and has a smoother transition, which makes it easier for the machine to automatically identify and locate the position of potential dangerous elements such as flames, thick smoke or other abnormally high temperature points. On this basis, continuing to perform the next stage of pattern identification tasks will greatly improve the accuracy and reliability of the final monitoring system. In one embodiment, this means that it is easier to lock the location features of those suspected initial fire sources from the video clip sequence that has been processed by adaptive correction, thereby gaining more time resources for the formulation of prevention and control strategies.

[0049] In one embodiment of the present invention, the present invention adds the following key steps by calibrating the temperature characteristics of objects of different materials in the detection area and generating compensation parameters: The system automatically learns the typical reflectance curves of various types of materials and builds a database based on them to guide the classification and labeling of new samples. In this step, different materials produce characteristic reflectances according to their respective spectral characteristics. In the scene where infrared and visible light cameras are linked, these reflectances help determine the identities of various materials in a specific fire environment.

[0050] In order to achieve more accurate data collection, a high-resolution multi-angle shooting plan was implemented to generate enough images for subsequent in-depth research. Since changes in viewing angles may affect the true performance of imaging results, this step is intended to ensure that each frame of the image obtained restores the details of the real object as much as possible. High-quality images are indispensable, especially when distinguishing between building materials with slight visual differences but very different essential properties. In one embodiment, this method was used to accurately distinguish between ordinary concrete walls and exterior walls with embedded insulation materials. By shooting the two at multiple different angles, subtle texture differences were obtained, and combined with existing data analysis, the boundary position of the two building components was successfully identified, improving the detection accuracy.

[0051] Since the surface emission characteristics vary with the viewing angle, a viewing angle adaptive algorithm is introduced to compensate for the result deviation caused by the tilt of the camera field of view. This process ensures that even when the camera cannot always face the target object, the recorded data can still accurately reflect the material properties, thereby reducing the incidence of misjudgment. Assuming that the parameter θ represents the angle value of the current viewing angle relative to the standard viewing direction, the calculation formula can be (where D represents the uncorrected measurement value or signal strength), where n is adjusted according to the specific material, and ideally is between 2 and 4 to ensure the rationality and stability of the corrected D. The reason for setting this formula is to achieve smooth correction of the error caused by the angle through reasonable trigonometric function transformation while maintaining the original physical meaning.

[0052] Finally, a comprehensive characterization system that fully covers various properties such as conductivity and insulation was established. This not only allows for a detailed division of the characteristics of each type of material, but also helps to build a more complete sensing framework, so that all monitored objects can obtain highly trusted and stable reading feedback. For example, in cases involving fires in electrical equipment, this system effectively distinguishes the changing trends of temperature and humidity states in high-temperature areas around metal lines and areas with insulation measures, greatly improving early warning efficiency and response capabilities. In addition, it also shows superior performance in non-conductive electromagnetic environments, such as the ability to efficiently identify abnormal surface temperatures of walls containing special component coatings.

[0053] In one embodiment of the present invention, the selection and configuration measures of the communication protocol of the present invention are strengthened and improved: first, a communication protocol that best matches the actual network conditions is selected from a series of predefined communication protocols. The specific selection criteria are not limited to stability, but also include factors such as bandwidth utilization efficiency. These evaluation factors ensure that the selected protocol can meet the real-time requirements and maintain efficient data transmission in unstable or low-bandwidth environments.

[0054] For example, considering that the application scenario of the fire monitoring method requires infrared and visible light cameras to work simultaneously, the system is required to have the flexibility to handle multiple data types and strong environmental adaptability. To this end, in one embodiment, the system dynamically adjusts to TCP or UDP protocol according to the quality of the current wireless network or wired connection. If the Wi-Fi signal in the on-site deployment area is poor, TCP with better reliability is preferred; if it is running in a high-speed Ethernet, it can switch to UDP with high efficiency but less fault tolerance to optimize the upload efficiency of video streams and other sensor information.

[0055] In one embodiment of the present invention, an efficient compression decoding unit is added to the sending end to convert the original collected infrared images and visible light videos into a form that is conducive to network transmission, aiming to reduce the size of the packet to improve the response speed. This link achieves a large compression rate without significantly losing important details by executing a high-quality distortion encoding algorithm on the original file, thereby reducing the network burden.

[0056] In one example scenario, the output of high-definition dual-mode monitoring equipment in a fire detection system usually generates several megabytes of unprocessed data per second. Direct push would cause great pressure on channel occupancy and easily cause delays or freezes. After being processed by a specific optimization solution, the volume of the generated data set is significantly reduced to about one-fifth of the original volume, greatly promoting the fluency and immediacy of long-distance information interaction.

[0057] In addition, advanced error control strategies are adopted, such as selecting appropriate ARQ (Automatic Repeat Request) or EC (Extended Character Check) coding mechanisms, and applying effective forward error correction technology to improve transmission robustness. The former allows the receiving end to detect errors and request retransmission of erroneous data packets until all are received correctly, and the latter enables the other end to independently complete error correction by adding a certain number of additional check codes to the sequence to be transmitted in advance, without the need for reverse communication.

[0058] Specifically, at a fire scene, high temperature and smoke may interfere with communication quality, resulting in data packet loss. At this time, the carefully designed EC coding structure can effectively resist the negative impact of such unfavorable factors and ensure continuous and stable video images are transmitted back to the headquarters command department for remote decision-making and analysis.

[0059] Further quantitative evaluation is made based on the actual measured parameters of each communication channel: the maximum throughput Q measured between all established links is used i Its arithmetic mean Q avg The ratio of the overall average delay T avg Together, they reflect the health status of the network; Under the above conditions, the network function is judged to be operating normally; otherwise, the alarm mechanism will be triggered and the self-diagnosis program will be started to try to restore the optimal performance level until the standard position is reached again. The variable Q involved here corresponds to the total transmission load per unit time in each period, that is, the traffic scale, and T represents the time limit for single task execution. For the above relationship expression, a reasonable threshold setting should ensure that both the fast feedback characteristics of the system are taken into account and the necessary waiting window margin is not ignored. Therefore, it is recommended to use the empirical value of 1 second ≤ T min ≤3 seconds is used as an ideal boundary setting standard.

[0060] Under this framework, accurate fault location capability can be achieved under comprehensive monitoring coverage, and even in the event of an emergency, the key data chain can be kept uninterrupted.

[0061] Reference Figure 5 The embodiment of the present invention further provides an outdoor fire monitoring device 10, which is used to perform the above method, and the device includes: A detection module 100, which calibrates the temperature characteristics of objects of different materials in the detection area and generates compensation parameters; An input module 200, which performs adaptive fusion of two image input data according to the compensation parameters to stabilize the image contrast under illumination changes; A positioning module 300, which uses the adjusted early-stage fire feature extraction rule to locate the suspected fire source in the fused image; The early warning module 400 realizes real-time monitoring of mobile heat sources and eliminates false positive alarms based on dynamic parameter adjustment of the time series abnormal temperature rise early warning model.

[0062] See also Figure 6 , Figure 6 A schematic diagram of an electronic device 20 that can implement an embodiment of the present invention is shown, and the electronic device is intended to represent various forms of control devices. The components shown herein, their connections and relationships, and their functions are only examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0063] The electronic device 20 includes at least one processor 21, and a memory connected to the at least one processor 21, such as a read-only memory (ROM) 22, a random access memory (RAM) 23, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 21 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 22 or the computer program loaded from the storage unit 28 to the random access memory (RAM) 13. In the RAM 23, various programs and data required for the operation of the electronic device 20 can also be stored. The processor 21, ROM 22 and RAM 23 are connected to each other through a bus 24. An input / output (I / O) interface 25 is also connected to the bus 24.

[0064] A number of components in the electronic device 20 are connected to the I / O interface 25, including: an input unit 26, such as a keyboard, a mouse, etc.; an output unit 27, such as various types of displays, speakers, etc.; a storage unit 28, such as a disk, an optical disk, etc.; and a communication unit 29, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 29 allows the electronic device 20 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0065] The processor 21 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 21 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 21 performs the various methods and processes described above.

[0066] In some embodiments, the method of the above embodiment may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 28. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 20 via the ROM 22 and / or the communication unit 29. When the computer program is loaded into the RAM 23 and executed by the processor 21, one or more steps in the method described above may be performed. Alternatively, in other embodiments, the processor 21 may be configured to perform the method of the above embodiment in any other appropriate manner (e.g., by means of firmware).

[0067] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0068] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0069] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0070] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0071] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0072] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0073] In addition, refer to the attached Figure 7 The embodiment of the present application provides an outdoor camera with fire prevention function, the camera includes an image acquisition unit, a coding research unit and a network transmission unit, the camera is connected to a central server through the network transmission unit, the central server is installed with the above-mentioned electronic device 20, and the electronic device 20 is built with the above-mentioned outdoor fire monitoring device. The selection and configuration measures of the communication protocol of multiple cameras based on the above-mentioned embodiment are strengthened and improved, so that multiple cameras form a camera group, and the camera group is connected to the central server to realize uninterrupted and timely fire monitoring of multiple outdoor points.

[0074] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0075] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for outdoor fire monitoring, characterized in that: The steps include: By calibrating the temperature characteristics of objects of different materials in the detection area and generating compensation parameters; Adaptively fusing the two image input data according to the compensation parameters to stabilize the image contrast under illumination changes; The adjusted early-stage fire feature extraction rules are used to locate the suspected fire source in the fused image; Based on the dynamic parameter adjustment of the time series abnormal temperature rise warning model, real-time monitoring of mobile heat sources can be achieved and false positive alarms can be eliminated.

2. An outdoor fire monitoring method according to claim 1, characterized in that: The step of adjusting the parameters of the time series abnormal temperature rise warning model to achieve real-time monitoring and eliminate false alarms further includes: Time delay correction is performed based on the heat source movement speed and historical data: the predicted movement distance D is calculated by introducing the time difference information Δt of the historical image frame and the relative position deviation Δx of the current detection frame p ,formula , where V represents the moving speed of the heat source; A multi-level filter is constructed using machine learning algorithms to initiate a more fine-grained dynamic monitoring mechanism after the initial identification of a suspected fire source to eliminate false positives caused by non-target objects; Real-time optimization of temperature change thresholds: When a significant temperature rise occurs at the same location in N consecutive images, a fire is confirmed; Fusion of data streams captured synchronously by multiple cameras, ensuring consistency and timeliness in data transmission.

3. An outdoor fire monitoring method according to claim 1, characterized in that: The method of using the adjusted early fire feature extraction rules to locate suspected fire sources is further embodied in four steps as follows: Select appropriate morphological operators to enhance the unique features of the fire source area and suppress the influence of noise and other non-important features; Use color models combined with texture analysis techniques to help identify possible early signs of fire; Filter out the potential target object set according to specific conditions, that is, , where T r is the minimum temperature difference reference value, T u is the upper limit of the maximum temperature difference, A min Indicates the lowest limit of the target area, A r , A u are the upper and lower ranges of the aspect ratio, respectively. These values ​​are pre-set by experiments or obtained by adaptive adjustment.

4. The outdoor fire monitoring method according to claim 1, characterized in that: The adaptive fusion step is improved to solve the contrast distortion problem caused by illumination, and includes the following process: Construct a series of parameters for evaluating the intensity of light sources, such as illumination L(x,y), color shift ΔRGB, and contrast sensitivity function CSF, to evaluate the degree of light fluctuation and the resulting image quality issues; When illuminance L(x,y)>L max Or ΔRGB>ΔR max When the image is processed, a special image enhancement processing path is activated to ensure stable and reliable image quality required by subsequent steps.

5. The outdoor fire monitoring method according to claim 1, characterized in that: The following key steps are added by calibrating the temperature characteristics of objects of different materials in the detection area and generating compensation parameters: The system automatically learns typical reflection coefficient curves of various types and builds a database based on them to guide the classification and labeling of new material samples; Implement a high-resolution, multi-angle photography program to generate enough samples for subsequent analysis, especially for subtle color differences between different types of building materials that are difficult to distinguish; In view of the fact that the surface emission characteristics show different degrees of change with the angle, the view angle adaptive algorithm is used to compensate for the result deviation caused by the tilt of the field of view; A comprehensive characterization system including but not limited to conductive materials and thermal insulation coatings has been established to ensure comprehensive coverage of all treated materials and guarantee high credibility and accuracy of measurement results.

6. An outdoor fire monitoring device, used to implement the method according to any one of claims 1 to 5, characterized in that: The device includes: A detection module, which calibrates the temperature characteristics of objects of different materials in the detection area and generates compensation parameters; An input module, wherein the input module adaptively fuses two types of image input data according to the compensation parameter to stabilize image contrast under illumination changes; A positioning module, wherein the positioning module uses the adjusted early-stage fire feature extraction rule to locate the suspected fire source in the fused image; The early warning module realizes real-time monitoring of mobile heat sources and eliminates false positive alarms based on dynamic parameter adjustment of a time series abnormal temperature rise early warning model.

7. An electronic device, characterized in that: The method comprises at least one processor, wherein the processor is communicatively connected to at least one memory, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method according to any one of claims 1 to 5.

8. An outdoor camera with fire prevention function, characterized in that: The outdoor camera is provided with an electronic device as claimed in claim 7.

Images