An outdoor camera with fire prevention function

Abstract

The present invention discloses an outdoor camera with a fire prevention function, including: calibrating the temperature characteristics of objects with different materials in the detection area to generate compensation parameters; adaptively fusing two kinds of image input data according to the compensation parameters to stabilize the image contrast under changing light; applying the adjusted early fire feature extraction rule to locate the suspected fire source position in the fused image; and realizing real-time monitoring of moving heat sources and eliminating false positive false alarms based on the dynamic parameter adjustment of the time series abnormal temperature rise warning model. The beneficial effects of the present invention are: solving the problem of temperature sensing errors of objects with different materials in the detection area, improving the sensitivity of fire monitoring, and solving the problem of camera communication delay.

Description

Technical Field

[0001] The present invention relates to the technical field of fire monitoring, and particularly to an outdoor camera with fire prevention function. Background Art

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

[0003] In this context, the fire monitoring technology based on the linkage of infrared and visible light cameras has emerged. 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, realizing all-weather and multi-angle real-time fire monitoring. Through the fusion analysis of two types of image data, the fire can be identified more accurately, greatly improving the ability and scope of fire monitoring.

[0004] However, this technology still faces many challenges in practical applications:

[0005] Problems with the linkage mechanism of infrared and visible light cameras: The reflection and absorption characteristics of different materials for infrared and visible light vary greatly, which leads to errors in temperature perception by the cameras. For example, in complex industrial scenarios where multiple materials such as metal, plastic, and wood coexist, the existing linkage mechanism is difficult to accurately compensate for this temperature perception difference, easily causing misreading and affecting the accuracy of fire monitoring.

[0006] Problems with image fusion algorithms: The light conditions in the actual environment are complex and changeable, such as strong direct sunlight during the day, light interference at night, and light changes under different weather conditions. The existing image fusion algorithms cannot well dynamically adapt to these light environment changes. When the light intensity changes suddenly, the image contrast is unstable, key information is easily lost, and the accuracy of fire detection is reduced.

[0007] Problems with the rules for extracting initial fire characteristics: In the initial stage of a fire, the characteristics of flames and smoke are not obvious and are easily confused with other similar phenomena. The currently established rules for discriminating initial fire characteristics are not perfect. In non-fire alarm situations, the system is prone to false alarms, frequently disturbing the normal working order; at the same time, due to the limitations of the rules, the response to a real fire may be slow, delaying the best opportunity for fire fighting.

[0008] 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.

[0009] 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

[0010] The object of the present invention is to provide an outdoor fire monitoring method to solve the problems raised in the above background technology.

[0011] To achieve the above object, the present invention provides the following technical solutions:

[0012] In a first aspect, the present invention provides an outdoor fire monitoring method, comprising the following steps:

[0013] By calibrating the temperature characteristics of objects of different materials in the detection area and generating compensation parameters;

[0014] Adaptively fusing the two image input data according to the compensation parameters to stabilize the image contrast under illumination changes;

[0015] The adjusted early-stage fire feature extraction rules are used to locate the suspected fire source in the fused image;

[0016] 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.

[0017] 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:

[0018] 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;

[0019] Construct a multi-level filter using machine learning algorithms, and start a more fine-grained dynamic monitoring mechanism after the initial determination of the suspected fire source to eliminate false positives caused by non-target objects;

[0020] Optimize the temperature change threshold in real time: when a significant temperature rise occurs at the same position in N consecutive images, confirm the occurrence of a fire;

[0021] Fuse the data streams captured synchronously between multiple cameras and ensure consistency and timeliness during data transmission.

[0022] Preferably, the method of applying the adjusted early fire feature extraction rules to locate the suspected fire source is further specified as the following four steps:

[0023] 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;

[0024] Use a combination of color models and texture analysis techniques to assist in identifying possible early flame signs;

[0025] Screen out a set of potential target objects according to specific conditions,

[0026] That is , where T r is the minimum temperature difference reference value, T u is the maximum temperature difference value, A min represents the lowest limit of the target area, A r 、A u are the upper and lower ranges of the shape ratio respectively, and these values are obtained by pre-setting or adaptive adjustment through experiments.

[0027] Preferably, the adaptive fusion step is improved to solve the problem of contrast distortion caused by illumination, including the following process:

[0028] Construct a series of parameters for evaluating the strength of the light source, such as illuminance L(x,y), color offset ΔRGB and contrast sensitivity function CSF, so as to evaluate the degree of illumination fluctuation and the resulting picture quality problems;

[0029] When the illuminance L(x,y)>L max or ΔRGB>ΔR max Activate a special image enhancement processing path to ensure the stable and reliable image quality required for subsequent steps.

[0030] Preferably, the calibration of the temperature characteristics of different material objects in the detection area and the generation of compensation parameters add the following key steps:

[0031] The system automatically learns the typical reflection coefficient curves of various types and establishes a database based on this to guide the classification and marking work of new substance samples;

[0032] Implement a high-resolution multi-angle shooting plan to generate sufficient samples for subsequent analysis and mining, especially for the subtle color differences between different types of building materials that are difficult to distinguish.

[0033] In view of the fact that the surface emission characteristics vary to different degrees with the change of angle, a view angle adaptive algorithm is adopted to compensate for the result offset caused by the tilt of the field of view;

[0034] A comprehensive characterization system including but not limited to conductive materials and thermal insulation coatings is established to ensure full coverage of all treated materials and guarantee the high reliability and accuracy of the measurement results.

[0035] 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 includes:

[0036] A detection module that calibrates the temperature characteristics of objects of different materials in the detection area and generates compensation parameters;

[0037] An input module that adaptively fuses two image input data according to the compensation parameters to stabilize the image contrast under changing illumination;

[0038] A positioning module that applies the adjusted fire early-stage feature extraction rules to locate the suspected fire source position in the fused image;

[0039] An early warning module that realizes real-time monitoring of moving heat sources and eliminates false positive alarms based on the dynamic parameter adjustment of the time series abnormal temperature rise early warning model.

[0040] In a third aspect, the present invention provides an electronic device including at least one processor, and the processor is communicatively connected to at least one memory. Among them, 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 described in any one of the above embodiments.

[0041] In a fourth aspect, the present invention provides an outdoor camera with a fire prevention function, and the outdoor camera is provided with an electronic device as described in the above embodiments.

[0042] Compared with the prior art, the beneficial effects of the present invention are as follows: The present invention calibrates the temperature characteristics of objects made of different materials in the detection area and generates compensation parameters; adaptively fuses two types of image input data according to the compensation parameters to stabilize the image contrast under changing lighting conditions; applies the adjusted early fire feature extraction rules to locate the suspected fire source position in the fused image; and realizes real-time monitoring of moving heat sources and eliminates false positive false alarms through dynamic parameter adjustment of the time series abnormal temperature rise warning model. Through the solution of the embodiments of the present disclosure, it is possible to solve:

[0043] 1. How to optimize the linkage mechanism of infrared and visible light cameras to solve the problem of temperature sensing errors of objects made of different materials in the detection area.

[0044] 2. How to regulate the fusion algorithm of two types of image input data to solve the problem of unstable image contrast caused by changing lighting conditions.

[0045] 3. How to regulate the rule setting for early fire feature extraction to solve the problem of frequent false positive false alarms.

[0046] 4. How to adjust the parameters of the abnormal temperature rise warning model based on time series to solve the problem of monitoring delay caused by the movement of heat sources.

[0047] 5. How to regulate the selection and configuration of communication protocols between cameras to solve the problem of large image transmission synchronization errors in a multi-camera system. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 is a schematic flow chart of the method of the present invention;

[0049] Figure 2 are the steps for the present invention to adjust the parameters of the time series abnormal temperature rise warning model to achieve real-time monitoring and eliminate false alarms;

[0050] Figure 3 are the steps for the present invention to apply the adjusted early fire feature extraction rules to locate the suspected fire source, which are specifically divided into four link processes;

[0051] Figure 4 are the improvement processes for the adaptive fusion step of the present invention to solve the problem of contrast distortion caused by lighting;

[0052] Figure 5 is a module diagram of an outdoor fire monitoring device provided by the present invention;

[0053] Figure 6 is a module diagram of an electronic device provided by the present invention;

[0054] Figure 7This is a schematic diagram of the network connection of the outdoor camera of the present invention. Detailed implementation manners

[0055] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0056] In an embodiment of the present invention, referring to the attached Figure 1 , this embodiment provides a fire monitoring method. The method includes:

[0057] S101. Calibrate the temperature characteristics of objects with different materials in the detection area and generate compensation parameters;

[0058] S102. Adaptively fuse two image input data according to the compensation parameters to stabilize the image contrast under the illumination change;

[0059] S103. Apply the adjusted fire early-stage feature extraction rule to locate the suspected fire source position in the fused image;

[0060] S104. Based on the dynamic parameter adjustment of the time series abnormal temperature rise warning model, perform real-time monitoring on the moving heat source and eliminate false alarms of false positives.

[0061] By calibrating the temperature characteristics of objects with different materials in the detection area and generating compensation parameters, the linkage mechanism between the infrared and visible light cameras is optimized, and the temperature sensing error caused by objects with different materials in the detection area is solved. First, before implementation, a detailed investigation of the monitoring area is required, and the temperature curves of different types of materials and their standard environments are recorded.

[0062] Then, according to the above-obtained compensation parameter adjustment method, an adaptive image processing fusion technology is implemented for two image input sources, namely the infrared imaging view and the optical vision captured picture, to address the problem of uneven image quality caused by drastically changing light under outdoor conditions. Specifically, after receiving the original material transmitted from the front end, the RGB color mode is converted into separate processing of the luminance Y component, chrominance U component, and V component. Different filter operations are applied respectively and then recombined to output the final version. For example, in a surveillance environment of a sunny afternoon square, if relying solely on a visible light camera, overexposure may occur, causing the loss of key contours around the fire, and it is difficult to accurately determine the shape of the smoke and other complex scenes relying solely on single-dimensional thermal-sensitive information. At this time, after calculating and mixing the advantages of the two pieces of data through a specific formula, a composite layout that is both bright and prominent in the location of the fire point and has clear background detail levels can be formed, providing a solid and reliable basis for the next step of recognition, reducing the misreading rate caused by environmental fluctuations, and ensuring the overall painting style is coordinated and unified.

[0063] In addition, regarding the issue of adjusting the early fire form recognition rules, a machine training method is adopted to pre-feed a large amount of teaching resource libraries containing real cases, including flame image samples at various angles and distances, into the computer system, and artificial labels for distinguishing true and false targets are added to build a deep neural network to complete the initial model creation. Then, the threshold setting parameters are continuously adjusted using the latest feedback results to adapt to the changing trend of the actual application scenario, reducing unnecessary alarm times and improving work efficiency. For example, in a warehouse goods storage area, there are often phenomena such as personnel carrying heating appliances into the operation area or the exhaust pipes of vehicle engines emitting sparks, which are likely to trigger misjudgments by the traditional fixed index system. The new solution can flexibly configure the discrimination conditions in combination with environmental information, allowing a certain error tolerance range. Only when most of the characteristic factor combinations are met will it be determined that there is a real risk of fire ignition, and then an alarm instruction will be issued to notify relevant personnel for emergency handling, avoiding frequent interruption of the normal work rhythm and enhancing the early warning accuracy at the same time.

[0064] Immediately regarding the part of establishing an abnormal body temperature rise pre-alarm plan based on the statistical laws of time series, considering that there are situations where fast-moving objects pass through the camera's field of view in some scenarios and the real-time dynamic capture and tracking function needs to be ensured. For this reason, a multi-dimensional feature point extraction and tracking algorithm is introduced to regularly capture key-frame images and compare the similarity between adjacent frames at intervals to determine whether there are signs of sudden increase in hot spots. At the same time, considering the possible delay phenomenon caused during the transfer of the heat source along the path, a buffer time slice window function is specially set to give an early warning to remind the relevant person in charge to pay attention to potential dangers and intervene in time to investigate and verify the on-site situation to prevent missing the best rescue opportunity. For example, near a forest fire lookout tower, once a suspected smoke signal is detected but it moves quickly, it is very likely that a wild animal is running by and the heat emitted by its body causes a short-term local hot spot. Then, at this time, it is necessary to rely on historical trajectory modeling to predict its action direction trend and combine geographical location analysis to determine the final confirmation plan, effectively eliminating non-hazardous sources and reducing the spread of panic.

[0065] Finally, regarding the work of selecting and matching the communication protocols between cameras, considering that modern large-scale distributed networking architectures often have the need to compatible and dock multiple products with different models and specifications. In order to ensure the coordinated operation of multiple devices, a stable and efficient communication method must be selected and various initialization configuration parameters such as IP addresses, port numbers, authentication keys, etc. must be set scientifically and reasonably to ensure smooth information transmission without obstruction. In a large enterprise park spanning different regions, hundreds or even thousands of monitoring devices produced by different manufacturers are installed. If no unified and standardized technical guidance criteria are formulated, it is very likely that there will be mutual interference, conflicts, or even disconnection and offline situations, resulting in serious consequences such as video loss and fragmentation. The security connection mechanism under the TCP / IP protocol family is preferably adopted, and the performance indicators of each link are centrally regulated through a dedicated management background. The firmware and software are regularly maintained and updated, and patches are repaired to ensure that the system is always in the optimal working state, meeting the continuous operation requirements of 7×24 hours. Fundamentally overcome the challenges brought by hardware and software differences, achieving the ideal effect of seamless docking and interconnection, thereby greatly improving the fire emergency response efficiency and laying a more solid intelligent technology foundation for protecting people's lives and property safety.

[0066] In one embodiment, a monitoring system composed 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 through programming. Among them, assume that there is a threshold formula T = I * R + B to determine the triggering of the alarm mechanism. In this formula, I represents the effective fire probability extracted from the infrared image; R represents the weighted proportionality factor, taking a real number between 0 and 1, and a good recognition accuracy can be obtained when the optimized value is 0.8; B is the background error coefficient, which is an integer starting from 0 and increasing, and the usually preferred value is 5.

[0067] Then refine the operation process and application scenarios. Specifically, it will describe the information transfer methods and timing control between various hardware units, etc. For example, the method mentioned in the above example will also include how to process image data and perform pattern matching judgments, that is, when it detects that the temperature exceeds the preset limit or discovers a specific color change trend, it will immediately generate corresponding warning information to ensure a quick response to abnormal situations, reduce the false alarm rate, and improve the security of the entire system.

[0068] In one embodiment of the present invention, refer to the attached Figure 2 , the steps of adjusting the parameters of the time series abnormal temperature rise warning model of the present invention to achieve real-time monitoring and eliminate false alarms.

[0069] S201. Perform time delay correction based on the heat source movement 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 predicted displacement D of the heat source p . The formula indicates that the predicted distance is equal to the heat source speed multiplied by the time period difference. Here, Δt can be defined within a range according to the system capture frequency; the optimal value depends on the object movement speed and camera device refresh rate in the specific scenario; V represents the real-time observed or expected heat source movement speed obtained from the previous learning process. In practical applications, if a certain heat source shows a trend of moving from the upper left corner to the lower right corner on both 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.

[0070] S202. Use machine learning to construct a multi-level filter to initiate a more refined and continuously monitored review process to identify the authenticity of potential alarms and eliminate 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 a fire, then a secondary verification mechanism is triggered at this time. It can analyze other data characteristics that may cause errors, such as smoke diffusion, light intensity, etc., and adjust the threshold or take additional verification actions accordingly until it is ensured that the detected target is indeed the ignition point.

[0071] S203. Implement dynamic fine-tuning of the temperature change threshold so that the condition that a significant temperature increase ( ) is shown in the same pixel area of N consecutive frames can be determined as a fire occurrence. Here, N refers to the number of images continuously captured to confirm a fire, and its value should be sufficient to filter out transient heat fluctuations but not overly increase the probability of delayed response. Among them, ΔT represents the temperature change amount, T th is the temperature change threshold, P(ΔT>T c | data from fire)≥P cis the probability that the data comes from a fire condition where ΔT is greater than T c , T c is a reference value of another temperature, P C is a set probability threshold, and N is the number of consecutive images used to determine a fire. For example, a fixed interval is set to extract several video frames for evaluation, and it is required that at least three consecutive time points reach a predetermined temperature increment, and based on Bayesian theory, the likelihood of a large fire occurring is estimated to be greater than a given confidence interval, then an anomaly is officially reported.

[0072] S204. Integrate the information flows provided by each video sensing unit to ensure stable, reliable, and timely data interaction, which is an essential part of realizing the overall monitoring efficiency. Specifically, for a network system with multiple complementary cameras installed at different perspectives, 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 and ensure that no important segment is lost, so as to maintain an effective full-course tracking ability. Suppose there is a set of deployed panoramic monitoring stations responsible for covering every corner inside a building, then by precisely adjusting the transmission rate of each camera, protocol adaptation, etc., it is ensured that there is no lag or replay phenomenon in the video stream transmission.

[0073] In an embodiment of the present invention, referring to Appendix Figure 3 , the steps of applying the adjusted fire early-stage feature extraction rules of the present invention to locate a suspected fire source are specifically divided into the following four links.

[0074] S301. Select a suitable morphological operator 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 various interfering objects or non-flame elements. The use of morphological operators is to highlight the target area through means such as opening operation, closing operation, or morphological gradient, remove small details in the image that do not affect the overall analysis result, and smooth the edges to make the boundary of the suspected fire area clear, facilitating subsequent precise processing and identification. For example, when detecting fire hazards in a forest, the closing operation with a rectangular kernel structural element can be used to merge the shadow parts caused by the shooting angle or small separated parts caused by dust coverage.

[0075] S302. Combine the color model and texture analysis technology to identify potential early flame signs. Utilize the contrast difference and spectral characteristics of the data captured by cameras in different bands, and comprehensively evaluate whether the suspected area has the typical color attributes of a flame from three aspects: hue, brightness, and saturation; simultaneously consider texture characteristics such as the strength of graininess and its arrangement pattern to judge the possibility of the dynamic propagation of the flame. In an embodiment, the infrared camera can sense the temperature gradient change to form a thermal image, while the conventional camera obtains a visual image, and then a dual-threshold algorithm is established by integrating the advantages of both to evaluate whether the local high-temperature point has the appearance of a flame and whether its surface fiber texture flickers regularly.

[0076] 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.

[0077] 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.

[0078] 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.

[0079] S401. Construct a series of parameters for evaluating the intensity of the light source, including illuminance L(x,y), color offset ΔRGB, and contrast sensitivity function CSF, to determine the degree of negative impact of the light state in the current environment on the image. Specifically, L(x,y) represents the spatial brightness distribution value measured at the position (x,y); the color offset ΔRGB is used to represent the degree of difference between the three primary colors of red, green, and blue. When this value is too large, it implies the existence of abnormal color tones caused by complex light conditions such as strong light or backlight; while the contrast sensitivity function CSF reflects the different sensitivity characteristics of the human visual system to 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 in a fire scene, due to uneven external lighting or direct natural light entering the range of the surveillance camera, a strong contrast is formed, then applying the above parameters to this situation can accurately identify the root cause of the problem.

[0080] S402. When it is detected that the illuminance L(x,y) exceeds the preset maximum value L max , or ΔRGB exceeds the critical upper limit ΔR max , then enable a special image enhancement program path to ensure a stable and reliable supply of original data required for subsequent processes. This limit is set because when the imaging quality exceeds the limit level, it will significantly decline, which is not conducive to analysis and interpretation. At the same time, it is also beneficial to protect sensitive electronic components from being damaged by excessive light. Specifically, when setting reasonable thresholds, the sensitivity of the device and the requirements of the actual application scenario are taken into account, and these key values are not selected randomly. Taking an example, in a case where smoke occlusion or bright flames appear and interfere with the coordinated work of infrared and visible light camera components, once the trigger condition is met, a specific algorithm is immediately called to improve the quality stability of the information content obtained by the acquisition port.

[0081] S403. Apply a novel method that combines local gradient estimation and global histogram equalization techniques to perform image enhancement operations. 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 brought by the shadow part; as for the latter, by rearranging the overall gray spectrum, the dynamic range becomes wider, so as to better reflect the detailed characteristics of the true appearance of the target. For example, accurately reproduce the temperature field distribution in the surrounding environment of the high-heat radiation area emitted by the burning object, which not only improves the clarity but also enhances the interpretability and practicality.

[0082] 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.

[0083] 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:

[0084] 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.

[0085] 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.

[0086] 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.

[0087] Finally, a comprehensive characterization system covering various characteristics such as conductivity and heat insulation is established. Doing so not only allows for a fine division according to the characteristics of each type of material, but also helps to construct a more perfect sensing framework, enabling all monitored objects to obtain highly reliable and stable reading feedback. For example, in cases involving electrical equipment fires, this system effectively distinguishes the temperature change trends in the high-temperature areas around metal lines and the temperature and humidity states in areas with thermal insulation measures, greatly improving the early warning efficiency and response capabilities for disposal. In addition, it also shows superior performance in non-conductive electromagnetic environments, such as the ability to efficiently identify abnormal surface temperatures of walls with special component coatings.

[0088] In one embodiment of the present invention, the selection and configuration measures regarding communication protocols of the present invention are strengthened and improved: First, the one that best matches the actual network conditions is selected from a series of predefined communication protocols. The specific selection basis is not limited to stability, but also includes factors such as bandwidth utilization efficiency. These evaluation factors ensure that the selected protocol can not only meet the real-time requirements but also maintain efficient data transmission in unstable or low-bandwidth environments.

[0089] For example, considering that the application scenario of the fire monitoring method requires the simultaneous operation of infrared and visible light cameras, which demands the system to have the flexibility to process multiple data types and strong environmental adaptability. To this end, in one embodiment, the system dynamically adjusts to the TCP or UDP protocol according to the quality of the current wireless or wired connection. If the Wi-Fi signal is poor in the on-site deployment area, TCP with better reliability is preferentially selected; while if it is operating within a high-speed Ethernet, it can be switched to UDP with high efficiency but slightly lower fault tolerance to optimize the upload efficiency of video streams and other sensor information.

[0090] In one embodiment of the present invention, an efficient compression and decoding unit is added at the sending end to be responsible for converting the originally collected infrared images and visible light videos into a form conducive to network transmission, aiming to reduce the packet size to improve the response speed. This link achieves a large compression ratio through an encoding algorithm with high-quality distortion for the original file without significantly losing important details, thereby reducing the network burden.

[0091] In an example scenario, for the output content of high-definition dual-modal monitoring devices in a fire alarm detection system, usually several megabytes of unprocessed data can be generated per second. Directly pushing it will cause a great pressure on channel occupancy and is prone to delays or lags. After being processed by a specific optimization scheme, the volume of the generated data set is significantly reduced to about one-fifth of the original volume, greatly promoting the smoothness and immediacy in the process of long-distance information interaction.

[0092] 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 techniques to improve transmission robustness. The former allows the receiving end to detect errors and request the retransmission of incorrect data packets until all are correctly received. The latter enables the peer 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.

[0093] Specifically, at a fire scene, the communication quality may be interfered by high temperature and smoke, resulting in data packet loss. At this time, with the help of a carefully designed EC coding structure, the negative impacts brought by such adverse factors can be effectively resisted, and continuous and stable video images can be transmitted back to the headquarters command department for remote decision-making and analysis.

[0094] Furthermore, a quantitative evaluation is made based on the actual measurement parameters of each communication channel: using the ratio of the maximum throughput Q i measured between all established links to its arithmetic mean Q avg and the overall average delay T avg together reflect the quality of the network health status; when under certain conditions, it is judged that the network functions properly; otherwise, the alarm mechanism will be triggered and the self-diagnosis program will be started to try to restore the optimal performance level until it reaches the standard again. Here, the variable Q corresponds to the total transmission load per unit time in each period, that is, the traffic scale, and T represents the time limit for a single task to execute. For the above relational expression, the reasonable threshold setting should ensure both the fast feedback characteristics of the system and not neglect the necessary waiting window margin, so it is recommended to use the empirical value 1 second ≤ T min ≤ 3 seconds as an ideal boundary setting standard.

[0095] Under this framework, the accurate fault location ability under full monitoring coverage is realized, and even in case of sudden accidents, the critical data chain can be maintained unbroken.

[0096] Refer to Appendix Figure 5 , an embodiment of the present invention further provides an outdoor fire monitoring device 10 for performing the above method, and the device includes:

[0097] A detection module 100, which calibrates the temperature characteristics of different material objects in the detection area and generates compensation parameters;

[0098] An input module 200, which adaptively fuses two image input data according to the compensation parameters to stabilize the image contrast under the change of illumination;

[0099] A positioning module 300, which locates the suspected fire source position in the fused image by applying the adjusted fire early - stage feature extraction rule;

[0100] An early - warning module 400, which realizes real - time monitoring of the moving heat source and eliminates false - positive false alarms based on the dynamic parameter adjustment of the time - series abnormal temperature - rise early - warning model.

[0101] Please refer to Figure 6 , Figure 6 , which shows a schematic diagram of the mechanism of the electronic device 20 in which the embodiments of the present invention can be implemented. The electronic device is intended to represent various forms of control devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described herein and / or claimed.

[0102] The electronic device 20 includes at least one processor 21 and a memory communicatively connected to the at least one processor 21, such as a read - only memory (ROM) 22, a random - access memory (RAM) 23, etc. Among them, the memory stores a computer program executable by the at least one processor. 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 into 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, the ROM 22, and the RAM 23 are connected to each other through a bus 24. The input / output (I / O) interface 25 is also connected to the bus 24.

[0103] A plurality 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 disc, 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.

[0104] The processor 21 can be various general - purpose and / or special - purpose 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 executes the various methods and processes described above.

[0105] In some embodiments, the method of the above embodiments may be implemented as a computer program tangibly embodied in a computer-readable storage medium, such as storage unit 28. In some embodiments, part or all of the computer program may be loaded and / or installed onto 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 of the methods described above may be performed. Alternatively, in other embodiments, the processor 21 may be configured to perform the methods of the above embodiments by any other suitable means (e.g., by means of firmware).

[0106] The various implementations of the systems and techniques described above in this document can be implemented in digital electronic circuitry, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-chip (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations may 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 may be a dedicated 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 the data and instructions to the storage system, the at least one input device, and the at least one output device.

[0107] The 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 dedicated computer, or other programmable data processing device, such that when the computer programs are executed by the processor, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The computer programs may be executed entirely on the machine, partially on the machine, as a stand-alone software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0108] In the context of the present invention, a computer-readable storage medium can be a tangible medium that can contain or store a computer program for use by or in connection with an instruction execution system, apparatus, or device. The computer-readable storage medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. Alternatively, the computer-readable storage medium can be a machine-readable signal medium. More specific examples of the machine-readable storage medium would include an electrical connection based on one or more wires, 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 disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0109] In order to provide interaction with a user, the systems and techniques described herein can 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 a trackball) by which the user can provide input to the electronic device. Other kinds of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0110] The systems and techniques described herein can 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 having a graphical user interface or a web browser through which the user can interact with an implementation 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 can 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.

[0111] A computing system may include a client and a server. The client and the server are generally far from each other and usually interact via a communication network. The client-server relationship is created by computer programs running on respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or a cloud host, which is a host product in the cloud computing service system, solving the defects of difficult management and weak business scalability existing in traditional physical hosts and VPS services.

[0112] In addition, referring to the attached Figure 7 , embodiments of the present application provide an outdoor camera with a fire prevention function. The camera includes an image acquisition unit, an encoding and 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 internally provided with the above-mentioned outdoor fire monitoring device. The selection and configuration measures regarding the communication protocol of multiple cameras based on the above embodiments are strengthened and improved, so that multiple cameras form a camera group. The camera group is connected to the central server to achieve continuous and timely fire monitoring of multiple outdoor points.

[0113] It should be understood that various forms of the processes shown above can be used, reordering, adding, or deleting steps. For example, the steps recited in the present invention can be executed in parallel, sequentially, or in a different order, as long as the desired results of the technical solution of the present invention can be achieved, and no limitation is made herein.

[0114] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An outdoor fire monitoring method, characterized in that, Including the following steps: Calibrating the temperature characteristics of objects with different materials in the detection area and generating compensation parameters; Adapting and fusing two kinds of image input data according to the compensation parameters to stabilize the image contrast under changing illumination; Applying the adjusted early fire feature extraction rules to locate the suspected fire source position in the fused image, including four steps: Selecting appropriate morphological operators to enhance the unique features of the fire source area and suppressing the influence of noise and other non-important features; Using a combination of color models and texture analysis techniques to assist in identifying possible early flame signs; Filter out a set of potential target objects according to specific conditions, that is , where Tr is the minimum temperature difference reference value, Tu is the upper limit of the maximum temperature difference, Amin represents the lowest limit of the target area, and Ar and Au are the upper and lower ranges of the shape ratio respectively, and these values are preset or adaptively adjusted by experiments; Reducing the background complexity based on the background modeling principle; Implementing real-time monitoring of moving heat sources and eliminating false positive alarms through dynamic parameter adjustment of the time series abnormal temperature rise warning model.

2. 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, including the following process: Constructing a series of parameters for evaluating the strength of the light source, such as illuminance L(x,y), color offset ΔRGB, and contrast sensitivity function CSF, to evaluate the degree of illumination fluctuation and the resulting image quality problems; When the illuminance L(x,y) > Lmax or ΔRGB > ΔRmax, activating a special image enhancement processing path to ensure the stable and reliable image quality required for subsequent steps.

3. The outdoor fire monitoring method according to claim 2, characterized in that: The step of calibrating the temperature characteristics of objects with different materials in the detection area and generating compensation parameters adds the following key steps: The system automatically learns the typical reflectance coefficient curves of various types and establishes a database based on this to guide the classification and marking of new substance samples; Implementing a high-resolution multi-angle shooting plan to generate sufficient samples for subsequent analysis and mining, especially for the subtle color differences between different types of building materials that are difficult to distinguish; In view of the fact that the surface emission characteristics change regularly to varying degrees with the change of angle, using a perspective adaptive algorithm to compensate for the result offset caused by the tilt of the field of view; Establishing a comprehensive characterization system including but not limited to conductive materials and thermal insulation coatings to ensure full coverage of all materials to be treated and guarantee the high credibility and accuracy of the measurement results.

4. An outdoor fire monitoring device for implementing the method according to any one of claims 1 to 3, characterized in that, The device includes: A detection module that calibrates the temperature characteristics of objects with different materials in the detection area and generates compensation parameters; An input module that adaptively fuses two kinds of image input data according to the compensation parameters to stabilize the image contrast under changing illumination; A positioning module that applies the adjusted early fire feature extraction rules to locate the suspected fire source position in the fused image; An early warning module that implements real-time monitoring of moving heat sources and eliminates false positive alarms through dynamic parameter adjustment of the time series abnormal temperature rise warning model.

5. An electronic device, characterized in that, Including at least one processor, 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 3.

6. An outdoor camera with a fire prevention function, characterized in that: The outdoor camera is provided with an electronic device as described in claim 5.

Images