Method for realizing unmanned aerial vehicle to patrol and extinguish fire in forest fire prevention

Abstract

This invention provides a method for using drones for forest fire prevention and inspection, including: the drone acquiring an electronic map of the area to be inspected, and determining the inspection route based on the electronic map and flight time; after takeoff, the drone's infrared camera captures color images of the inspection area, and analyzes the flame images from the color images; if there are no flames, the inspection continues; if flames are present, an alarm is issued and firefighting begins; the captured and extinguished color images are wirelessly transmitted to a control platform to confirm the completion of firefighting. This invention analyzes the images to determine the number of sharp corners, and determines whether the number of sharp corners reaches a preset threshold. If the threshold is reached, the drone descends; if an open flame is confirmed, firefighting begins. This achieves multi-functionality for the drone, enabling it not only to inspect but also to extinguish fires, greatly improving the drone's intelligent control level and realizing automatic inspection, automatic detection, automatic judgment, and automatic firefighting.

Description

Technical Field

[0001] This invention relates to the field of forest fire prevention technology, and in particular to a method for enabling drones to patrol and extinguish forest fires. Background Technology

[0002] Forests are biological communities dominated by woody plants, including arbor forests, bamboo forests, and shrublands specifically designated by the state. Forest vegetation is crucial for maintaining the atmospheric environment. Vegetation absorbs carbon dioxide and produces oxygen through photosynthesis, creating a favorable environment for human survival and development, purifying the air, and maintaining the Earth's ecological balance. However, forest fires, caused by both natural and human factors, result in the destruction of vast amounts of vegetation, and the resulting dust and other pollutants cause environmental pollution. Due to the large area of ​​forests, traditional fire prevention relied on manual methods, such as patrolling to detect fires or potential hazards and using fire extinguishers, water, or tools to put them out. This approach was not only inefficient in detecting fires in a timely manner but also had low patrol and firefighting efficiency. With continuous technological advancements, drones have played an important role in forest fire prevention; however, their intelligence level is relatively low, and their functions are limited, making it difficult to simultaneously perform patrol and firefighting functions.

[0003] Existing technology 1, CN201811573497.0, describes an inspection robot and its fire extinguishing equipment and method. The fire extinguishing equipment includes a mounting frame and a fire extinguisher and a telescopic mechanism, both mounted on the mounting frame. A push-button switch is located at the end of the fire extinguisher facing the telescopic mechanism. The telescopic mechanism extends towards the push-button switch and actuates it, opening the fire extinguisher. Using a telescopic mechanism to open the fire extinguisher improves the mechanization of the fire extinguishing equipment and facilitates automated control. While this fire extinguishing equipment, when installed on an inspection robot, allows the robot to extinguish fires immediately upon discovery, preventing them from spreading in their early stages, it is unsuitable for forest fire prevention. Furthermore, the robot's ground-based movement limits its inspection area and results in low inspection efficiency.

[0004] Existing technology two, CN115300828A, describes a forest fire rescue and extinguishing device, including a monitoring station, multiple rescue drones, a satellite, a fire situation handling platform, a forest fire prevention center, and a data transmission module. A charging station for charging the rescue drones is located between two adjacent monitoring stations. When any of the rescue drones, the fire situation handling platform, or the satellite detects a fire, the data transmission module transmits the fire information to the forest fire prevention center. While this device integrates multiple methods for forest fire prevention and rescue, including monitoring stations, remote patrols with rescue drones, and satellite monitoring, and features an improved and innovative charging pad that allows for remote control of the rescue drones to automatically plug in and disconnect power, making it simple and convenient to operate and enabling fully unmanned operation, particularly suitable for forest areas, and low in cost, it only achieves the functions of inspection and fire detection, lacking fire extinguishing capabilities. This limited functionality results in low forest fire prevention efficiency.

[0005] Existing technology three, CN114949664A, describes a land-air linkage path planning and control method for forest fire patrol robots, which includes the following steps: When a satellite remote sensing system detects a fire in a certain area of ​​the forest, it transmits an alarm signal to the forest fire control console. Upon receiving the alarm, the forest fire control console first dispatches a drone formation as an advance force, followed by a fire patrol mobile robot to the designated location for fire suppression. The advance drone formation first finds a suitable path for the fire patrol mobile robot based on the rough terrain data provided by the remote sensing satellite, flies towards the fire point, and promptly reports any environmental obstacles affecting the robot's movement. Although the fire patrol mobile robot can travel quickly based on the correct path provided by the advance drone formation, saving time and introducing a multi-functional intelligent robot to improve the efficiency of forest fire patrol, its ground patrol movement speed is slow, and it cannot quickly reach the fire point, resulting in poor patrol and fire suppression effects.

[0006] Currently, existing technologies 1, 2, and 3 suffer from the problem that existing intelligent forest fire prevention equipment has limited functionality, low inspection and fire extinguishing efficiency, and cannot effectively play a role in forest fire prevention. Therefore, this invention provides a method for using drones for inspection and fire extinguishing in forest fire prevention. By using drones, large-scale forest inspection and fire extinguishing can be achieved, greatly reducing the occurrence of fires. Summary of the Invention

[0007] To address the aforementioned technical problems, this invention provides a method for using drones for inspection and firefighting in forest fire prevention, comprising the following steps:

[0008] The drone acquires an electronic map of the area to be inspected and determines the inspection route based on the electronic map and flight time.

[0009] After the drone takes off, its infrared camera captures color images of the inspection area and analyzes the flame images from the color images. If there are no flames, the inspection continues; if there are flames, an alarm is triggered and the fire is extinguished.

[0010] Color images of the captured and extinguished fire are wirelessly transmitted to the control platform to confirm that the fire has been extinguished.

[0011] Optionally, the process of confirming the inspection route using a drone includes:

[0012] The drone downloads an electronic map of the area to be inspected via the internet;

[0013] The area to be inspected is divided into several nodes according to the topography and vegetation distribution. The nodes include important nodes and non-important nodes.

[0014] The drone plans its inspection route according to the preset inspection path rules within the safe flight time;

[0015] The drones inspect the areas to be inspected according to the planned inspection routes.

[0016] Optionally, the electronic map includes the boundaries, topography, and vegetation distribution of the area to be inspected; important nodes are those prone to fire, while non-important nodes are those unlikely to be prone to fire.

[0017] Optional, preset inspection path rules include: not exceeding the boundary of the area to be inspected, the proportion of important nodes being 20% ​​higher than that of non-important nodes, and the time allotted for the drone to reach the next charging station.

[0018] Optionally, the preset inspection path rules also include: the UAV optimizes the dynamic path based on obstacles, flight path deviation and UAV energy consumption, specifically including obstacle collision function, flight path deviation function and UAV energy consumption effective function, and plans the dynamic path based on the obstacle collision function, flight path deviation function and UAV energy consumption effective function.

[0019] Optionally, the process of analyzing a flame image from an image includes:

[0020] The infrared camera captures a color image of the inspection area, and the image processing module of the infrared camera converts the color image to grayscale to obtain a first grayscale image.

[0021] The image processing module performs image segmentation on the first grayscale image to obtain a flame mask image corresponding to the first grayscale image, removes the background of the flame mask image, and obtains a second grayscale image with the background removed.

[0022] The image processing module scans the second grayscale image, labels white pixels as 0, and when it encounters a black pixel, it determines whether to label the current pixel based on whether the surrounding pixels are labeled. If labeled, it records it as 1, thus obtaining a labeled third grayscale image.

[0023] The image processing module then scans the third grayscale image and counts the number of sharp corners that are not less than 8. If there are not less than 8 sharp corners, it is determined that the third grayscale image contains a flame image, that is, there is an open flame in the inspection area, which could lead to a fire.

[0024] If an open flame is detected, the drone's controller receives an open flame command from the image processing module, issues an alarm, and activates the fire extinguishing grenade launching system to extinguish the fire. After launch, the infrared camera re-captures a color image to determine if a flame image exists. If no flame image is found, the drone continues its inspection; if a flame image is found, the drone continues to perform fire extinguishing actions.

[0025] Optionally, the image processing module performs grayscale conversion on the color image, including:

[0026] Extract the R, G, and B channels of the color image and vectorize them into a column vector; then perform preliminary grayscale conversion.

[0027] The subjective perceived brightness of the neighborhood is obtained by unilateral filtering of the initially grayscale color image;

[0028] Based on the difference between the perceived brightness in the neighborhood and the actual light intensity, the local contrast of the initially grayscale color image is linearly adjusted to obtain the first grayscale image.

[0029] Optionally, the image processing module performs image segmentation on the first grayscale image, including:

[0030] The image processing module inputs a training color image of the inspection area in the preset training samples into the preset network model to determine the predicted color image mask corresponding to the training color image; the training color image contains the flame area.

[0031] Obtain the flame edges in the predicted color image mask, and determine the weight map corresponding to the predicted color image mask based on the flame edges;

[0032] The first grayscale image segmentation model is obtained by training the preset network model based on the weight map, the predicted color image mask map, and the flame image mask map corresponding to the training color image.

[0033] The first grayscale image is input into the first grayscale image segmentation model to determine the flame mask image corresponding to the first grayscale image.

[0034] Optionally, the background of the flame mask image is removed to obtain a second grayscale image with the background removed, including:

[0035] Color sampling is performed on the main area of ​​the flame mask image;

[0036] The range of sampled colors is expanded based on the color tolerance value to obtain the retained color range;

[0037] The flame mask image is processed to remove the background based on the preserved color range, resulting in a second grayscale image with the background removed.

[0038] Optionally, the process of wirelessly transmitting color images of the captured and extinguished fire to the control platform includes:

[0039] The control platform receives and stores color images captured by the drone and used in firefighting operations.

[0040] If the control platform detects that the fire extinguishing has not met the set standard, it will control the drone fire extinguishing projectile launching system to continue executing the fire extinguishing command;

[0041] Once the fire is confirmed to be extinguished, an instruction is issued for the drone to continue its inspection.

[0042] This invention first uses a drone to acquire an electronic map of the area to be inspected, and then determines the inspection route based on the electronic map and flight time. Second, after takeoff, the drone's infrared camera captures color images of the inspection area and analyzes these images to identify flames. If no flames are found, the inspection continues; if flames are present, an alarm is triggered, and firefighting begins. Finally, the captured and extinguished color images are wirelessly transmitted to the control platform to confirm firefighting completion. The invention utilizes the drone's infrared camera to capture images of the inspection area, analyzes these images to determine the number of sharp corners, and determines if the number reaches a preset threshold. If the threshold is reached, the drone descends; if an open flame is confirmed, firefighting begins. This invention achieves multi-functionality for the drone, enabling both inspection and firefighting, significantly improving the drone's intelligent control level. It enables automatic inspection, detection, judgment, and firefighting, solving the problem of traditional manual inspection and firefighting, and greatly improving the efficiency and quality of forest fire prevention.

[0043] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description, claims, and drawings.

[0044] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0045] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0046] Figure 1 This is a flowchart illustrating the method for using drones in forest fire prevention and control in Embodiment 1 of the present invention.

[0047] Figure 2 This is a diagram illustrating the process of the UAV confirming the inspection route in Embodiment 2 of the present invention.

[0048] Figure 3 This is a diagram illustrating the process of analyzing a flame image from an image in Embodiment 4 of the present invention.

[0049] Figure 4 This is a flowchart of the image processing module in Embodiment 5 of the present invention for converting a color image to grayscale.

[0050] Figure 5 This is a diagram illustrating the process of image segmentation of the first grayscale image by the image processing module in Embodiment 6 of the present invention.

[0051] Figure 6 This is a diagram illustrating the process of wirelessly transmitting color images captured and extinguished during fire suppression to the control platform in Embodiment 9 of the present invention. Detailed Implementation

[0052] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0053] The terminology used in the embodiments of this application is for the purpose of describing particular embodiments only and is not intended to limit the embodiments of this application. The singular forms “a,” “the,” and “the” used in the embodiments of this application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any or all possible combinations of one or more of the associated listed items.

[0054] In the following description, when referring to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims. In the description of this application, it should be understood that the terms "first," "second," "third," etc., are used only to distinguish similar objects and are not necessarily used to describe a specific order or sequence, nor should they be construed as indicating or implying relative importance. Those skilled in the art can understand the specific meaning of the above terms in this application according to the specific circumstances.

[0055] Example 1

[0056] like Figure 1 As shown, this embodiment of the invention provides a method for using drones for inspection and firefighting in forest fire prevention, including the following steps:

[0057] S100: The drone acquires an electronic map of the area to be inspected and confirms the inspection route based on the electronic map and flight time.

[0058] S200: After the drone takes off, the infrared camera captures a color image of the inspection area and analyzes the flame image from the color image. If there is no flame, the inspection continues; if there is a flame, an alarm is issued and the fire is extinguished.

[0059] S300: The captured and extinguished color images are wirelessly transmitted to the control platform to confirm that the fire has been extinguished.

[0060] The working principle and beneficial effects of the above technical solution are as follows: First, the drone acquires an electronic map of the area to be inspected and confirms the inspection route based on the electronic map and flight time. Second, after the drone takes off, its infrared camera captures color images of the inspection area and analyzes the flame images from the color images. If there are no flames, the inspection continues; if there are flames, an alarm is issued and firefighting is initiated. Finally, the captured and extinguished color images are wirelessly transmitted to the control platform to confirm the completion of firefighting. The drone's infrared camera captures images of the inspection area, and the number of sharp corners is obtained by analyzing the images. It is determined whether the number of sharp corners reaches a preset threshold. If the threshold is reached, the drone descends. If it confirms that there is an open flame, it begins firefighting. This solution realizes the multi-functionality of the drone, enabling it not only to inspect but also to extinguish fires, greatly improving the intelligent control level of the drone. It achieves automatic inspection, automatic detection, automatic judgment, and automatic firefighting, solving the problem of traditional manual inspection and firefighting, and greatly improving the efficiency and quality of forest fire prevention.

[0061] Example 2

[0062] like Figure 2 As shown, based on Example 1, the process of confirming the inspection route using a drone provided in this embodiment of the invention includes:

[0063] S101: The drone downloads an electronic map of the area to be inspected via the Internet. The electronic map includes the boundaries, topography, and vegetation distribution of the area to be inspected.

[0064] S102: The area to be inspected is divided into several nodes according to the topography and vegetation distribution. The nodes include important nodes and non-important nodes. Important nodes are nodes that are prone to fire, and non-important nodes are nodes that are not prone to fire.

[0065] S103: The drone plans its inspection route within the safe flight time according to the preset inspection path rules. The preset inspection path rules include: not exceeding the boundary of the area to be inspected, the proportion of important nodes being 20% ​​higher than that of non-important nodes, and reserving time for the drone to reach the next charging station.

[0066] S104: The drone inspects the area to be inspected according to the planned inspection route.

[0067] The working principle and beneficial effects of the above technical solution are as follows: First, the drone downloads an electronic map of the area to be inspected via the internet. This map includes the boundaries, topography, and vegetation distribution of the area. Second, the area is divided into several nodes based on the topography and vegetation distribution. These nodes include important and unimportant nodes. Important nodes are those prone to fire, while unimportant nodes are those less likely to be fire-prone. Then, the drone plans its inspection route within a safe flight time according to preset inspection path rules. These rules include: not exceeding the boundaries of the area to be inspected, ensuring that the proportion of important nodes is 20% higher than that of unimportant nodes, and allowing time for the drone to reach the next charging station. Finally, the drone inspects the area according to the planned inspection route. The forest area is divided into nodes for inspection, allowing drones to focus their inspections and minimize the probability of fires. Operators use historical fire data to differentiate between fire-prone and fire-resistant areas, setting a ratio of important to non-important nodes in each inspection route. This ensures effective forest fire prevention inspections, making them more planned and eliminating the time and labor costs of manual inspections. The intelligent design of the drone inspection routes undoubtedly enhances the drone's control capabilities and makes its application more practical. Many factors, such as boundaries and flight time, are considered in the planning of the inspection routes to ensure both drone safety and optimal inspection results. Through the planning of inspection paths, the effectiveness of forest fire prevention is improved.

[0068] Example 3

[0069] Based on Example 2, the preset inspection path rules provided in this embodiment of the invention further include: the UAV dynamically optimizes the path based on obstacles, flight path deviation, and UAV energy consumption, specifically including an obstacle collision function, a flight path deviation function, and a UAV energy consumption efficiency function, and optimizes the dynamic path G based on the obstacle collision function, flight path deviation function, and UAV energy consumption efficiency function. u The plan;

[0070] obstacle collision function A risk,uTo characterize the risk of collisions between dynamic paths and forest obstacles, the circumcircle of the forest obstacle is used as a mathematical model to describe the forest obstacle, and the expression is:

[0071]

[0072] Where risk represents an obstacle, u identifies the label of the dynamic path, u = 1, 2, 3, ..., M, M represents the total number of dynamic paths, and C u This indicates the detection results for forest obstacles, with obstacle C present. u =1, no obstacle C u =0, σ represents the standard deviation of forest obstacle collision risk, which determines the effective range of forest obstacle collision detection, and k represents the correction coefficient of dynamic path;

[0073] Flight path offset function B guid,u The result is obtained by integrating the difference between the dynamic path and the flight centerline:

[0074]

[0075] Where guid represents the flight path offset, l u (x) represents a function for dynamic paths, l lane The x represents the positional offset in the direction of the flight centerline normal; x represents the arc length along the flight centerline. end x represents the end of the arc length along the flight centerline. sta Indicates the beginning of the arc length along the flight centerline;

[0076] Effective energy consumption function E of drones u The expression is:

[0077]

[0078]

[0079]

[0080]

[0081] Where Q1 represents the internal power consumption of the drone's battery, C represents the air drag coefficient, S represents the cross-sectional area of ​​the drone, ρ represents the air density, and L(m,n) represents the distance between adjacent coordinate sampling points during the drone's flight. V represents the power conversion efficiency of the drone motor. i-1 v represents the flight speed of the drone at sampling point i-1. iQ1 represents the flight speed of the UAV at sampling point i; Q2 represents the UAV motor loss; η represents the UAV motor efficiency; P represents the UAV motor rated power; Q3 represents the lateral air resistance loss; F represents the lateral air resistance; v represents the flight speed; α represents the fuselage sideslip angle; E u Let Q represent the effective energy consumption of the drone, Q represent the total energy of the drone when fully charged, c represent the correlation coefficient between the fuselage sideslip angle and the drone's turning angle, and θ represent the drone's turning angle, where θ∈1,…,N;

[0082] G u =ω1A risk,u +ω2B guid,u +ω3E u

[0083] Where ω1 represents the obstacle collision function A risk,u The weights, ω2 represents the flight path offset function C guid,u The weights, ω3, represent the effective energy consumption function E of the UAV. u The weight.

[0084] The working principle and beneficial effects of the above technical solution are as follows: The above solution optimizes the dynamic path of the UAV during flight by calculating the obstacle collision function, flight path offset function, and UAV energy consumption effective function, thereby realizing dynamic path G based on the obstacle collision function, flight path offset function, and UAV energy consumption effective function. u The planning process includes: taking into account potential obstacles, route deviations, and effective energy consumption during drone forest fire patrols; optimizing alternative routes to make forest fire patrols more efficient, ensuring both efficiency and equipment safety and reliability; considering the complex environment of forests, with mountains, lowlands, high-altitude areas, and birds posing objective risks to drone flight, improving flight safety; and optimizing patrol routes by calculating drone deviations to ensure patrol quality, save on drone energy consumption, and extend patrol time and area, as drones are not always ready for patrols when fully charged. The energy consumption of drones in non-operational states is calculated to obtain the effective energy required for operation. When the effective energy is below a threshold, the system is recharged to ensure the efficiency of forest fire patrols.

[0085] Example 4

[0086] like Figure 3 As shown, based on Example 1, the process of analyzing a flame image from an image provided by this embodiment of the invention includes:

[0087] S201: The infrared camera captures a color image of the inspection area, and the image processing module of the infrared camera converts the color image to grayscale to obtain a first grayscale image.

[0088] S202; The image processing module performs image segmentation on the first grayscale image to obtain a flame mask image corresponding to the first grayscale image, removes the background of the flame mask image, and obtains a second grayscale image with the background removed.

[0089] S203: The image processing module scans the second grayscale image, labels white pixels as 0, and when it encounters a black pixel, it determines whether to label the current pixel based on whether the surrounding pixels are labeled. If labeled, it records it as 1, thus obtaining the labeled third grayscale image.

[0090] S204: The image processing module scans the third grayscale image again and counts the number of sharp corners that are not less than 8. If there are not less than 8 sharp corners, it is determined that the third grayscale image contains a flame image, that is, there is an open flame in the inspection area, which could cause a fire.

[0091] S205: Open flame present. The drone's controller receives an open flame command from the image processing module, issues an alarm, and activates the fire extinguishing grenade launching system to extinguish the fire. After launch, the infrared camera re-captures a color image to determine if a flame image exists. If no flame image is found, the drone continues its inspection; if a flame image is found, the drone continues its fire extinguishing actions.

[0092] The working principle and beneficial effects of the above technical solution are as follows: First, the infrared camera captures a color image of the inspection area. The image processing module of the infrared camera converts the color image to grayscale to obtain a first grayscale image. Second, the image processing module performs image segmentation on the first grayscale image to obtain a flame mask image corresponding to the first grayscale image. The background of the flame mask image is removed to obtain a second grayscale image with the background removed. Third, the image processing module scans the second grayscale image, marking white pixels as 0. When encountering a black pixel, it determines whether to mark the black pixel based on whether the surrounding pixels are marked. If marked, it is recorded as 1, resulting in a marked third grayscale image. Then, the image processing module scans the third grayscale image again, counting the number of sharp corners not less than 8. If there are at least 8 sharp corners, it is determined that the third grayscale image contains... The system includes images of flames, indicating the presence of open flames in the inspected area, which could potentially lead to a fire. Upon the presence of an open flame, the drone's controller receives a fire command from the image processing module, issues an alarm, and activates the fire extinguishing grenade launching system. After launch, the infrared camera re-captures a color image to determine if a flame image is present. If no flame image is found, the inspection continues; if a flame image is found, the fire extinguishing action continues. Alternatively, by capturing and processing color images to obtain grayscale images, these images are labeled, and the number of sharp corners is counted to determine the presence of a flame image. This simple and fast approach, relying solely on the presence of a flame image, and executed independently by the drone without the need for collaboration, enables rapid detection and handling of open flames, significantly improving the efficiency of fire suppression and minimizing the probability of fires.

[0093] Example 5

[0094] like Figure 4 As shown, based on Embodiment 4, the image processing module provided in this embodiment of the invention performs grayscale conversion of a color image, including:

[0095] S2011: Extract the R, G, and B channels of the color image and vectorize them into a column vector; perform preliminary grayscale conversion.

[0096] S2012: Obtain the neighborhood subjective perceived brightness by unilateral filtering of the initially grayscale color image;

[0097] S2013: Based on the difference between the obtained subjective perceived brightness in the neighborhood and the actual light intensity, the local contrast of the initially grayscale color image is linearly adjusted to obtain the first grayscale image.

[0098] The working principle and beneficial effects of the above technical solution are as follows: First, the R, G, and B channels of the color image are extracted and vectorized into a column vector; preliminary grayscale conversion is then performed; then, the preliminary grayscale color image is unilaterally filtered to obtain the neighborhood subjective perceived brightness; finally, based on the difference between the obtained neighborhood subjective perceived brightness and the actual light intensity, the local contrast of the preliminary grayscale color image is linearly adjusted to obtain the first grayscale image; by performing local contrast linear adjustment on the preliminary grayscale color image based on the difference between the obtained neighborhood subjective perceived brightness and the actual light intensity, local enhancement of the grayscale image is achieved, which can further improve the clarity of the first grayscale image, better conform to the actual situation of the forest inspection area, help improve the recognition accuracy of flame images, and improve the accuracy of forest fire judgment.

[0099] Example 6

[0100] like Figure 5 As shown, based on Embodiment 4, the image processing module provided in this embodiment of the invention performs image segmentation on the first grayscale image, including:

[0101] S2021: The image processing module inputs the training color image of the inspection area in the preset training sample into the preset network model to determine the predicted color image mask corresponding to the training color image; the training color image contains the flame area.

[0102] S2022: Obtain the flame edges in the predicted color image mask, and determine the weight map corresponding to the predicted color image mask based on the flame edges;

[0103] S2023: The preset network model is trained based on the weight map, the predicted color image mask, and the flame image mask corresponding to the training color image to obtain the first grayscale image segmentation model;

[0104] S2024: Input the first grayscale image into the first grayscale image segmentation model to determine the flame mask image corresponding to the first grayscale image.

[0105] The working principle and beneficial effects of the above technical solution are as follows: First, the image processing module inputs the training color image of the inspection area in the preset training samples into the preset network model to determine the predicted color image mask corresponding to the training color image; the training color image contains the flame region; second, the flame edge in the predicted color image mask is obtained, and the weight map corresponding to the predicted color image mask is determined based on the flame edge; then, the preset network model is trained based on the weight map, the predicted color image mask, and the flame image mask corresponding to the training color image to obtain the first grayscale image segmentation model; finally, the first grayscale image is input into the first grayscale image segmentation model to determine the flame mask corresponding to the first grayscale image; by training the preset network model with the weight map, the predicted color image mask, and the flame image mask corresponding to the training color image, the first grayscale image segmentation model is obtained, realizing the segmentation of the first grayscale image, obtaining the flame mask, improving the clarity of the first grayscale image, and providing an accurate image for flame image recognition.

[0106] Example 7

[0107] Based on Example 4, the embodiment of the present invention provides a method for removing the background of a flame mask image to obtain a second grayscale image with the background removed, including:

[0108] Color sampling is performed on the main area of ​​the flame mask image;

[0109] The range of sampled colors is expanded based on the color tolerance value to obtain the retained color range;

[0110] The flame mask image is processed to remove the background based on the preserved color range, resulting in a second grayscale image with the background removed.

[0111] The working principle and beneficial effects of the above technical solution are as follows: First, the main area of ​​the flame mask image is sampled for color; then, the range of the sampled colors is expanded according to the color tolerance value to obtain the retained color range; finally, the flame mask image is processed to remove the background according to the retained color range to obtain a second grayscale image with the background removed. By removing the background from the flame mask image according to the retained color range, the influence of the background image on the flame image recognition is removed, which helps to improve the quality of drone inspection in forest fire prevention, realize early detection of fires, and enable early handling, thus ensuring the safety of the forest ecosystem.

[0112] Example 8

[0113] Based on Example 4, the present invention provides a method for determining whether to label a pixel by judging whether surrounding pixels are labeled, including:

[0114] Process the first row of the second grayscale image. If the pixels of the second grayscale image to the left are not labeled, then label the first black pixel encountered. Otherwise, set it to the same label as the pixel to the left of the second grayscale image. Next, process the first column of the second grayscale image. If the pixels at the top of the first column are not labeled, then label the first black pixel encountered. Otherwise, set it to the same label as the pixel above.

[0115] The working principle and beneficial effects of the above technical solution are as follows: The above solution processes the first row of the second grayscale image. If the pixels of the second grayscale image on the left are not labeled, the first black pixel encountered is labeled; otherwise, it is set to the same label as the pixel on the left. Then, the first column of the second grayscale image is processed. If the pixels at the top of the first column are not labeled, the first black pixel encountered is labeled; otherwise, it is set to the same label as the pixel above. The image processing module scans the second grayscale image, labels white pixels as 0, and when a black pixel is encountered, it determines whether to label the pixel based on whether the surrounding pixels are labeled. If labeled, it is recorded as 1, resulting in a labeled third grayscale image. This achieves the statistical analysis of the number of sharp corners in the second grayscale image, laying a technical and data foundation for accurate identification of flame images.

[0116] Example 9

[0117] like Figure 6 As shown, based on Example 1, the process of wirelessly transmitting color images of fire capture and extinguishing to the control platform provided in this embodiment of the invention includes:

[0118] S301: The control platform receives and stores color images captured by the drone and used for firefighting.

[0119] S302: If the control platform detects that the fire extinguishing has not met the set standard, it will control the UAV fire extinguishing projectile launching system to continue executing the fire extinguishing command;

[0120] S303: After confirming that the fire has been extinguished, issue a command for the drone to continue its inspection.

[0121] The working principle and beneficial effects of the above technical solution are as follows: First, the control platform receives and stores the color images captured by the drone and the fire extinguishing process; second, if the control platform finds that the fire extinguishing has not reached the set standard, it controls the drone's fire extinguishing projectile launching system to continue executing the fire extinguishing command; finally, after confirming that the fire has been extinguished, it issues a command for the drone to continue patrolling; this achieves intelligent supervision of the drone, ensures the quality of patrolling and fire extinguishing, and provides a barrier for the ecological environment safety of the forest.

[0122] Example 10:

[0123] Based on Example 4, the fire extinguishing projectile launching system provided in this embodiment of the invention includes: a fire extinguishing agent filling module, a propulsion module, and an activation module;

[0124] The extinguishing agent filling module is responsible for filling the extinguishing bombs.

[0125] The propulsion module is responsible for ejecting the fire extinguishing bombs from the drone's magazine, ensuring that the fire extinguishing bombs accurately reach the surface of the open flame.

[0126] The activation module is responsible for launching the extinguishing agent inside the fire extinguishing bomb, which then adheres directly to the surface of the open flame.

[0127] The working principle and beneficial effects of the above technical solution are as follows: The fire extinguishing agent filling module is responsible for filling the fire extinguishing bomb; the propulsion module is responsible for ejecting the fire extinguishing bomb from the UAV's magazine, ensuring that the fire extinguishing bomb accurately reaches the surface of the open flame; the ignition module is responsible for launching the fire extinguishing agent inside the fire extinguishing bomb, which directly adheres to the surface of the open flame; the solution achieves self-controlled launch of the fire extinguishing bomb. Upon receiving the control command from the UAV, the propulsion module immediately works to launch the fire extinguishing bomb, and under the action of the ignition module, the fire extinguishing agent accurately reaches the surface of the open flame to extinguish the fire, eliminating the tediousness of manual fire extinguishing and improving the safety of fire extinguishing.

[0128] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for implementing the patrol and fire extinguishing of unmanned aerial vehicles in forest fire prevention, characterized in that, The method comprises the following steps: The unmanned aerial vehicle acquires an electronic map of the area to be inspected, and confirms the inspection route according to the electronic map and the flight time; After the unmanned aerial vehicle takes off, the infrared camera shoots a color image of the inspection area, and analyzes a flame image from the color image; if there is no flame, the inspection continues; if there is a flame, an alarm is sent out, and fire extinguishing is performed; The color image of shooting and fire extinguishing is wirelessly transmitted to the control platform to confirm that the fire extinguishing is completed; The process of confirming the inspection route by the unmanned aerial vehicle comprises: The unmanned aerial vehicle downloads the electronic map of the area to be inspected through the Internet; The area to be inspected is divided into a plurality of nodes according to the topography and vegetation distribution, wherein the nodes include important nodes and non-important nodes; The unmanned aerial vehicle plans the inspection route according to a preset inspection path rule within the safe flight time; The unmanned aerial vehicle performs the inspection of the area to be inspected according to the planned inspection route; The preset inspection path rule comprises: the boundary of the area to be inspected cannot be exceeded, the proportion of important nodes is 20% more than that of non-important nodes, and the unmanned aerial vehicle reserves the time to arrive at the next charging station; The preset inspection path rule further comprises: dynamic path optimization of the unmanned aerial vehicle according to obstacles, flight route deviation degree and unmanned aerial vehicle energy consumption, specifically comprising obstacle collision function, flight route deviation function and unmanned aerial vehicle energy consumption effective function, dynamic path planning according to the obstacle collision function, the flight route deviation function and the unmanned aerial vehicle energy consumption effective function ; Obstacle collision function The risk of collision between dynamic path and forest obstacle is characterized, and a circumscribed circle of the forest obstacle is taken as a mathematical model for describing the forest obstacle, and an expression is: wherein, risk represents an obstacle, a label identifying a dynamic path, M represents the total number of dynamic paths, represents the detection result of forest obstacles, with obstacles without obstacles , represents the standard deviation of the collision risk of forest obstacles, determines the effective range of forest obstacle collision detection, and k represents the correction coefficient of the dynamic path. Flight path offset function The integral of the difference between the dynamic path and the flight centerline is calculated as follows: wherein guid represents a flight path offset, a function representing a dynamic path, a position offset representing a flight center line normal direction; an arc length along a flight center line, an end of the arc length along the flight center line, a start of the arc length along the flight center line; Drone energy consumption efficiency function The expression is: wherein, represents the UAV battery internal loss, C represents the air resistance coefficient, S represents the UAV cross-sectional area, represents the air density, represents the distance of the UAV in the adjacent coordinate sampling points in the navigation, represents the electric energy conversion efficiency of the UAV motor, represents the flight speed of the UAV at the sampling point i-1, represents the flight speed of the UAV at the sampling point i; represents the UAV motor loss, represents the UAV motor efficiency, represents the rated power of the UAV motor, represents the lateral air resistance loss, represents the lateral air resistance, represents the flight speed, represents the body side slip angle, represents the effective value of the UAV energy consumption, represents the total energy of the UAV in the full power state, represents the body side slip angle and the UAV rotation angle correlation coefficient, the UAV rotation angle, ; wherein, represents a weight of the obstacle collision function , represents a weight of the flight route deviation function , represents a weight of the UAV effective energy consumption function .

2. The method for implementing the patrol and fire extinguishing of the unmanned aerial vehicle in the forest fire prevention according to claim 1, characterized in that, The electronic map comprises the boundary, topography and vegetation distribution of the area to be inspected; the important nodes are nodes prone to fire, and the non-important nodes are nodes not prone to fire. 3.The method of claim 1, wherein the unmanned aerial vehicle is configured to fly to the fire location and spray the fire-extinguishing liquid on the fire location. The preset inspection path rule further comprises: the unmanned aerial vehicle performs dynamic path optimization according to obstacles, flight route deviation degree and energy consumption of the unmanned aerial vehicle, specifically comprising an obstacle collision function, a flight route deviation function and an unmanned aerial vehicle energy consumption effective function, and the dynamic path is planned according to the obstacle collision function, the flight route deviation function and the unmanned aerial vehicle energy consumption effective function.

4. The method for implementing the patrol and fire extinguishing of the unmanned aerial vehicle in the forest fire prevention according to claim 1, characterized in that, The process of analyzing the flame image from the image comprises: The infrared camera shoots a color image of the inspection area, and the image processing module of the infrared camera performs gray scale processing on the color image to obtain a first gray scale image; The image processing module performs image segmentation on the first gray scale image to obtain a flame mask image corresponding to the first gray scale image, removes the background of the flame mask image to obtain a second gray scale image with the background removed; The image processing module scans the second gray scale image, labels the points with white pixels as 0, and when a black pixel is encountered, determines whether the surrounding pixels are labeled to decide whether to label the pixel, if yes, marks it as 1, to obtain a third gray scale image with labels; The image processing module scans the third gray scale image again, counts the number of sharp corners not less than 8, and if there are sharp corners not less than 8, it is determined that the third gray scale image contains a flame image, that is, there is an open fire in the inspection area, which can cause a fire; If there is an open fire, the controller of the unmanned aerial vehicle receives the open fire instruction sent by the image processing module, sends out an alarm, starts the fire extinguishing bomb launching system to extinguish the fire, and after launching, the infrared camera re-shoots the color image to determine whether there is a flame image, and if there is no flame image, the inspection continues, and if there is a flame image, the fire extinguishing action is continued.

5. The method for implementing the patrol and fire extinguishing of the unmanned plane in the forest fire prevention according to claim 4, characterized in that, The gray scale processing of the color image by the image processing module comprises: Extracting R, G and B three channel vectors of the color image to become a column vector, and performing preliminary gray scale processing; Unilateral filtering of the color image subjected to the preliminary gray scale processing to obtain the neighborhood subjective brightness; According to the difference between the obtained neighborhood subjective perceived brightness and the actual light intensity, the preliminarily gray-scale color image is locally linearly adjusted in contrast to obtain a first gray-scale image.

6. The method for implementing the patrol and fire extinguishing of the unmanned plane in the forest fire prevention according to claim 4, characterized in that, The image processing module performs image segmentation on the first gray-scale image, and the process includes: The image processing module inputs the training color image of the inspection area in the training sample into the preset network model to determine a predicted color image mask corresponding to the training color image; the training color image contains a flame area; Obtain the flame edge in the predicted color image mask, and determine a weight map corresponding to the predicted color image mask based on the flame edge; Train the preset network model based on the weight map, the predicted color image mask, and a flame image mask corresponding to the training color image to obtain a first gray-scale image segmentation model; Input the first gray-scale image into the first gray-scale image segmentation model to determine a flame mask corresponding to the first gray-scale image.

7. The method for implementing the patrol and fire extinguishing of the unmanned plane in the forest fire prevention according to claim 4, characterized in that, Remove the background of the flame mask to obtain a second gray-scale image with the background removed, including: Color sampling is performed on the main body area of the flame mask; According to the color tolerance value, the sampled color is expanded in range to obtain a reserved color range; According to the reserved color range, the flame mask is processed to remove the background to obtain the second gray-scale image with the background removed. 8.The method of claim 1, wherein the unmanned aerial vehicle is configured to fly to a location of a fire in the forest based on the information received from the sensor. The process of wirelessly transmitting the color images of shooting and fire extinguishing to the control platform includes: The control platform receives the color images of shooting and fire extinguishing by the unmanned aerial vehicle and stores them; If the control platform finds that the fire extinguishing does not reach the set standard, the control platform controls the unmanned aerial vehicle fire extinguishing bomb launching system to continue executing the fire extinguishing instruction; After confirming that the fire extinguishing is completed, the control platform issues an instruction for the unmanned aerial vehicle to continue inspection.

Images