Fire source positioning method and system based on combination of thermal imaging and laser ranging
By combining thermal imaging and laser ranging technologies, a fire source spatial positioning system was constructed. By utilizing Lie algebra fusion and adversarial identification models in symplectic geometric space, high-precision fire source positioning was achieved, solving the accuracy and adaptability problems of traditional fire source positioning technology in complex environments.
Patent Information
- Application Number
- CN202511484221.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-10-17
- Publication Date
- 2026-01-16
AI Technical Summary
Traditional fire source location technology is susceptible to environmental interference and cannot achieve continuous all-round scanning in large space scenarios, resulting in low positioning accuracy and inability to accurately obtain the distance between the fire source and the detection device.
By combining thermal imaging and laser ranging technologies, the approximate location of the fire source can be quickly determined using a thermal imager, and the distance to the fire source can be accurately measured using a laser rangefinder. This constructs a complete spatial positioning system for the fire source. By combining Lie algebra fusion in symplectic geometric space and adversarial recognition model to dynamically adjust the covariance weight of laser ranging, the system maps the fire source to its spatial coordinates in cylindrical coordinates.
It significantly improves the accuracy and robustness of fire source location, maintains stable location performance in complex environments, and enhances the accuracy and environmental adaptability of fire source location.
Smart Images
Figure CN121348346A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the field of fire prevention and control, more particularly to a fire source positioning method and system based on the combination of thermal imaging and laser ranging. BACKGROUND
[0002] In the field of fire prevention and control, accurate positioning of the fire source is the key to reducing fire loss, but the traditional fire source positioning technology has obvious shortcomings. The positioning method based on infrared or ultraviolet sensors is easily disturbed by smoke, dust and other heat sources in the environment, resulting in distorted detection signals, making it difficult to accurately identify the location of the fire source, and may also result in false positives or false negatives. At the same time, such technology can only achieve local area monitoring and cannot meet the needs of continuous scanning in large space scenarios, further reducing the positioning accuracy and causing great inconvenience to the early disposal of fires.
[0003] The emergence of thermal imaging technology provides a new direction for fire source detection. It converts temperature differences into visual thermal imaging by capturing infrared radiation from the surface of an object, can penetrate smoke, dust and other obstructions, quickly identify high-temperature abnormal areas, accurately lock the location of the fire source, and effectively solve the problem of environmental interference of traditional sensors, showing significant advantages in early fire detection. However, thermal imaging technology can determine the approximate direction of the fire source, but it is difficult to accurately obtain the distance between the fire source and the detection device, and cannot form a complete fire source spatial coordinate, limiting the further improvement of positioning accuracy.
[0004] Laser ranging technology has the advantage of high-precision ranging. It can accurately calculate the distance between the target and the device by emitting laser pulses, recording the round-trip time of the pulses, and combining the speed of light. It has been widely used in the fields of construction engineering and topographic surveying. Therefore, how to organically combine thermal imaging technology and laser ranging technology to improve the positioning accuracy of the fire source still needs further research by technical personnel in the field. SUMMARY
[0005] Therefore, the present application provides a fire source positioning method and system based on the combination of thermal imaging and laser ranging, which combines thermal imaging technology and laser ranging technology, uses thermal imaging technology to quickly lock the approximate direction of the fire source, and then uses laser ranging technology to accurately measure the distance of the fire source, thereby constructing a complete fire source spatial positioning system, effectively making up for the shortcomings of traditional technology and single technology, greatly improving the positioning accuracy of the fire source, and providing more reliable technical support for fire prevention and control.
[0006] In order to achieve the above purpose, the present application adopts the following technical solutions: A fire source positioning method based on the combination of thermal imaging and laser ranging, comprising the following steps: S1: Obtain the thermal imaging image, the original image and the point cloud data of the target area by the thermal imager, the high-definition camera and the laser range finder carried by the mobile carrier, identify the fire source candidate area, construct a cylindrical coordinate system with one end of the center axis of the fire source candidate area as the origin, and generate a three-dimensional model containing the space posture of the fire source and obstacle information; S2: Adjust the shooting angle of the thermal imager in combination with the three-dimensional model, plan a surrounding path containing effective detection points, control the mobile carrier to move along the path, obtain the distance data of the mobile carrier and the fire source candidate area in the cylindrical coordinate system by the laser range finder, and generate a three-dimensional space constraint matrix; S3: Perform Lie algebra fusion on the thermal imaging signal features and the three-dimensional space constraint matrix in the symplectic geometry space, dynamically adjust the laser ranging covariance weight by the adversarial identification model, and map it to the spatial coordinates of the fire source in the cylindrical coordinate system. S4: Calculate the Riemannian manifold distance between the positioning coordinates and the pre-path data, and reposition after resampling the data and updating the model weight when the distance exceeds the threshold.
[0007] Optionally, it also includes positioning verification and correction, specifically: calculate the Riemannian manifold distance between the fire source positioning coordinates and the mobile carrier pre-path data in the cylindrical coordinate system, the Riemannian manifold distance is calculated by the Frobenius norm of the three-dimensional space constraint matrix and the determinant of the inertial feature covariance matrix; if the distance exceeds the preset threshold, trigger the fire source thermal signal detection and space modeling step to resample the data, update the adversarial identification model weight coefficient and reposition.
[0008] Optionally, the inertial feature covariance matrix is constructed by the angular velocity and acceleration data obtained by the inertial measurement unit carried by the mobile carrier; when the Riemannian manifold distance exceeds the threshold, the preprocessed thermal imaging image needs to be preprocessed, the preprocessing includes removing noise and unifying temperature scale, and the laser ranging data needs to be measured multiple times to take the average value to reduce the error.
[0009] Optionally, in S1, it specifically includes the following steps: identifying the high-temperature abnormal area as the fire source candidate area based on the thermal imaging image, extracting the fire source candidate area edge geometric descriptor from the original image using a deformable convolution kernel, the deformation parameters of the deformable convolution kernel are adjusted by the reflectivity gradient field of the fire source candidate area, and the gradient direction of the reflectivity gradient field is determined by the relative position of the mobile carrier and the fire source candidate area; simultaneously collect the point cloud data of the fire source candidate area and the surrounding environment by the laser range finder carried by the mobile carrier, generate a three-dimensional point cloud model containing the space posture of the fire source candidate area and the surrounding obstacle information in combination with the high-definition camera image, construct a cylindrical coordinate system with one end of the center axis of the fire source candidate area as the origin, and the three axes of the cylindrical coordinate system are respectively the vertical direction of the origin and the center axis of the fire source candidate area, the circumferential direction rotating around the center axis, and the axial direction coinciding with the center axis.
[0010] Optionally, in S3, the generator of the adversarial identification model tensor contracts the thermal signal features with the three-dimensional space constraint matrix to generate the symplectic geometry structure feature basis; the discriminator encodes the pre-path data into a dual quaternion motion constraint chain, and realizes the adversarial iteration of the generator and the discriminator through Lie algebra bracket operation, and the iteration termination condition is determined by the exponential moving average of the laser ranging covariance matrix trace.
[0011] Optionally, the mobile carrier is a drone or a ground mobile robot; when it is a drone, the axial height data of the cylindrical coordinate system is calibrated in combination with the GPS positioning information of the drone; when it is a ground mobile robot, the horizontal direction data of the cylindrical coordinate system is corrected in combination with the wheel odometer data of the robot.
[0012] Optionally, in S3, the angle interval constraint is that the central angle of adjacent point positions in the circumferential direction of the cylindrical coordinate system is ≤30°, and the distance constraint is that the straight line distance of the point position from the outer surface of the fire source candidate area is in the range of 1-3 meters; the space range of the stereoscopic detection box is ±0.5 meters in the distance direction, ±5° in the circumferential direction, and ±0.3 meters in the axial direction of the cylindrical coordinate system, and when the volume overlap rate is ≥15%, it is determined that the supplementary point position is invalid.
[0013] Optionally, in S3, during the Lie algebra fusion process of the symplectic geometry space, the angular velocity and acceleration of the mobile carrier inertial data are converted into Lie group elements, and the affine transformation of the connection operator on the symplectic geometry space and the thermal signal features and the laser ranging features establishes a differential homeomorphism relationship, realizing the cross-domain feature coupling of multi-modal data.
[0014] A fire source positioning system based on the combination of thermal imaging and laser ranging, comprising: A fire source thermal signal detection and space modeling module: used for acquiring target area thermal imaging, original image and point cloud data through a thermal imager, a high-definition camera and a laser range finder carried by a mobile carrier, identifying a fire source candidate area, constructing a cylindrical coordinate system with one end of the center axis of the fire source candidate area as the origin, and generating a three-dimensional model containing the space posture of the fire source and obstacle information; A shooting angle precise adjustment module: used for adjusting the shooting angle of the thermal imager in combination with the three-dimensional model, planning a surrounding path containing effective detection point positions; controlling the mobile carrier to move along the path, acquiring the distance data of the mobile carrier and the fire source candidate area in the cylindrical coordinate system through the laser range finder, and generating a three-dimensional space constraint matrix; A detection path planning and laser ranging module: used for performing Lie algebra fusion on the thermal imaging signal features and the three-dimensional space constraint matrix in the symplectic geometry space, dynamically adjusting the laser ranging covariance weight through the adversarial identification model, and mapping the space coordinates of the fire source in the cylindrical coordinate system; A fire source space positioning module: used for calculating the Riemannian manifold distance of the positioning coordinates and the pre-path data, resampling the data and updating the model weight when the threshold value is exceeded, and repositioning.
[0015] Optionally, it also includes a positioning verification and correction module: for calculating the Riemannian manifold distance of the fire source positioning coordinates and the moving carrier pre-path data in the cylindrical coordinate system, the Riemannian manifold distance is calculated by the Frobenius norm of the three-dimensional space constraint matrix and the determinant of the inertia characteristic covariance matrix; if the distance exceeds the preset threshold, trigger the fire heat signal detection and space modeling step to resample the data, update the weight coefficient of the adversarial identification model and reposition.
[0016] Through the above technical solution, compared with the prior art, the fire source positioning method and system based on the combination of thermal imaging and laser ranging provided by the present application have the following beneficial effects: first, with the help of multi-source data acquisition of thermal imager, high-definition camera and laser range finder, combined with cylindrical coordinate system modeling, the spatial posture and obstacle information of the fire source candidate area can be fully captured, laying a precise data foundation for subsequent positioning; second, by planning the surrounding detection path and generating a three-dimensional space constraint matrix, the effectiveness and spatial coverage of data acquisition are effectively improved, and the positioning deviation caused by a single perspective is reduced; third, the Lie algebra fusion of the symplectic geometry space and the dynamic weight adjustment of the adversarial identification model realize the deep cooperation of the thermal imaging signal features and the distance data, significantly improving the accuracy of the fire source spatial coordinate calculation; finally, based on the threshold judgment and dynamic resampling mechanism of the Riemannian manifold distance, the positioning deviation can be corrected in real time, ensuring stable positioning performance in complex environments, greatly improving the precision, robustness and environmental adaptability of fire source positioning, and providing more reliable technical support for fire monitoring, emergency rescue and other scenes. BRIEF DESCRIPTION OF DRAWINGS
[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced as follows. Obviously, the drawings in the following description are only embodiments of the present application, and those skilled in the art can obtain other drawings according to the provided drawings without creative labor.
[0018] Figure 1 The method flowchart provided by the present application is shown in the figure; Figure 2 The system structure schematic diagram provided by the present application is shown in the figure. DETAILED DESCRIPTION
[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0020] This invention discloses a fire source localization method based on a combination of thermal imaging and laser ranging, such as... Figure 1 As shown, it includes the following steps: S1: Using a thermal imager, high-definition camera and laser rangefinder mounted on a mobile carrier, thermal images, raw images and point cloud data of the target area are acquired, candidate fire source areas are identified, a cylindrical coordinate system with one end of the central axis of the candidate fire source area as the origin is constructed, and a three-dimensional model containing the spatial attitude and obstacle information of the fire source is generated. S2: Combine the 3D model to adjust the shooting angle of the thermal imager and plan a surrounding path containing effective detection points; control the mobile vehicle to move along the path, and obtain the distance data between the mobile vehicle and the candidate area of the fire source in the cylindrical coordinate system through the laser rangefinder to generate a 3D spatial constraint matrix. S3: The thermal imaging signal features and the three-dimensional spatial constraint matrix are fused with Lie algebra in the symplectic geometric space. The laser ranging covariance weights are dynamically adjusted by the adversarial recognition model and mapped to the spatial coordinates of the fire source in the cylindrical coordinate system. S4: Calculate the Riemannian distance between the positioning coordinates and the pre-path data. If the distance exceeds the threshold, resample the data, update the model weights, and then reposition.
[0021] More specifically, in this embodiment, the above steps are further refined as follows: The target area is scanned by a thermal imager and a high-definition camera mounted on a mobile platform to obtain thermal images and original images. Based on the thermal images, high-temperature anomaly areas are identified as candidate fire sources. Deformable convolution kernels are used to extract the edge geometric descriptors of the candidate fire sources from the original images. The deformation parameters of the deformable convolution kernels are adjusted by the reflectivity gradient field of the surface of the candidate fire sources, and the gradient direction of the reflectivity gradient field is determined by the relative position of the mobile platform and the candidate fire sources. Simultaneously, a laser rangefinder mounted on the mobile platform collects point cloud data of the candidate fire sources and the surrounding environment. Combined with the high-definition camera images, a three-dimensional point cloud model containing the spatial attitude of the candidate fire sources and information on surrounding obstacles is generated. A cylindrical coordinate system is constructed with one end of the central axis of the candidate fire sources as the origin. The three axes of the cylindrical coordinate system are the direction perpendicular to the central axis of the candidate fire sources, the circumferential direction of rotation around the central axis, and the axial direction coinciding with the central axis. The shooting angle fine adjustment step comprises the following steps: extracting the cylindrical coordinates of two end points of the center axis of the fire source candidate region according to the three-dimensional point cloud model, calculating the inclination angle of the axis and the horizontal plane, and coarsely adjusting the shooting angle of the thermal imager according to the angle; identifying the same calibration point from the thermal imaging images of different detection points, converting the pixel coordinates of the calibration point into physical coordinates of the image, calculating the deviation of the theoretical cylindrical coordinates of the calibration point in the three-dimensional point cloud model, generating an angle correction amount through Lie algebra operation in the symplectic geometry space, and finely adjusting the shooting angle of the thermal imager; The detection path planning and laser ranging step comprises the following steps: presetting detection points on the outer periphery of the fire source candidate region according to the cylindrical coordinate system, and screening and retaining the points in combination with the obstacle information in the three-dimensional point cloud model; planning and supplementing points through the angle interval constraint and distance constraint of adjacent retained points, verifying the volume overlap rate of the obstacles in the three-dimensional detection box constructed at the supplementing points, and planning a surrounding detection path after eliminating invalid supplementing points; controlling the mobile carrier to move along the path, emitting laser pulses to the fire source candidate region through the laser range finder at each point, recording the round-trip time of the pulses, calculating the distance data between the mobile carrier and the fire source candidate region in the cylindrical coordinate system in combination with the speed of light, generating a three-dimensional space constraint matrix, and the elements of the matrix representing the error covariance of each ranging point; The fire source space positioning step comprises the following steps: inputting the fire source candidate region thermal signal features of the thermal imaging image and the three-dimensional space constraint matrix of the laser ranging into a data processing unit, and performing Lie algebra fusion in the symplectic geometry space; dynamically adjusting the laser ranging covariance weight coefficient through the adversarial recognition model, calculating the covariant derivative of the geometric descriptor curvature and the inertial angular velocity of the mobile carrier, mapping the fused data into the space coordinates of the fire source candidate region in the cylindrical coordinate system, and realizing accurate positioning of the fire source.
[0022] In the embodiment, the positioning verification and correction also comprises the following steps: calculating the Riemannian manifold distance between the fire source positioning coordinates and the mobile carrier pre-path data in the cylindrical coordinate system, wherein the Riemannian manifold distance is calculated in combination with the Frobenius norm of the three-dimensional space constraint matrix and the determinant of the inertial feature covariance matrix; if the distance exceeds a preset threshold, triggering the fire source thermal signal detection and space modeling step to resample data, updating the adversarial recognition model weight coefficient, and repositioning.
[0023] Further, in the positioning verification and correction step, the inertial feature covariance matrix is constructed by the angular velocity and acceleration data obtained by the inertial measurement unit carried by the mobile carrier; when the Riemannian manifold distance exceeds the threshold, the reacquired thermal imaging image needs to be preprocessed, and the preprocessing includes removing noise and unifying temperature scale; the laser ranging data needs to be measured multiple times to take an average value to reduce the error.
[0024] Further, the inertia feature covariance matrix is constructed by the angular velocity and acceleration data obtained by the inertia measurement unit carried by the mobile carrier; when the Riemannian manifold distance exceeds the threshold value, the reacquired thermal image needs to be preprocessed, and the preprocessing includes removing noise, unifying temperature scale, and the laser ranging data needs to be measured multiple times to take the average value to reduce the error.
[0025] Further, in S1, specifically comprising the following steps: identifying a high-temperature abnormal area as a fire source candidate area based on the thermal image, extracting a fire source candidate area edge geometric descriptor from the original image using a deformable convolution kernel, the deformation parameters of the deformable convolution kernel are adjusted by the surface reflectivity gradient field of the fire source candidate area, and the gradient direction of the reflectivity gradient field is determined by the relative position of the mobile carrier and the fire source candidate area; synchronously collecting point cloud data of the fire source candidate area and the surrounding environment by a laser range finder carried by the mobile carrier, combining the high-definition camera image to generate a three-dimensional point cloud model containing the spatial pose of the fire source candidate area and the surrounding obstacle information, and constructing a cylindrical coordinate system with one end of the center axis of the fire source candidate area as the origin, and the three axes of the cylindrical coordinate system are the vertical direction of the origin and the center axis of the fire source candidate area, the circumferential direction rotating around the center axis, and the axial direction coinciding with the center axis.
[0026] Further, in S3, the generator of the adversarial identification model tensor contracts the thermal signal feature and the three-dimensional space constraint matrix to generate a symplectic geometric structure feature base; the discriminator encodes the pre-path data into a dual quaternion motion constraint chain, and realizes the adversarial iteration of the generator and the discriminator through Lie algebra bracket operation, and the iteration termination condition is determined by the exponential moving average of the trace of the laser ranging covariance matrix.
[0027] Further, the mobile carrier is a UAV or a ground mobile robot; when it is a UAV, the axial height data of the cylindrical coordinate system is calibrated in combination with the UAV GPS positioning information; when it is a ground mobile robot, the horizontal direction data of the cylindrical coordinate system is corrected in combination with the robot wheel odometer data.
[0028] Further, in S3, the angle interval constraint is that the central angle of adjacent points in the circumferential direction of the cylindrical coordinate system is ≤30°, and the distance constraint is that the straight line distance between the point and the outer surface of the fire source candidate area is within 1-3 meters; the spatial range of the stereo detection box is ±0.5 meters in the distance direction, ±5° in the circumferential direction, and ±0.3 meters in the axial direction of the cylindrical coordinate system, and when the volume overlap rate is ≥15%, it is determined that the supplementary point is invalid.
[0029] Further, in S3, during the Lie algebra fusion process of the symplectic geometric space, the angular velocity and acceleration of the mobile carrier inertia data are converted into Lie group elements, and the affine transformation of the connection operator on the symplectic geometric space and the thermal signal feature and the laser ranging feature establishes a differential homeomorphism relationship, realizing the cross-domain feature coupling of multi-modal data.
[0030] With Figure 1 Corresponding to the method shown, the application also discloses a fire source positioning system based on the combination of thermal imaging and laser ranging, which is used for positioning Figure 1 The implementation of the method, and the specific structure is as shown in Figure 2 The method comprises the following steps: The fire source thermal signal detection and space modeling module is used for acquiring the thermal imaging image, the original image and the point cloud data of the target area, identifying the fire source candidate area, constructing the cylindrical coordinate system with the center axis of the fire source candidate area as the origin, and generating the three-dimensional model containing the space posture of the fire source and the obstacle information through the thermal imager, the high-definition camera and the laser range finder carried by the mobile carrier. The shooting angle precision adjustment module is used for adjusting the shooting angle of the thermal imager in combination with the three-dimensional model, planning the surrounding path containing the effective detection point, controlling the mobile carrier to move along the path, acquiring the distance data of the mobile carrier and the fire source candidate area in the cylindrical coordinate system through the laser range finder, and generating the three-dimensional space constraint matrix. The detection path planning and laser ranging module is used for performing Lie algebra fusion on the thermal imaging signal features and the three-dimensional space constraint matrix in the symplectic geometry space, dynamically adjusting the laser ranging covariance weight through the adversarial identification model, and mapping the space coordinates of the fire source in the cylindrical coordinate system. The fire source space positioning module is used for calculating the Riemannian manifold distance of the positioning coordinates and the pre-path data, resampling the data and updating the model weight when the distance exceeds the threshold, and then repositioning.
[0031] Further, the positioning verification and correction module is further included, which is used for calculating the Riemannian manifold distance of the fire source positioning coordinates and the pre-path data of the mobile carrier in the cylindrical coordinate system, wherein the Riemannian manifold distance is calculated by the Frobenius norm of the three-dimensional space constraint matrix and the determinant of the inertia feature covariance matrix; if the distance exceeds the preset threshold, the fire source thermal signal detection and space modeling step is triggered to resample the data, update the adversarial identification model weight coefficient, and then reposition.
[0032] The embodiments in the specification are described in a progressive manner, and each embodiment focuses on the difference from other embodiments. The same and similar parts of each embodiment can be referred to each other. For the device disclosed by the embodiments, since it corresponds to the method disclosed by the embodiments, the description is relatively simple, and the related parts can be referred to the method part.
[0033] The foregoing description of the disclosed embodiments enables a person skilled in the art to make or use the application. Modifications of these embodiments will occur to persons of skill in the art, and that the appended claims are intended to cover all such modifications that do not depart from the true spirit and scope of the application. Therefore, the application is not limited to the embodiments shown but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A fire source positioning method based on the combination of thermal imaging and laser ranging, characterized in that, The method comprises the following steps: S1: Obtain the thermal imaging image, the original image and the point cloud data of the target area by the thermal imager, the high-definition camera and the laser range finder carried by the mobile carrier, identify the fire source candidate area, construct a cylindrical coordinate system with one end of the center axis of the fire source candidate area as the origin, and generate a three-dimensional model containing the space posture of the fire source and obstacle information; S2: Adjust the shooting angle of the thermal imager in combination with the three-dimensional model, plan a surrounding path containing effective detection points, control the mobile carrier to move along the path, and obtain the distance data of the mobile carrier and the fire source candidate area in the cylindrical coordinate system by the laser range finder, and generate a three-dimensional space constraint matrix; S3: Perform Lie algebra fusion on the thermal signal features and the three-dimensional space constraint matrix in the symplectic geometry space, dynamically adjust the laser ranging covariance weight by the adversarial identification model, and map the space coordinates of the fire source in the cylindrical coordinate system; S4: Calculate the Riemannian manifold distance between the positioning coordinates and the pre-path data, and when the distance exceeds the threshold, resample the data and update the model weight to reposition.
2. The fire source positioning method based on the combination of thermal imaging and laser ranging according to claim 1, characterized in that, Further comprising positioning verification and correction, specifically: calculate the Riemannian manifold distance between the fire source positioning coordinates and the mobile carrier pre-path data in the cylindrical coordinate system, the Riemannian manifold distance is calculated by the Frobenius norm of the three-dimensional space constraint matrix and the determinant of the inertial feature covariance matrix; if the distance exceeds the preset threshold, trigger the fire source thermal signal detection and space modeling step to resample the data, update the adversarial identification model weight coefficient and reposition.
3. The fire source positioning method based on the combination of thermal imaging and laser ranging according to claim 2, characterized in that, The inertial feature covariance matrix is constructed by the angular velocity and acceleration data obtained by the inertial measurement unit carried by the mobile carrier; when the Riemannian manifold distance exceeds the threshold, the thermal imaging image reacquired needs to be preprocessed, the preprocessing includes removing noise and unifying temperature scale, and the laser ranging data needs to be measured multiple times to take the average value to reduce the error.
4. The fire source positioning method based on the combination of thermal imaging and laser ranging according to claim 1, characterized in that, In S1, specifically comprising the following steps: identifying the high-temperature abnormal area as the fire source candidate area based on the thermal imaging image, extracting the fire source candidate area edge geometric descriptor from the original image using a deformable convolution kernel, the deformation parameters of the deformable convolution kernel are adjusted by the surface reflectivity gradient field of the fire source candidate area, and the gradient direction of the reflectivity gradient field is determined by the relative position of the mobile carrier and the fire source candidate area; simultaneously, the point cloud data of the fire source candidate area and the surrounding environment are collected by the laser range finder carried by the mobile carrier, and a three-dimensional point cloud model containing the space posture of the fire source candidate area and the surrounding obstacle information is generated combined with the high-definition camera image, a cylindrical coordinate system is constructed with one end of the center axis of the fire source candidate area as the origin, and the three axes of the cylindrical coordinate system are respectively the vertical direction of the origin and the center axis of the fire source candidate area, the circumferential direction rotating around the center axis, and the axial direction coinciding with the center axis.
5. The fire source positioning method based on the combination of thermal imaging and laser ranging according to claim 1, characterized in that, In S3, the generator of the adversarial identification model performs tensor contraction on the thermal signal features and the three-dimensional space constraint matrix to generate a symplectic geometry structure feature basis; The discriminator encodes the pre-path data into a dual quaternion motion constraint chain, and realizes the adversarial iteration of the generator and the discriminator through Lie algebra bracket operation, and the iteration termination condition is determined by the exponential moving average of the trace of the laser ranging covariance matrix.
6. The fire source positioning method based on the combination of thermal imaging and laser ranging according to claim 1, characterized in that, The mobile carrier is a drone or a ground mobile robot; when it is a drone, the axial height data of the cylindrical coordinate system is calibrated in combination with the GPS positioning information of the drone; when it is a ground mobile robot, the horizontal direction data of the cylindrical coordinate system is corrected in combination with the wheel odometer data of the robot.
7. The fire source positioning method based on the combination of thermal imaging and laser ranging according to claim 1, characterized in that, In S3, the angular interval constraint is that the central angle of adjacent points in the circumferential direction of the cylindrical coordinate system is ≤ 30°, and the distance constraint is that the straight line distance between the point and the outer surface of the fire source candidate region is within the range of 1-3 meters; the spatial range of the stereoscopic detection box is ± 0.5 meters in the distance direction, ± 5° in the circumferential direction, and ± 0.3 meters in the axial direction in the cylindrical coordinate system, and when the volume overlap rate is ≥ 15%, it is determined that the supplementary point is invalid.
8. The fire source positioning method based on the combination of thermal imaging and laser ranging according to claim 1, characterized in that, In S3, during the Lie algebra fusion process of the octonion geometry space, the angular velocity and acceleration of the mobile carrier inertial data are converted into Lie group elements, and the affine transformation of the connection operator on the octonion geometry space and the thermal signal features and laser ranging features establishes a differential homeomorphism relationship, realizing the cross-domain feature coupling of multi-modal data.
9. A fire source positioning system based on the combination of thermal imaging and laser ranging, characterized in that, It comprises: a fire source thermal signal detection and space modeling module: used to obtain target area thermal imaging, original image and point cloud data through the thermal imager, high-definition camera and laser range finder carried by the mobile carrier, identify the fire source candidate region, construct a cylindrical coordinate system with the center axis of the fire source candidate region as the origin, and generate a three-dimensional model containing the space attitude and obstacle information of the fire source; a shooting angle precise adjustment module: used to adjust the shooting angle of the thermal imager in combination with the three-dimensional model, and plan a surrounding path containing effective detection points; control the mobile carrier to move along the path, and obtain the distance data of the mobile carrier and the fire source candidate region in the cylindrical coordinate system through the laser range finder, and generate a three-dimensional space constraint matrix; a detection path planning and laser ranging module: used to fuse the thermal imaging signal features and the three-dimensional space constraint matrix in the Lie algebra of the octonion geometry space, dynamically adjust the laser ranging covariance weight through the adversarial recognition model, and map it into the spatial coordinates of the fire source in the cylindrical coordinate system; a fire source space positioning module: used to calculate the Riemannian manifold distance between the positioning coordinates and the pre-path data, and when the threshold is exceeded, resample the data and update the model weight to reposition.
10. The fire source locating system based on the combination of thermal imaging and laser ranging according to claim 9, characterized in that, It also comprises a positioning verification and correction module: used to calculate the Riemannian manifold distance between the fire source positioning coordinates and the mobile carrier pre-path data in the cylindrical coordinate system, which is jointly calculated by the Frobenius norm of the three-dimensional space constraint matrix and the determinant of the inertial feature covariance matrix; if the distance exceeds the preset threshold, trigger the fire source thermal signal detection and space modeling step to resample the data, update the adversarial recognition model weight coefficient, and reposition.