A forest fire point positioning method based on a gimbal camera and GIS elevation fusion

Abstract

The application relates to the technical field of forest fire prevention positioning, and discloses a forest fire point positioning method based on a cloud platform camera and GIS elevation fusion, which comprises the following steps: S1, camera calibration and basic parameter input processing of a forest fire point monitoring system are carried out, camera installation parameter sets and video stream superposition configuration data are generated; S2, fire monitoring and fire point image recognition processing are carried out based on the camera installation parameter sets and the video stream superposition configuration data, and fire point pixel coordinate data are generated; real-time horizontal rotation angle, vertical pitch angle and optical zoom parameter are superimposed and displayed in a video picture through configuration of the cloud platform camera, and the parameters are automatically extracted by using an optical character recognition technology, so that the automatic and high-synchronism acquisition of real-time posture parameters of the camera is realized, manual reading errors and time delay are avoided, and the real-time performance and reliability of data relied on subsequent space calculation are ensured.

Description

Technical Field

[0001] This invention relates to the field of forest fire prevention and positioning technology, specifically a method for locating forest fire points based on the fusion of a PTZ camera and GIS elevation data. Background Technology

[0002] Forests are an important component of terrestrial ecosystems, playing a vital role in maintaining ecological balance, improving the ecological environment, and preserving biodiversity. However, in recent years, due to factors such as global warming, intercropping between forests and farms, and frequent forest tourism activities, forest fires have occurred frequently. Because forest fires often occur suddenly, are unpredictable, highly dangerous, and difficult to control, if they are not detected and controlled in time, they can easily spread rapidly, causing incalculable losses.

[0003] Currently, forest fire monitoring relies on visible light or thermal imaging PTZ cameras for fire identification. After determining the fire point through video footage, it is usually necessary to manually read the real-time attitude parameters of the camera, such as horizontal rotation angle and vertical pitch angle, from the video overlay information or from the camera communication protocol. This process is prone to human error or communication delay, resulting in insufficient real-time performance and accuracy of the key parameters obtained for spatial positioning.

[0004] Therefore, a forest fire location method based on the fusion of PTZ camera and GIS elevation is proposed to solve the above problems. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a forest fire location method based on the fusion of a PTZ camera and GIS elevation data, which solves the problem mentioned in the background technology that the real-time performance and accuracy of the key parameters acquired for spatial positioning are insufficient due to human error in reading or communication delay.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for locating forest fire points based on the fusion of a PTZ camera and GIS elevation data, the method comprising the following steps: S1. Perform camera calibration and basic parameter input processing for the forest fire monitoring system, and generate camera installation parameter set and video stream overlay configuration data. S2. Based on the camera installation parameter set and video stream overlay configuration data, perform fire monitoring and fire point image recognition processing to generate fire point pixel coordinate data; S3. Perform optical character recognition processing on the fire point video image with attitude parameters superimposed, and extract and generate real-time attitude parameter data of the camera. S4. Based on the camera installation parameter set, the camera real-time attitude parameter data and the fire point pixel coordinate data, a virtual spherical model is constructed and a spatial line-of-sight vector is calculated to generate fire point spatial line-of-sight vector data. S5. Based on the camera installation parameter set, the fire point spatial line-of-sight vector data, and the preset geographic information system elevation data, perform spatial ray and terrain surface intersection calculation processing to generate three-dimensional geographic coordinate data of the forest fire point.

[0007] Preferably, step S1 involves camera calibration and basic parameter input processing for the forest fire monitoring system, including the following steps: S11. Install a heavy-duty PTZ camera on the fire tower, adjust the camera's installation height and calibrate the camera's initial spatial orientation, and then reset the PTZ camera's horizontal rotation angle, vertical pitch angle, and optical zoom parameters to the preset initial zero position. S12. Configure the pan-tilt camera so that the current horizontal rotation angle parameter value, vertical pitch angle parameter value and optical zoom parameter value are always superimposed on the video output of the camera. S13. Enter the geographic coordinates of the installation point of the pan-tilt camera, the altitude of the installation point, the height of the camera's optical center above the ground, and the spatial orientation azimuth angle after the initial calibration of the camera into the background forest fire monitoring system. Also enter the inherent horizontal and vertical field of view parameters of the camera and integrate them to generate the camera installation parameter set and video stream overlay configuration data.

[0008] Preferably, the fire monitoring and fire point image recognition processing in S2 includes the following steps: S21. The background forest fire monitoring system receives and processes in real time the video stream data with superimposed and displayed attitude parameter values ​​transmitted back by the pan-tilt camera; S22. When the system detects a preset fire alarm signal and triggers a fire warning through real-time video analysis, it calls a preset machine vision target detection algorithm to perform image analysis on the current video frame, identify and select the suspected fire point area in the video screen. S23. The image pixel coordinates corresponding to the midpoint of the bottom edge of the fire point region recognition box output by the machine vision target detection algorithm are determined as fire point pixel coordinate data representing the position of the fire point on the two-dimensional image plane of the video.

[0009] Preferably, the optical character recognition processing in S3 includes the following steps: S31. Obtain the current video frame image containing the fire point area and the overlaid attitude parameter characters; S32. Call the pre-trained optical character recognition model to perform text detection and recognition on the area superimposed in the current video frame image, and extract the horizontal rotation angle parameter value, vertical pitch angle parameter value and optical zoom parameter value displayed in real time in the video screen. S33. Perform data format standardization verification on the extracted values ​​to generate real-time camera attitude parameter data that is strictly synchronized with the fire point pixel coordinate data in time. The real-time camera attitude parameter data includes horizontal rotation angle P value, vertical pitch angle T value and optical zoom parameter value.

[0010] Preferably, the virtual spherical model construction and spatial line-of-sight vector calculation in S4 includes the following steps: S41. Calculate the effective horizontal field of view and effective vertical field of view under the current zoom level based on the camera's inherent horizontal field of view, vertical field of view, and optical zoom parameter values. S42. Using the optical center of the PTZ camera as the origin, construct a virtual unit sphere model, where the center pixel of the video image is mapped to the point on the virtual unit sphere corresponding to the direction of the camera's optical axis. S43. Based on the positional offset of the fire point pixel coordinates relative to the center of the video frame, the total pixel size of the video frame, the effective horizontal field of view, and the effective vertical field of view, calculate the longitude and latitude offset angles of the fire point pixel coordinates on the virtual unit sphere. S44. Combining the horizontal rotation angle P and vertical pitch angle T values ​​in the real-time attitude parameter data of the camera, and the initial orientation azimuth angle in the camera installation parameter set, the points on the virtual unit sphere determined by the longitude offset angle and the latitude offset angle are converted into fire point spatial line-of-sight vector data based on the geographic coordinate system of the camera installation point. The fire point spatial line-of-sight vector data is a three-dimensional direction vector.

[0011] Preferably, the specific steps for calculating the longitude offset angle and the latitude offset angle in step S43 are as follows: S431. Obtain the total number of pixels in the horizontal direction and the total number of pixels in the vertical direction of the video frame, and determine the pixel coordinates of the center point of the video frame; S432. Calculate the pixel offset of the fire point pixel coordinates relative to the center point of the image in the horizontal and vertical directions. S433. The longitude offset angle is calculated by using a linear proportional relationship based on the horizontal pixel offset, the total number of horizontal pixels, and the effective horizontal field of view. S434. The latitude offset angle is calculated by using a linear proportional relationship based on the vertical pixel offset, the total number of vertical pixels, and the effective vertical field of view.

[0012] Preferably, the calculation and processing of the intersection of the spatial ray and the terrain surface in S5 includes the following steps: S51. Based on the geographical coordinates of the installation point, the altitude parameters, and the ground clearance of the camera's optical center in the camera installation parameter set, calculate and determine the precise physical starting point coordinates of the PTZ camera's optical center in three-dimensional space. S52. Using the precise physical starting point coordinates of the camera's optical center as the ray starting point and the direction of the fire point's spatial line-of-sight vector data as the ray direction, establish a spatial ray equation pointing towards the fire point target. S53. Call the preset geographic information system elevation data model, wherein the geographic information system elevation data model is a three-dimensional digital elevation model containing topographic relief information of the monitoring area; S54. Using the spatial geometric ray tracing algorithm, calculate the intersection point between the spatial ray equation and the terrain surface represented by the geographic information system elevation data model; S55. Extract the three-dimensional coordinates of the intersection point and output them as the final three-dimensional geographic coordinate data of the forest fire point. The three-dimensional geographic coordinate data of the forest fire point includes longitude coordinates, latitude coordinates and altitude coordinates.

[0013] Preferably, the specific operation steps of the spatial geometry ray tracing algorithm in S54 are as follows: S541. Starting from the origin of the space ray, perform progressive sampling along the ray direction with a preset step size; S542. At each sampling point, query the elevation data model of the geographic information system to obtain the elevation value corresponding to the latitude and longitude coordinates of that point. S543. Compare the spatial height of the sampling point with the elevation value obtained by querying its latitude and longitude coordinates in the geographic information system elevation data model; S544. When the spatial height of a sampling point is less than or equal to its corresponding elevation value for the first time, it is determined that the spatial ray and the terrain surface intersect in the interval near this sampling point. S545. Using linear interpolation, perform precise calculations between the last sampling point whose spatial height is higher than the terrain elevation and the first sampling point whose spatial height is lower than or equal to the terrain elevation to determine the final intersection point coordinates, i.e., the three-dimensional geographic coordinate data of the forest fire point.

[0014] Preferably, the operation of performing data format standardization verification on the extracted values ​​in S33 includes: The text string extracted by optical character recognition is converted into a floating-point number format, and its value is verified to be within the reasonable range allowed by the physical parameters of the pan-tilt camera. Abnormal values ​​and data with recognition errors are discarded, and re-recognition and alarms are triggered.

[0015] Preferably, after outputting the three-dimensional geographic coordinate data of forest fire points in S55, the method further includes: The three-dimensional geographic coordinates of forest fire points are encapsulated with the fire identification timestamp and source camera identification information to form a standard format fire point location information message, which is then transmitted to the forest fire prevention command and dispatch system to trigger subsequent alarm issuance, path planning and rescue resource dispatch operations.

[0016] Compared with existing technologies, this invention provides a method for locating forest fire points based on the fusion of PTZ camera and GIS elevation data, which has the following advantages: 1. In this invention, by configuring a PTZ camera to overlay and display real-time horizontal rotation angle, vertical pitch angle, and optical zoom parameters on the video screen, and by using optical character recognition technology to automatically extract these parameters, the real-time attitude parameters of the camera are obtained automatically and with high synchronization, avoiding errors and delays from manual reading, and ensuring the real-time performance and reliability of the data on which subsequent spatial calculations depend.

[0017] 2. In this invention, the pixel coordinates of the fire point are converted into a spatial line-of-sight vector centered on the camera by constructing a virtual spherical model. The spatial geometric ray tracing algorithm is then used to accurately calculate the intersection of this vector with the elevation data model of the geographic information system. This fully considers the undulations of the actual terrain, thereby directly solving the two-dimensional image coordinates into three-dimensional geographic coordinates of the forest fire point, including longitude, latitude, and altitude, thus improving the positioning accuracy in complex mountainous environments.

[0018] 3. In this invention, the generated three-dimensional geographic coordinate data of forest fire points, along with the fire identification timestamp and source camera identification information, are encapsulated into a standard format fire point location information message and transmitted to the forest fire prevention command and dispatch system. This enables the location results to be directly and quickly applied to alarm issuance, path planning, and rescue resource dispatch by the backend system, achieving full-process automation from monitoring and identification to location and then to command and disposal, thereby improving the overall response speed and decision-making efficiency of the forest fire prevention system. Attached Figure Description

[0019] Figure 1 This is a flowchart of a forest fire location method based on the fusion of a PTZ camera and GIS elevation data according to the present invention. Figure 2 This is a schematic diagram of the virtual spherical mapping principle of a forest fire location method based on the fusion of a PTZ camera and GIS elevation data according to the present invention. Detailed Implementation

[0020] 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.

[0021] For specific implementation examples, please refer to: Figures 1-2 A method for locating forest fire points based on the fusion of PTZ camera and GIS elevation data, the method includes the following steps: S1. Perform camera calibration and basic parameter input processing for the forest fire monitoring system, and generate camera installation parameter set and video stream overlay configuration data. S2. Based on the camera installation parameter set and video stream overlay configuration data, fire monitoring and fire point image recognition processing are performed to generate fire point pixel coordinate data. S3. Perform optical character recognition processing on the fire point video image with attitude parameters superimposed, and extract and generate real-time attitude parameter data of the camera. S4. Based on the camera installation parameter set, camera real-time attitude parameter data and fire point pixel coordinate data, construct a virtual spherical model and perform spatial line-of-sight vector calculation to generate fire point spatial line-of-sight vector data. S5. Based on the camera installation parameter set, the spatial line-of-sight vector data of the fire point, and the preset geographic information system elevation data, perform spatial ray and terrain surface intersection calculation processing to generate three-dimensional geographic coordinate data of the forest fire point.

[0022] In S1, the camera calibration and basic parameter input processing for the forest fire monitoring system are performed, including the following steps: S11. Install a heavy-duty PTZ camera on the fire tower, adjust the camera's installation height and calibrate the camera's initial spatial orientation, and then reset the PTZ camera's horizontal rotation angle, vertical pitch angle, and optical zoom parameters to the preset initial zero position. S12. Configure a pan-tilt camera so that the current horizontal rotation angle parameter value, vertical tilt angle parameter value, and optical zoom parameter value are always superimposed on the video output of the camera. S13. Enter the geographic coordinates of the installation point of the pan-tilt camera, the altitude of the installation point, the height of the camera's optical center above the ground, and the spatial orientation azimuth angle after the initial calibration of the camera into the background forest fire monitoring system. Also, enter the inherent horizontal and vertical field of view parameters of the camera and integrate them to generate the camera installation parameter set and video stream overlay configuration data.

[0023] S2 fire monitoring and fire point image recognition processing includes the following steps: S21. The background forest fire monitoring system receives and processes in real time the video stream data with superimposed and displayed attitude parameter values ​​transmitted back by the pan-tilt camera. S22. When the system detects a preset fire alarm signal and triggers a fire warning through real-time video analysis, it calls a preset machine vision target detection algorithm to perform image analysis on the current video frame, identify and select the suspected fire point area in the video screen. In the specific implementation of the machine vision target detection algorithm, a pre-built flame and smoke detection model based on a deep convolutional neural network is invoked to analyze the video frames transmitted back by the PTZ camera in real time. This model takes RGB images as input, extracts multi-scale features through multi-layer convolution and pooling operations, and finally outputs the bounding box coordinates and corresponding confidence scores of suspected fire points through regression and classification heads. To balance false positives and false negatives, an adjustable confidence threshold is set, and only detection boxes with confidence scores higher than this threshold are retained. The formula is expressed as: ; in This represents the set of valid fire point bounding boxes after confidence filtering. Indicates the first A preliminary bounding box of suspected fire points identified by the object detection model. Represents bounding box The corresponding confidence score, This represents the preset confidence threshold; by continuously collecting labeled data, the model is fine-tuned offline and incrementally learned online to continuously improve the detection accuracy and robustness of fires in different forest environments and under different lighting conditions; S23. Determine the image pixel coordinates corresponding to the midpoint of the bottom edge of the fire point region recognition box output by the machine vision target detection algorithm as the fire point pixel coordinate data representing the position of the fire point on the two-dimensional image plane of the video. Among them, determining the fire point pixel coordinates The data process is defined by the following formula: ; ; in The x-coordinate of the fire point pixel. The vertical coordinate of the fire point pixel. and These represent the pixel coordinates of the left and right boundaries of the bounding box in the horizontal direction of the image, respectively. This represents the pixel coordinates of the bottom boundary of the recognition box in the vertical direction of the image.

[0024] The optical character recognition processing in S3 includes the following steps: S31. Obtain the current video frame image containing the fire point area and the overlaid attitude parameter characters; S32. Call the pre-trained optical character recognition model to perform text detection and recognition on the superimposed area in the current video frame image, and extract the horizontal rotation angle parameter value, vertical pitch angle parameter value and optical zoom parameter value displayed in real time in the video picture. In the specific implementation of the optical character recognition model, an image region with overlaid pose parameters is extracted from a video frame containing fire points. First, this image region is preprocessed, including grayscale conversion and adaptive threshold binarization, to enhance the contrast between characters and background and reduce the impact of lighting changes. Then, connection-based component analysis and deep learning text detection algorithms are used to locate and segment independent regions for characters with horizontal rotation angles, vertical pitch angles, and zoom parameters. For the segmented character images, a pre-trained lightweight convolutional neural network (CRNN) is used for end-to-end sequence recognition. This network can simultaneously perform feature extraction and sequence modeling, directly outputting the corresponding parameter strings. The recognition results are verified by similarity comparison with a pre-set character template library, and finally parsed and output as structured numerical data for subsequent steps. S33. Perform data format standardization verification on the extracted values ​​to generate real-time camera attitude parameter data that is strictly synchronized with the fire point pixel coordinate data in time. The real-time camera attitude parameter data includes the horizontal rotation angle P value, the vertical pitch angle T value, and the optical zoom parameter value.

[0025] The virtual sphere model construction and spatial view vector calculation in S4 include the following steps: S41. Calculate the effective horizontal field of view and effective vertical field of view under the current zoom level based on the camera's inherent horizontal field of view, vertical field of view, and optical zoom parameter values. The process of calculating the effective field of view is defined by the following formula: ; ; in and These represent the calculated effective horizontal field of view and effective vertical field of view, respectively. and These represent the camera's inherent maximum horizontal field of view and maximum vertical field of view at the minimum zoom level, respectively. Indicates the optical zoom parameter value and ; S42. Using the optical center of the PTZ camera as the origin, construct a virtual unit sphere model, where the center pixel of the video image is mapped to the point on the virtual unit sphere corresponding to the direction of the camera's optical axis. S43. Based on the positional offset of the fire point pixel coordinates relative to the center of the video frame, the total pixel size of the video frame, the effective horizontal field of view, and the effective vertical field of view, calculate the longitude and latitude offset angles of the fire point pixel coordinates on the virtual unit sphere. S44. Combining the horizontal rotation angle P and vertical pitch angle T values ​​in the real-time attitude parameter data of the camera, and the initial orientation azimuth angle in the camera installation parameter set, the points on the virtual unit sphere determined by the longitude offset angle and the latitude offset angle are converted into fire point spatial line-of-sight vector data based on the geographic coordinate system of the camera installation point. The fire point spatial line-of-sight vector data is a three-dimensional direction vector. The process of converting spherical coordinates into spatial line-of-sight vector data is defined by the following formula: ; ; ; ; in This represents the initial gaze vector in the camera's local coordinate system. and These respectively indicate that the gimbal angle has been taken into account. and pixel offset angle The total horizontal azimuth and total pitch angles are then calculated. and These are the horizontal rotation angle P value and the vertical pitch angle T value extracted from S33, respectively. This represents the fire point space line-of-sight vector data transformed to the world coordinate system. Indicates the initial orientation azimuth angle of rotation about the vertical axis. The rotation matrix.

[0026] The specific steps for calculating the longitude and latitude offset angles in S43 are as follows: S431. Obtain the total number of pixels in the horizontal direction and the total number of pixels in the vertical direction of the video frame, and determine the pixel coordinates of the center point of the video frame; ; ; in This represents the pixel coordinates of the center point of the video frame in the horizontal direction of the image. This represents the pixel coordinates of the center point of the video frame in the vertical direction of the image. This indicates the total number of pixels in the horizontal direction of the video frame. This represents the total number of pixels in the vertical direction of the video frame. S432. Calculate the pixel offset of the fire point pixel coordinates relative to the center point of the image in the horizontal and vertical directions. ; ; in , These represent the pixel offsets of the fire point's pixel coordinates relative to the center point of the image in the horizontal and vertical directions, respectively. S433. The longitude offset angle is calculated by using a linear proportional relationship based on the horizontal pixel offset, the total number of horizontal pixels, and the effective horizontal field of view. ; in This represents the calculated longitude offset angle. This indicates the total number of pixels in the horizontal direction of the video frame. S434. The latitudinal offset angle is calculated using a linear proportional relationship based on the vertical pixel offset, the total number of vertical pixels, and the effective vertical field of view. ; in This represents the calculated latitude offset angle.

[0027] The calculation and processing of the intersection of spatial rays and terrain surfaces in S5 includes the following steps: S51. Based on the geographical coordinates of the installation point, the altitude parameters, and the ground clearance of the camera's optical center in the camera installation parameter set, calculate and determine the precise physical starting point coordinates of the PTZ camera's optical center in three-dimensional space. ; in This represents the precise physical starting point coordinates of the optical center of the PTZ camera in three-dimensional space, obtained through final calculation. This represents the Cartesian coordinates of the camera mounting point obtained after Gaussian projection transformation from its geodetic coordinates. Indicates the altitude of the camera installation point. This indicates the height of the camera's optical center relative to the mounting point above the ground. The plane coordinates representing the optical center of the camera, This indicates the elevation coordinates of the camera's optical center. ; S52. Using the precise physical starting coordinates of the camera's optical center as the ray origin and the direction of the fire point's spatial line-of-sight vector data as the ray direction, establish a spatial ray equation pointing towards the fire point target: ; ; in Representing the equation of a space ray, Represents the coordinates of a point on the ray. This is the distance parameter along the ray direction; S53. Call the preset geographic information system elevation data model. The geographic information system elevation data model is a three-dimensional digital elevation model that includes the topographic relief information of the monitoring area. In practical implementation, the geographic information system (GIS) elevation data model adopts a digital elevation model (DEM). First, raw elevation data sources for the monitoring area are collected, which can originate from airborne LiDAR point clouds, densely matched point clouds generated by aerial photogrammetry, and contour vector data. The raw data is then cleaned to remove gross errors, and interpolation algorithms are used to generate continuous elevation surfaces. This method constructs a digital elevation model with a regular grid structure, which can be mathematically represented as a two-dimensional matrix. Each grid point stores an elevation value and establishes a linear mapping relationship with geographic coordinates through its row and column numbers; to improve the efficiency and accuracy of ray tracing calculations, multi-resolution pyramid construction and smoothing processing can be performed on the DEM data; the final generated DEM data is stored together with geographic projection information to provide a terrain elevation query interface for spatial intersection calculations in S54; S54. Using the spatial geometric ray tracing algorithm, calculate the intersection point between the spatial ray equation and the terrain surface represented by the geographic information system elevation data model; S55. Extract the three-dimensional coordinates of the intersection point and output them as the final three-dimensional geographic coordinate data of the forest fire point. The three-dimensional geographic coordinate data of the forest fire point includes longitude coordinates, latitude coordinates and altitude coordinates.

[0028] The specific steps for using the spatial geometry ray tracing algorithm in S54 are as follows: S541. Starting from the origin of the space ray, perform progressive sampling along the ray direction with a preset step size; ; ; in Indicates the first The coordinates of each sampling point This indicates the preset sampling step size. Indicates the sampling point number; S542. At each sampling point, query the geographic information system elevation data model to obtain the elevation value corresponding to the latitude and longitude coordinates of that point. S543. Compare the spatial height of the sampling point with the elevation value obtained by querying its latitude and longitude coordinates in the geographic information system elevation data model; S544. When the spatial height of a sampling point is less than or equal to its corresponding elevation value for the first time, it is determined that the spatial ray and the terrain surface intersect in the interval near this sampling point. S545. Using linear interpolation, perform precise calculations between the last sampling point whose spatial height is higher than the terrain elevation and the first sampling point whose spatial height is lower than or equal to the terrain elevation to determine the final intersection point coordinates, i.e., the three-dimensional geographic coordinate data of the forest fire point. The linear interpolation calculation is defined by the following formula: ; ; in This represents the final calculated coordinates of the intersection point. and These represent the last sampling point whose spatial height is higher than the terrain elevation and the first sampling point whose spatial height is lower than or equal to the terrain elevation, respectively. This is the interpolation scaling factor. and These represent the sampling points. and Spatial height, and These represent the sampling points respectively. and The elevation value obtained from the elevation data model of the geographic information system at the latitude and longitude coordinates.

[0029] The data format standardization and validation operations performed on the extracted values ​​in S33 include: The text string extracted by optical character recognition is converted into a floating-point number format, and its value is verified to be within the reasonable range allowed by the physical parameters of the pan-tilt camera. Abnormal values ​​and data with recognition errors are discarded, and re-recognition and alarms are triggered.

[0030] After completing the output of the three-dimensional geographic coordinate data of forest fire points in S55, the method also includes: The three-dimensional geographic coordinates of forest fire points are encapsulated with the fire identification timestamp and source camera identification information to form a standard format fire point location information message, which is then transmitted to the forest fire prevention command and dispatch system to trigger subsequent alarm issuance, path planning and rescue resource dispatch operations.

[0031] The operational steps of a forest fire location method based on the fusion of PTZ camera and GIS elevation data are as follows: Step 1: Perform camera calibration and basic parameter input. A heavy-duty PTZ camera was installed and calibrated on the fire tower, with its horizontal rotation angle, vertical tilt angle, and optical zoom parameters returned to their initial zero positions. The camera was configured to continuously overlay these attitude parameter values ​​onto its video feed. Subsequently, the camera's installation point's geographical coordinates, altitude, optical center height above ground, initial spatial orientation azimuth, and its inherent horizontal and vertical field of view were entered into the backend system, integrating these data to generate a set of camera installation parameters and video stream overlay configuration data.

[0032] Step 2: Fire monitoring and fire point image recognition based on the above parameters: The backend system receives and processes video streams from cameras in real time, overlaid with attitude parameter values. When the system detects a fire alarm signal, it calls a machine vision target detection algorithm to analyze the current video frame, identify, and select suspected fire areas in the image. Finally, the image pixel coordinates corresponding to the midpoint of the bottom edge of the identified box are determined as the fire point pixel coordinates.

[0033] Step 3: Perform optical character recognition on the fire point video image containing attitude parameters: The system acquires the current video frame image containing the fire point and superimposed parameter characters. A pre-trained optical character recognition model is then invoked to perform text detection and recognition on the superimposed area in the image, thereby extracting the real-time displayed horizontal rotation angle, vertical pitch angle, and optical zoom parameter values. After standardizing and validating these values, real-time camera attitude parameter data, synchronized with the fire point pixel coordinate data, is generated, including the horizontal rotation angle P value, vertical pitch angle T value, and optical zoom parameter values.

[0034] Step 4: Construct a virtual spherical model and calculate the spatial line-of-sight vector based on multi-source data: Based on the camera's inherent field of view and optical zoom parameters, the effective horizontal and vertical field of view at the current zoom level are calculated. Next, a virtual unit spherical model is constructed with the camera's optical center as the origin, and the center of the video image is mapped to a point on this sphere representing the optical axis direction. Then, based on the offset of the fire point pixel coordinates relative to the image center, the total pixel size of the image, and the effective field of view, the longitude and latitude offset angles corresponding to that pixel on the virtual unit sphere are calculated. Finally, combining the real-time attitude parameters and the initial orientation azimuth angle from the installation parameters, the points on the sphere are converted into fire point spatial line-of-sight vector data based on the geographic coordinate system of the camera installation point.

[0035] Step 5: Calculate the intersection of spatial rays and terrain surfaces based on spatial line-of-sight vectors and GIS elevation data. Based on the geographic coordinates, altitude, and optical center height of the installation point, the precise physical starting point coordinates of the camera's optical center in three-dimensional space are calculated. Using this starting point as the ray origin and the direction of the fire point's spatial line-of-sight vector as the ray direction, a spatial ray equation is established. Subsequently, a pre-set geographic information system elevation data model is invoked; using a spatial geometric ray tracing algorithm, the intersection point between this spatial ray and the terrain surface represented by the elevation data model is calculated. Finally, the three-dimensional coordinates of this intersection point are extracted and output as the three-dimensional geographic coordinate data of the forest fire point.

[0036] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0037] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for locating forest fire points based on the fusion of PTZ camera and GIS elevation data, characterized in that: The method includes the following steps: S1. Perform camera calibration and basic parameter input processing for the forest fire monitoring system, and generate camera installation parameter set and video stream overlay configuration data. S2. Based on the camera installation parameter set and video stream overlay configuration data, perform fire monitoring and fire point image recognition processing to generate fire point pixel coordinate data; S3. Perform optical character recognition processing on the fire point video image with attitude parameters superimposed, and extract and generate real-time attitude parameter data of the camera. S4. Based on the camera installation parameter set, the camera real-time attitude parameter data and the fire point pixel coordinate data, a virtual spherical model is constructed and a spatial line-of-sight vector is calculated to generate fire point spatial line-of-sight vector data. S5. Based on the camera installation parameter set, the fire point spatial line-of-sight vector data, and the preset geographic information system elevation data, perform spatial ray and terrain surface intersection calculation processing to generate three-dimensional geographic coordinate data of the forest fire point.

2. The forest fire location method based on the fusion of PTZ camera and GIS elevation data according to claim 1, characterized in that: The S1 step involves camera calibration and basic parameter input processing for the forest fire monitoring system, including the following steps: S11. Install a heavy-duty PTZ camera on the fire tower, adjust the camera's installation height and calibrate the camera's initial spatial orientation, and then reset the PTZ camera's horizontal rotation angle, vertical pitch angle, and optical zoom parameters to the preset initial zero position. S12. Configure a pan-tilt camera so that the current horizontal rotation angle parameter value, vertical tilt angle parameter value, and optical zoom parameter value are always superimposed on the video output of the camera. S13. Enter the geographic coordinates of the installation point of the pan-tilt camera, the altitude of the installation point, the height of the camera's optical center above the ground, and the spatial orientation azimuth angle after the initial calibration of the camera into the background forest fire monitoring system. Also, enter the inherent horizontal and vertical field of view parameters of the camera and integrate them to generate the camera installation parameter set and video stream overlay configuration data.

3. The forest fire location method based on the fusion of PTZ camera and GIS elevation data according to claim 2, characterized in that: The fire monitoring and fire point image recognition processing in S2 includes the following steps: S21. The background forest fire monitoring system receives and processes in real time the video stream data with superimposed and displayed attitude parameter values ​​transmitted back by the pan-tilt camera. S22. When the system detects a preset fire alarm signal and triggers a fire warning through real-time video analysis, it calls a preset machine vision target detection algorithm to perform image analysis on the current video frame, identify and select the suspected fire point area in the video screen. S23. The image pixel coordinates corresponding to the midpoint of the bottom edge of the fire point region recognition box output by the machine vision target detection algorithm are determined as fire point pixel coordinate data representing the position of the fire point on the two-dimensional image plane of the video.

4. The forest fire location method based on the fusion of PTZ camera and GIS elevation data according to claim 3, characterized in that: The optical character recognition processing in S3 includes the following steps: S31. Obtain the current video frame image containing the fire point area and the overlaid attitude parameter characters; S32. Call the pre-trained optical character recognition model to perform text detection and recognition on the area superimposed in the current video frame image, and extract the horizontal rotation angle parameter value, vertical pitch angle parameter value and optical zoom parameter value displayed in real time in the video screen. S33. Perform data format standardization verification on the extracted values ​​to generate real-time camera attitude parameter data that is strictly synchronized with the fire point pixel coordinate data in time. The real-time camera attitude parameter data includes horizontal rotation angle P value, vertical pitch angle T value and optical zoom parameter value.

5. A forest fire location method based on the fusion of a PTZ camera and GIS elevation data according to claim 4, characterized in that: The virtual spherical model construction and spatial line-of-sight vector calculation in S4 include the following steps: S41. Calculate the effective horizontal field of view and effective vertical field of view under the current zoom level based on the camera's inherent horizontal field of view, vertical field of view, and optical zoom parameter values. S42. Using the optical center of the PTZ camera as the origin, construct a virtual unit sphere model, where the center pixel of the video image is mapped to the point on the virtual unit sphere corresponding to the direction of the camera's optical axis. S43. Based on the positional offset of the fire point pixel coordinates relative to the center of the video frame, the total pixel size of the video frame, the effective horizontal field of view, and the effective vertical field of view, calculate the longitude and latitude offset angles of the fire point pixel coordinates on the virtual unit sphere. S44. Combining the horizontal rotation angle P and vertical pitch angle T values ​​in the real-time attitude parameter data of the camera, and the initial orientation azimuth angle in the camera installation parameter set, the points on the virtual unit sphere determined by the longitude offset angle and the latitude offset angle are converted into fire point spatial line-of-sight vector data based on the geographic coordinate system of the camera installation point. The fire point spatial line-of-sight vector data is a three-dimensional direction vector.

6. The forest fire location method based on the fusion of PTZ camera and GIS elevation data according to claim 5, characterized in that: The specific steps for calculating the longitude and latitude offset angles in S43 are as follows: S431. Obtain the total number of pixels in the horizontal direction and the total number of pixels in the vertical direction of the video frame, and determine the pixel coordinates of the center point of the video frame; S432. Calculate the pixel offset of the fire point pixel coordinates relative to the center point of the image in the horizontal and vertical directions. S433. The longitude offset angle is calculated by using a linear proportional relationship based on the horizontal pixel offset, the total number of horizontal pixels, and the effective horizontal field of view. S434. The latitude offset angle is calculated by using a linear proportional relationship based on the vertical pixel offset, the total number of vertical pixels, and the effective vertical field of view.

7. A forest fire location method based on the fusion of a PTZ camera and GIS elevation data as described in claim 6, characterized in that: The calculation and processing of the intersection of spatial rays and terrain surfaces in S5 includes the following steps: S51. Based on the geographical coordinates of the installation point, the altitude parameters, and the ground clearance of the camera's optical center in the camera installation parameter set, calculate and determine the precise physical starting point coordinates of the PTZ camera's optical center in three-dimensional space. S52. Using the precise physical starting point coordinates of the camera's optical center as the ray starting point and the direction of the fire point's spatial line-of-sight vector data as the ray direction, establish a spatial ray equation pointing towards the fire point target. S53. Call the preset geographic information system elevation data model, wherein the geographic information system elevation data model is a three-dimensional digital elevation model containing topographic relief information of the monitoring area; S54. Using the spatial geometric ray tracing algorithm, calculate the intersection point between the spatial ray equation and the terrain surface represented by the geographic information system elevation data model; S55. Extract the three-dimensional coordinates of the intersection point and output them as the final three-dimensional geographic coordinate data of the forest fire point. The three-dimensional geographic coordinate data of the forest fire point includes longitude coordinates, latitude coordinates and altitude coordinates.

8. A method for locating forest fire points based on the fusion of a PTZ camera and GIS elevation data according to claim 7, characterized in that: The specific steps of the spatial geometry ray tracing algorithm in S54 are as follows: S541. Starting from the origin of the space ray, perform progressive sampling along the ray direction with a preset step size; S542. At each sampling point, query the geographic information system elevation data model to obtain the elevation value corresponding to the latitude and longitude coordinates of that point. S543. Compare the spatial height of the sampling point with the elevation value obtained by querying its latitude and longitude coordinates in the geographic information system elevation data model; S544. When the spatial height of a sampling point is less than or equal to its corresponding elevation value for the first time, it is determined that the spatial ray and the terrain surface intersect in the interval near this sampling point. S545. Using linear interpolation, perform precise calculations between the last sampling point whose spatial height is higher than the terrain elevation and the first sampling point whose spatial height is lower than or equal to the terrain elevation to determine the final intersection point coordinates, i.e., the three-dimensional geographic coordinate data of the forest fire point.

9. A method for locating forest fire points based on the fusion of a PTZ camera and GIS elevation data according to claim 4, characterized in that: The data format standardization verification operation of the extracted values ​​in S33 includes: The text string extracted by optical character recognition is converted into a floating-point number format, and its value is verified to be within the reasonable range allowed by the physical parameters of the pan-tilt camera. Abnormal values ​​and data with recognition errors are discarded, and re-recognition and alarms are triggered.

10. A method for locating forest fire points based on the fusion of a PTZ camera and GIS elevation data according to claim 7, characterized in that: After completing the output of the three-dimensional geographic coordinate data of forest fire points in S55, the method also includes: The three-dimensional geographic coordinates of forest fire points are encapsulated with the fire identification timestamp and source camera identification information to form a standard format fire point location information message, which is then transmitted to the forest fire prevention command and dispatch system to trigger subsequent alarm issuance, path planning and rescue resource dispatch operations.

Images