UAV thermal infrared and oblique photography equipment
By integrating a UAV with a thermal infrared camera, a tilting camera, and a lidar, and combining it with satellite-based differential signals and a POS system, the problems of sensor imaging quality and data fusion accuracy in extremely high-altitude environments have been solved, enabling high-precision geological disaster monitoring.
Patent Information
- Application Number
- CN202611245668.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-17
- Publication Date
- 2026-11-17
AI Technical Summary
In extremely high-altitude and complex terrain environments, existing UAV oblique photography technology is difficult to achieve high-precision geological disaster monitoring, sensor imaging quality is reduced, multi-sensor data fusion accuracy is low, and satellite remote sensing and ground surveying are costly and time-consuming.
The system employs UAV thermal infrared and oblique photography equipment, integrating a thermal infrared camera, an oblique camera group, a lidar, and a POS system. It receives satellite-based differential signals through the POS system for real-time positioning, dynamically calibrates sensors, generates images and point cloud data with a unified spatiotemporal reference, and achieves high-precision data fusion by combining exterior orientation elements and image reprojection error parameters.
The system enables the simultaneous acquisition and fusion of high-resolution images and three-dimensional morphological information at extremely high altitudes, meeting the multi-dimensional fusion perception requirements of ice and rock avalanches and snow lines, achieving decimeter-level positioning accuracy, and satisfying the high-precision monitoring needs of extremely high mountain areas.
Smart Images

Figure CN123111206A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of aerial remote sensing and surveying technology, specifically to a UAV thermal infrared and oblique photography equipment. Background Technology
[0002] Extremely high-altitude regions possess significant extreme environments and complex geological features, characterized by large elevation differences, widespread steep slopes, and common occurrences of snow cover and dense vegetation. They also exhibit complex geological structures, intense plate activity, and severe high-altitude, remote geological hazards. Limited by extreme environmental conditions such as steep terrain, high altitude and dense fog, and snow / vegetation cover, widely used space-based and ground-based observation technologies and equipment are insufficient to meet the demands for highly reliable geological hazard risk early warning. These technologies face challenges such as insufficient effective observation, extremely high observation difficulty, low accuracy in geological body modeling, and difficulties in weak signal transmission from sensor networks. Satellite remote sensing methods have limited accuracy, while ground-based surveying methods are costly and time-consuming. Unmanned aerial vehicle (UAV) oblique photogrammetry is currently one of the more mature and widely used methods. As a mature and widely applied method for real-world 3D modeling, UAV oblique photogrammetry offers advantages such as flexibility, high coverage efficiency, and relatively reliable modeling accuracy.
[0003] Current oblique photogrammetry technology still faces technical bottlenecks in complex high-altitude applications. Existing aerial oblique photogrammetry equipment is mostly designed for low- to mid-altitude regions. In extremely high-altitude environments, due to steep terrain, complex topography, and limited communication, conventional ground-based differential signals have limited coverage and insufficient stability, making it difficult to rely on existing ground-based differential networks to achieve high-precision photogrammetric operations without image control for UAVs. The direct ground-based positioning accuracy of UAV oblique photogrammetry is also difficult to guarantee. Furthermore, the thin air and large temperature differences lead to decreased sensor imaging quality and insufficient equipment reliability. A single sensor cannot simultaneously meet the requirements of high-precision 3D modeling, target recognition, and thermal feature acquisition, while simultaneous calibration of multi-sensor equipment is difficult, resulting in low data fusion accuracy. Summary of the Invention
[0004] This invention provides a drone-based thermal infrared and oblique photography equipment, which can support high-precision identification of massive ice and rock avalanches and snowline monitoring in high-altitude areas.
[0005] This invention proposes a UAV thermal infrared and oblique photography equipment, comprising a thermal infrared camera, an oblique camera group, a lidar, a POS system, and a UAV uncontrolled photogrammetry system mounted side by side on the UAV. The thermal infrared camera integrates multiple thermal infrared imaging units; The unmanned aerial vehicle (UAV) photogrammetry system is used to dynamically and jointly calibrate thermal infrared cameras, tilt camera groups, and lidar based on a POS system. It determines real-time positioning data through satellite-based differential signals received by the POS system. Based on the real-time positioning data, it controls the thermal infrared cameras, tilt camera groups, and lidar to generate tilted photogrammetric images, thermal infrared orthophotos, and three-dimensional laser point clouds with a unified spatiotemporal reference for the work area. It also combines calibrated exterior orientation elements, antenna arm parameters, and image reprojection error parameters to obtain visible light orthophotos and infrared temperature field data.
[0006] Preferably, the unmanned aerial vehicle (UAV) photogrammetry system dynamically calibrates the thermal infrared camera, tilt camera group, and lidar in conjunction with each of the following steps: Determine the calibration test site; Reference images and ground control points of the calibration test field are obtained using other aerial cameras; absolute orientation is performed using ground control points to obtain geographic reference data and establish a collinearity model; Aerial photography was used to collect images of the calibration test site to be calibrated using thermal infrared and oblique photography equipment. A multimodal automatic matching method is used to obtain multiple corresponding points between the image to be calibrated and the reference image, which serve as control points for the geometric calibration of thermal infrared and oblique photography equipment. Based on the collinearity model and control points, a lever error and attitude deviation model is introduced. With the minimum image point reprojection error as the objective function, the least squares method is used to iteratively calculate the external orientation parameters, lever error parameters, and attitude deviation parameters of each camera device in the output thermal infrared and oblique photography equipment.
[0007] Preferably, in the unmanned aerial vehicle (UAV) photogrammetry system, the geographic reference data is a collinearity model between ground points and image points of high-resolution aerial images, and the collinearity model is a three-dimensional similarity transformation model.
[0008] Preferably, in the unmanned aerial vehicle (UAV) photogrammetry system, the least squares method is used to iteratively calculate the internal and external orientation parameters, antenna boom error parameters, and attitude deviation parameters of each camera device in the thermal infrared and oblique photography equipment, specifically including: Based on the collinearity model, the geometric relationship between image points and ground points is established; Linearize the geometric relationship between the image point and the ground point to obtain the image point reprojection error; Considering the spatial relationship between the sensor and the inertial navigation system, lever error and attitude deviation parameters are introduced; Substitute the lever error and attitude deviation parameters into the collinearity model to establish a complete observation model, and take the minimum image point reprojection error of the geometric calibration control point as the objective function. Taylor expansion of the complete observation model yields linearized equations, and the parameter corrections are solved using the least squares method. Based on the objective function, iterative calculations are performed until convergence, ultimately obtaining the sensor orientation parameters, stick error parameters, and attitude deviation parameters, thus completing the flight dynamic calibration.
[0009] Preferably, in the unmanned aerial vehicle (UAV) photogrammetry system, the acquisition of the exterior orientation elements, antenna boom parameters, and image reprojection error parameters specifically includes: Acquire 3D laser point cloud and oblique photogrammetric image data, determine the set of spatial points and image observation points, and construct image reprojection error based on the projection function of the collinearity relationship model; Considering the fixed spatial offset between the camera and the GNSS antenna, determine the geometric relationship between the boom arm and the image position; transform the geometric relationship into an error form and construct GNSS constraints; By combining image reprojection error and GNSS constraints, a weighted overall optimization objective function is constructed. Taylor expansion of the complete observation model yields linearized equations, and the parameter corrections are solved using the least squares method. Based on the overall optimization objective function, iterative calculations are performed until convergence to obtain the optimal estimated exterior orientation elements, arm parameters, and image reprojection error parameters.
[0010] Preferably, the construction of the image reprojection error specifically includes: Projection function based on collinearity model The image reprojection error is: ; in, Indicates the first The external orientation elements of an image include position and posture. Indicates the camera's internal orientation parameters; Represents a point in three-dimensional space. Representing a spatial point In images The observed image point on; Image Set of observation points in the middle; This indicates the system error and additional parameters.
[0011] Preferably, the construction of the GNSS constraints specifically includes: Determine the geometric relationship between the lever arm and the image position: ; in, Indicates the first The camera center coordinates of the image. This indicates the location obtained from GNSS measurements. Represents the attitude rotation matrix. Represents the lever arm vector; Transform the above relationships into error form to construct GNSS constraints: .
[0012] Preferably, the thermal infrared imaging unit comprises 6 groups; the tilt camera group comprises 5 groups.
[0013] Preferably, the POS system supports BeiDou-based RTK.
[0014] The beneficial effects of this invention are: This invention addresses the urgent need for high-resolution imagery, high-precision 3D topography, and precise differentiation of ice, rock, and snow line types in monitoring ice and rock avalanche disasters in extremely high-altitude and cold regions. It provides an unmanned aerial vehicle (UAV) thermal infrared and oblique photography system adapted for these areas. By integrating the sensor structure and combining image control-free processing with a high-precision multi-source data synchronization mechanism, stable flight and reliable imaging are achieved at extremely high altitudes. This allows for the simultaneous acquisition and fusion processing of high-resolution imagery, detailed 3D morphology, and temperature field information. Consequently, decimeter-level positioning accuracy for oblique photography planes and elevations is achieved without control, resulting in multi-dimensional fusion perception results for ice and rock avalanche and snow line detection. This meets the application requirements for decimeter-level positioning of ice and rock masses and precise snow line identification in extremely high-altitude regions. Attached Figure Description
[0015] Figure 1 This is a schematic diagram of the integrated payload for UAV thermal infrared and oblique photogrammetry according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating the calibration process of the UAV thermal infrared and oblique photography equipment according to an embodiment of the present invention. Figure 3 This is a functional module architecture diagram of the UAV thermal infrared and oblique photography equipment according to an embodiment of the present invention.
[0016] Among them, 1. Five-lens tilting camera; 2. Thermal infrared imaging unit; 3. LiDAR payload. Detailed Implementation
[0017] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.
[0018] Example 1 This example presents a UAV thermal infrared and oblique photography equipment, the structural diagram of which is shown below. Figure 1As shown, the system includes a thermal infrared camera, a tilting camera group, a lidar, a POS system, and an uncontrolled photogrammetry system mounted side-by-side on the UAV. The thermal infrared camera integrates multiple thermal infrared imaging units. The uncontrolled photogrammetry system is used to dynamically and jointly calibrate the thermal infrared camera, tilting camera group, and lidar based on the POS system. It determines real-time positioning data using satellite-based differential signals received from the POS system. Based on the real-time positioning data, it controls the thermal infrared camera, tilting camera group, and lidar to generate tilted photogrammetric images, thermal infrared orthophotos, and 3D laser point clouds with a unified spatiotemporal reference for the operational area. It then combines calibrated exterior orientation elements, antenna arm parameters, and image reprojection error parameters to obtain visible light orthophotos and infrared temperature field data.
[0019] In this embodiment, the thermal infrared and oblique photography equipment includes three types of payloads: a five-lens oblique camera, a lidar payload, and a thermal infrared camera payload. Among them:
[0020] Figure 1 On the left is the five-lens tilt camera 1, with each lens having a 61-megapixel imaging unit, and the total effective pixels of the five-lens tilt camera is 350 million. For high-altitude and cold regions, the tilt angle of the sub-cameras has been optimized to enable them to acquire more comprehensive high-resolution texture information of the sides of unstable rock masses.
[0021] Figure 1 The central part consists of six thermal infrared imaging units, each with 2.07 million pixels and a temperature measurement sensitivity of ≤35mk. The total effective pixels are 12.44 million. The six thermal infrared imaging units enable the final acquired thermal infrared images to have both high resolution and a wide field of view, making them suitable for long-endurance UAV platforms.
[0022] Figure 1 On the right is lidar payload 3, with an effective range of 800 meters, a transmission frequency of 300kHz, and 7 echoes. It can penetrate vegetation and obtain accurate three-dimensional terrain information.
[0023] The integrated payload for UAV thermal infrared and oblique photogrammetry can simultaneously acquire five-camera oblique image data, thermal infrared camera data, and LiDAR data in a single operation. The oblique camera has a lateral field of view of 54.4 degrees, the thermal infrared payload has a lateral field of view of 65.8 degrees, and the LiDAR payload has a lateral field of view of 70 degrees. During operation, by designing the UAV flight path according to the overlap of the oblique cameras, the overlap requirements of the thermal infrared camera and LiDAR payload can be met. This enables the acquisition of high-resolution texture images of high-altitude building facades.
[0024] Specifically, such as Figure 2 As shown, the dynamic joint calibration of the thermal infrared camera, tilt camera group, and lidar of the unmanned aerial vehicle (UAV) photogrammetry system includes the following steps: The calibration test field is determined; reference images and ground control points of the calibration test field are acquired using other aerial cameras; absolute orientation is performed using ground control points to obtain geographic reference data and establish a collinearity model; aerial photography of the calibration test field is collected using thermal infrared and oblique photography equipment; multiple corresponding points of the images to be calibrated and the reference images are obtained using a multimodal automatic matching method, which serve as control points for the geometric calibration of the thermal infrared and oblique photography equipment.
[0025] Furthermore, based on the collinearity model, the geometric relationship between the image point and the ground point is established; the geometric relationship between the image point and the ground point is linearized to obtain the image point reprojection error; considering the spatial relationship between the sensor and the inertial navigation system, the stick arm error and attitude deviation parameters are introduced; the stick arm error and attitude deviation parameters are substituted into the collinearity model to establish a complete observation model, and the minimum image point reprojection error of the geometric calibration control point is taken as the objective function; the complete observation model is subjected to Taylor expansion to obtain the linearized equation, and the parameter correction is solved using the least squares method. Based on the objective function, iterative calculation is performed until convergence, and finally the sensor exterior orientation parameters, stick arm error parameters, and attitude deviation parameters are obtained, completing the flight dynamic calibration.
[0026] Joint calibration of thermal infrared and oblique photography equipment involves using the image space coordinate system of the oblique camera as a reference to reduce the phase centers of the thermal infrared camera, lidar, and GNSS antenna to the spatial coordinate system of the oblique camera, achieving spatiotemporal consistency of multi-source payload data. The main calibration contents include: the interior orientation elements and distortion coefficients of the oblique camera; the interior orientation elements and distortion coefficients of the thermal infrared imaging unit; the relative attitude of the oblique camera and the POS system; and the relative attitude of the lidar and the POS system. Because the thermal infrared and oblique photography equipment has a long focal length and many imaging units, conventional indoor or outdoor airborne calibration fields cannot obtain a sufficient number of control points within the sub-imaging units. Therefore, a dynamic calibration method is used to complete the joint calibration of the integrated equipment.
[0027] First, high-resolution aerial images of the calibration area are acquired using other aerial cameras, and high-precision georeferenced data is obtained by using a large number of ground control points for absolute orientation. The collinearity between ground points and image points is established, mathematically expressed as a three-dimensional similarity transformation model.
[0028] ; in, Represents the three-dimensional coordinates of the ground control points. Indicates the coordinates of the corresponding image point. Indicates the camera position. For rotation matrix elements, Focal length These are the pixel coordinates.
[0029] Second, a large number of corresponding points are obtained between the thermal infrared and oblique photography equipment and the reference image data through a multimodal automatic matching method. The spatial coordinates of the ground markers in the reference data are then used as the control point coordinates for the geometric calibration of the thermal infrared and oblique photography equipment. Let the matched corresponding point pairs be:
[0030] ; in, Represents a point in the reference image. This indicates the corresponding point in the image to be calibrated.
[0031] Establish the geometric relationship between image points and ground points based on the collinearity equation: ; Linearizing the above nonlinear equations yields the error equation: in, The vector of parameters to be estimated contains exterior orientation elements; It is a coefficient matrix; For observation vectors; This is the residual vector.
[0032] This step obtains a set of corresponding points from the cross-modal imagery, providing constraints for subsequent parameter solving.
[0033] Third, perform overall adjustment on all images from all camera sites to obtain the final calibration parameters for the thermal infrared and oblique photography equipment, and output the calibration parameters in a fixed format.
[0034] Based on the obtained corresponding points and reference coordinates, a global adjustment model incorporating multi-source observations is constructed. Considering the spatial relationship between the sensors and the inertial navigation system, lever arm error is introduced. With attitude deviation parameters Its expression is:
[0035] ; Simultaneously, an attitude deviation model is introduced: ; in, The coordinates of the camera's center of photography, For IMU position coordinates, For the camera pose rotation matrix, This is the IMU attitude rotation matrix.
[0036] Substituting the above parameters into the collinearity equation, a complete observation model is established, with the objective function being the minimum image point reprojection error: ; Performing a Taylor expansion on the model yields the linearized equations: ; Solving for parameter corrections using the least squares method: ; Through iterative calculations until convergence, the sensor orientation parameters, stick error parameters, and attitude deviation parameters are finally obtained, thus achieving flight dynamic calibration.
[0037] In the unmanned photogrammetry system for photographic equipment, the acquisition of exterior orientation elements, antenna mast parameters, and image reprojection error parameters of the photographic equipment includes: Acquire 3D laser point cloud and oblique photogrammetric image data, determine the set of spatial points and image observation points, and construct image reprojection error based on the projection function of the collinearity relationship model; Considering the fixed spatial offset between the camera and the GNSS antenna, determine the geometric relationship between the boom arm and the image position; transform the geometric relationship into an error form and construct GNSS constraints; By combining image reprojection error and GNSS constraints, a weighted overall optimization objective function is constructed. Taylor expansion of the complete observation model yields linearized equations, and the parameter corrections are solved using the least squares method. Based on the overall optimization objective function, iterative calculations are performed until convergence to obtain the optimal estimated exterior orientation elements, arm parameters, and image reprojection error parameters.
[0038] This invention integrates a high-precision satellite-based differential card into thermal infrared and oblique photography equipment, enabling it to directly obtain differential information from the satellite-based augmentation system during operation, replacing conventional ground-based differential network information.
[0039] This invention develops a thermal infrared and oblique photography equipment that integrates a high-precision POS system and supports BeiDou satellite-based RTK. During operation, the integrated equipment receives satellite-based differential signals in real time to acquire high-precision spatial position data from the oblique camera. By combining satellite-based differential positioning technology with lidar point cloud data, it replaces the geometric constraints of traditional image control points on oblique photogrammetry, enabling UAV-based oblique photogrammetry without image control points in extremely high mountain areas with multi-source constraints. The specific implementation process is as follows:
[0040] Utilizing satellite-based differential services (such as PPP-RTK and SBAS) of the Global Navigation Satellite System (GNSS), real-time positioning data at the centimeter to millimeter level is provided for thermal infrared and oblique photogrammetry equipment. This data serves as the initial values for the exterior orientation elements of the oblique image, reducing the accumulation of positioning errors. Three-dimensional point clouds of the operational scene are acquired through lidar point clouds, providing high-precision point cloud spatial coordinates. These point clouds can serve as "virtual control points," matched with corresponding points in the oblique image to constrain the image's rotation and translation parameters, compensating for the deficiencies of traditional image control points. Using the satellite-based differential positioning results as prior constraints for the exterior orientation elements, combined with feature matching information from the lidar point cloud and the oblique image, bundle adjustment optimizes the image's interior and exterior orientation elements, achieving a plane and elevation positioning accuracy of ≤1 meter in uncontrolled oblique photogrammetry in extremely high mountain areas. The core principle is as follows:
[0041] (1) Image reprojection error model Assume there is a total Zhang Image, Spatial Point Set as Image observation points are Then the projection function based on the collinearity equation is denoted as The image reprojection error is defined as: ; in, Indicates the first The external orientation elements of an image include position and posture. Indicates the camera's internal orientation parameters; Represents a point in three-dimensional space. Representing a spatial point In images The observed image point on; Image Set of observation points in the middle; This indicates the system error and additional parameters.
[0042] (2) Star-based difference and lever arm error constraint model Considering a fixed spatial offset (arm error) between the camera and the GNSS antenna, the geometric relationship is as follows: ; in, Indicates the first The camera center coordinates of the image. This indicates the location obtained from GNSS measurements. Represents the attitude rotation matrix. This represents the lever arm vector.
[0043] To standardize the coordinate representation, the above relationships are transformed into error form, and constraint functions are constructed as follows: ; (3) Multi-source joint optimization model Combining image reprojection error and GNSS constraints, an overall optimization objective function is constructed: ; in, This is the image reprojection error term. For GNSS and lever arm constraints, These are the weighting coefficients.
[0044] By performing Taylor expansion and linearization on the above nonlinear model, the error equation is constructed; the parameter correction is solved iteratively using the least squares method; until the convergence condition is met, the optimal estimated exterior orientation elements, arm parameters, and image reprojection error parameters are obtained.
[0045] This invention constructs a joint optimization model that integrates image observation, GNSS constraints, and point cloud constraints to achieve uncontrolled oblique photogrammetry with multi-source constraints, thereby significantly improving the automation and accuracy stability of measurements in complex environments.
[0046] Furthermore, this invention proposes unmanned aerial vehicle (UAV) photogrammetry software. After the thermal infrared and oblique photography equipment completes its operation, observation data from multiple sources can be acquired. Based on the UAV-based unmanned oblique photography method and the calibration results of the thermal infrared and oblique photography equipment, multi-source remote sensing data products with a unified spatiotemporal reference, such as thermal infrared orthophotos, visible light orthophotos, and 3D laser point clouds of the operational area, can be generated.
[0047] Based on the calibration parameters of the thermal infrared and oblique photography equipment, a 3D laser point cloud can be generated first. Based on the 3D laser point cloud and oblique photography image data, multi-source constrained image-controlled oblique photogrammetry is performed to obtain high-precision exterior orientation elements of the oblique photography image. Then, based on the high-precision exterior orientation elements, an oblique visible light orthophoto image is generated. Based on the acquired thermal infrared sensor image data, a large field-of-view stitched image of the thermal infrared sensor is generated. Combining the high-precision oblique photography exterior orientation elements, thermal infrared temperature field data of the operational scene can be generated. The software architecture is as follows: Figure 3 As shown, the system includes a satellite-based differential calculation module, a thermal infrared array stitching module, and a lidar and oblique photography data processing module. The satellite-based differential calculation module corrects the satellite-based differential data, ultimately outputting high-precision positioning results. The thermal infrared array stitching module uses the POS data output from the satellite-based differential calculation as a basis to stitch infrared images. The lidar and oblique photography data processing module supports dual-processing of visible light and lidar point clouds, outputting standardized mapping results.
[0048] Compared to existing aerial oblique photogrammetry or multi-sensor remote sensing equipment primarily designed for low- to mid-altitude applications, this invention integrates an oblique photogrammetry camera, thermal infrared sensor, or lidar. Furthermore, it addresses the core technical challenges of achieving high-precision measurements in extremely high-altitude, frigid mountainous regions where communication is limited, environments are extreme, and there is no image control. The invention incorporates targeted designs in system architecture, sensor collaboration methods, and data acquisition and processing modes. Existing technologies typically rely on ground-based differential mapping or ground control points to ensure photogrammetric accuracy, which can lead to problems such as time asynchrony, inconsistent spatial benchmarks, and accumulated data fusion errors in complex environments. This invention, however, is based on uncontrolled measurement applications at extremely high altitudes and constructs a unified multi-sensor installation benchmark and synchronous acquisition system. This enables the collaborative acquisition of oblique images, thermal infrared images, and lidar point clouds within the same spatiotemporal framework, providing a foundation for subsequent high-precision fusion and processing. Meanwhile, existing oblique photography equipment focuses on visible light imaging and three-dimensional geometric reconstruction, making it difficult to effectively identify the thermophysical differences of targets such as ice, rocks, and snow lines. However, this invention introduces thermal infrared sensing and combines it with high-resolution three-dimensional morphological information to achieve multi-dimensional feature fusion detection of ice and rock avalanches and snow lines, breaking through the limitations of single sensors or simple multi-sensor combinations in target identification dimensions.
[0049] In summary, this invention differs from existing similar technical solutions in terms of application targets, system architecture, multi-sensor collaboration methods, and adaptability to extremely high altitude environments, forming a technical route for multi-sensor integrated oblique photography equipment for fine detection of ice and rock avalanches and snow lines in extremely high mountain areas.
[0050] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
Claims
1. A UAV thermal infrared and oblique photography equipment, characterized in that, This includes a thermal infrared camera, a tilting camera group, a lidar, a POS system, and an uncontrolled photogrammetry system mounted side-by-side on the UAV; The thermal infrared camera integrates multiple thermal infrared imaging units; The unmanned aerial vehicle (UAV) photogrammetry system is used to dynamically and jointly calibrate thermal infrared cameras, tilt camera groups, and lidar based on a POS system. It determines real-time positioning data through satellite-based differential signals received by the POS system. Based on the real-time positioning data, it controls the thermal infrared cameras, tilt camera groups, and lidar to generate tilted photogrammetric images, thermal infrared orthophotos, and three-dimensional laser point clouds with a unified spatiotemporal reference for the work area. It also combines calibrated exterior orientation elements, antenna arm parameters, and image reprojection error parameters to obtain visible light orthophotos and infrared temperature field data.
2. The UAV thermal infrared and oblique photography equipment according to claim 1, characterized in that, The unmanned aerial vehicle (UAV) photogrammetry system dynamically calibrates the thermal infrared camera, tilt camera group, and lidar in conjunction with each other, including the following steps: Determine the calibration test site; Reference images and ground control points of the calibration test field are obtained using other aerial cameras; absolute orientation is performed using ground control points to obtain geographic reference data and establish a collinearity model; Aerial photography was used to collect images of the calibration test site to be calibrated using thermal infrared and oblique photography equipment. A multimodal automatic matching method is used to obtain multiple corresponding points between the image to be calibrated and the reference image, which serve as control points for the geometric calibration of thermal infrared and oblique photography equipment. Based on the collinearity model and control points, a lever error and attitude deviation model is introduced. With the minimum image point reprojection error as the objective function, the least squares method is used to iteratively calculate the external orientation parameters, lever error parameters, and attitude deviation parameters of each camera device in the output thermal infrared and oblique photography equipment.
3. The UAV thermal infrared and oblique photography equipment according to claim 2, characterized in that, In the unmanned aerial vehicle (UAV) photogrammetry system, the geographic reference data is a collinearity model between ground points and image points in high-resolution aerial photographs, and the collinearity model is a three-dimensional similarity transformation model.
4. The UAV thermal infrared and oblique photography equipment according to claim 3, characterized in that, In the unmanned aerial vehicle (UAV) photogrammetry system, the least squares method is used to iteratively calculate the internal and external orientation parameters, antenna arm error parameters, and attitude deviation parameters of each camera device in the thermal infrared and oblique photography equipment. Specifically, these parameters include: Based on the collinearity model, the geometric relationship between image points and ground points is established; Linearize the geometric relationship between the image point and the ground point to obtain the image point reprojection error; Considering the spatial relationship between the sensor and the inertial navigation system, lever error and attitude deviation parameters are introduced; Substitute the lever error and attitude deviation parameters into the collinearity model to establish a complete observation model, and take the minimum image point reprojection error of the geometric calibration control point as the objective function. Taylor expansion of the complete observation model yields linearized equations, and the parameter corrections are solved using the least squares method. Based on the objective function, iterative calculations are performed until convergence, ultimately obtaining the sensor orientation parameters, stick error parameters, and attitude deviation parameters, thus completing the flight dynamic calibration.
5. The UAV thermal infrared and oblique photography equipment according to claim 3, characterized in that, In the unmanned aerial vehicle (UAV) photogrammetry system, the acquisition of the exterior orientation elements, antenna mast parameters, and image reprojection error parameters specifically includes: Acquire 3D laser point cloud and oblique photogrammetric image data, determine the set of spatial points and image observation points, and construct image reprojection error based on the projection function of the collinearity relationship model; Considering the fixed spatial offset between the camera and the GNSS antenna, determine the geometric relationship between the boom arm and the image position; transform the geometric relationship into an error form and construct GNSS constraints; By combining image reprojection error and GNSS constraints, a weighted overall optimization objective function is constructed. Taylor expansion of the complete observation model yields linearized equations, and the parameter corrections are solved using the least squares method. Based on the overall optimization objective function, iterative calculations are performed until convergence to obtain the optimal estimated exterior orientation elements, arm parameters, and image reprojection error parameters.
6. The UAV thermal infrared and oblique photography equipment according to claim 5, characterized in that, The construction of the image reprojection error specifically includes: Projection function based on collinearity model The image reprojection error is: ; in, Indicates the first The external orientation elements of an image include position and posture. Indicates the camera's orientation parameters; Represents a point in three-dimensional space. Representing a spatial point In images The observed image point on; Image Set of observation points in the middle; This indicates the system error and additional parameters.
7. The UAV thermal infrared and oblique photography equipment according to claim 5, characterized in that, The construction of the GNSS constraints specifically includes: Determine the geometric relationship between the lever arm and the image position: ; in, Indicates the first The camera center coordinates of the image. This indicates the location obtained from GNSS measurements. Represents the attitude rotation matrix. Represents the lever arm vector; Transform the above relationships into error form to construct GNSS constraints: 。 8. The UAV thermal infrared and oblique photography equipment according to claim 1, characterized in that, The thermal infrared imaging unit consists of 6 groups; the tilt camera group consists of 5 groups.
9. The UAV thermal infrared and oblique photography equipment according to claim 1, characterized in that, The POS system supports BeiDou-based RTK.
