Global map-guided digital pan-tilt camera intersection measurement method and device

Abstract

This invention relates to a global map-guided digital pan-tilt camera intersection measurement method and apparatus. The method includes: determining the area to be measured using images acquired by a global camera; generating coarse pointing control quantities for each digital pan-tilt camera station based on a pre-calibrated coordinate relationship between the global camera and two digital pan-tilt camera stations; controlling the low-precision pan-tilt units of each digital pan-tilt camera station to rotate according to the coarse pointing control quantities, so that a local field-of-view telephoto camera mounted on the pan-tilt unit covers the area to be measured; using an angle measurement camera mounted on the pan-tilt unit's fixed support to observe a cooperative marker block fixed to the pan-tilt unit's rotating component, and determining the rotation angle or attitude of the pan-tilt unit's rotating component; based on the rotation angle or attitude, and combined with a pre-calibrated rigid transformation relationship, recovering the real-time exterior orientation parameters of the telephoto camera; acquiring local images containing the points to be measured by the two telephoto cameras whose real-time exterior orientation parameters have been recovered, extracting pixel coordinates, constructing corresponding spatial lines of sight, and performing spatial intersection calculations to obtain the three-dimensional coordinates of the points to be measured.

Description

Technical Field

[0001] This invention belongs to the fields of photogrammetry, visual measurement, precision photoelectric measurement and gimbal control technology. Specifically, it relates to a method and apparatus for guiding a large target area using a global camera, performing long-distance local high-resolution imaging using a digital gimbal camera, and realizing spatial intersection measurement based on two or more digital gimbal camera stations. Background Technology

[0002] In 3D measurements of large-scale structures, distant targets, or large-scale scenes, a single wide-angle camera can cover a large area, but due to limitations in object resolution, distant targets typically occupy only a few pixels in the image, making it difficult to meet the precision measurement requirements at the millimeter or sub-millimeter level. Using a telephoto camera can improve the resolution of distant objects, but telephoto cameras have a narrow field of view, and without effective guidance, it is difficult to quickly locate targets in large scenes, especially in continuous measurements at multiple points, where there are problems such as low search efficiency, easy omissions, and insufficient automation.

[0003] Existing motorized pan-tilt units can be used to drive telephoto cameras to point at target areas, but the angle coding accuracy, hysteresis, assembly and adjustment errors, and repeatability of low-cost or low-precision pan-tilt units are difficult to directly meet the requirements of precise intersection measurements. If the pan-tilt control angle is directly used as the exterior azimuth angle of the telephoto camera, the mechanical error of the pan-tilt unit will be directly converted into spatial line-of-sight error, resulting in a decrease in the accuracy of intersection measurements. Although high-precision pan-tilt units can improve the above problems, they are costly, bulky, and complex to deploy, which is not conducive to large-scale deployment with multiple stations.

[0004] Furthermore, multi-camera intersection measurements typically require that the lines of sight of each camera accurately point to the same target point, and that the extrinsic parameters of each camera be stable and known. For composite measurement systems consisting of a global camera, multiple pan-tilt cameras, close-range angle measurement cameras, and cooperative marker blocks, there is still a lack of systematic technical solutions for establishing a unified coordinate relationship, quickly guiding multiple narrow-field telephoto cameras to the target area using global images, accurately recovering the actual pointing of each telephoto camera under low-precision pan-tilt conditions, and further completing multi-station intersection calculations. Summary of the Invention

[0005] The purpose of this invention is to provide a global map-guided digital gimbal camera intersection measurement method and device. By combining "global camera large-scale discovery and guidance, low-precision gimbal coarse coverage, cooperative marker block visual angle measurement compensation, long-focus local high-resolution imaging, and multi-station spatial intersection", it can achieve rapid guidance, accurate pointing recovery, and high-precision three-dimensional coordinate measurement of any point to be measured in a large-scale scene without relying on a high-precision mechanical gimbal.

[0006] The technical solution of this invention is the global map-guided digital pan-tilt camera intersection measurement method, which is characterized by including the following steps:

[0007] (1) Survey area layout and coordinate system establishment: Use a global camera to collect images of a large area and determine the points or areas to be measured;

[0008] (2) Generate coarse pointing control quantities for each digital pan-tilt camera station based on the pre-calibrated coordinate relationship between the global camera and at least two digital pan-tilt camera stations.

[0009] (3) According to the coarse pointing control amount, control the low-precision gimbal of each digital gimbal camera station to turn so that the local field of view telephoto camera installed on the gimbal covers the point to be measured or the area to be measured.

[0010] (4) Use an angle measuring camera mounted on the gimbal fixed bracket to observe the cooperative marker block fixed to the gimbal rotating component, and determine the actual rotation angle or actual attitude of the gimbal rotating component based on the observed image.

[0011] (5) Based on the actual rotation angle or actual attitude, and in conjunction with the pre-calibrated rigid transformation relationship, recover the real-time exterior orientation parameters of the telephoto camera;

[0012] (6) At least two telephoto cameras with recovered real-time exterior orientation parameters acquire local images containing the point to be measured, extract the pixel coordinates of the point to be measured, construct the corresponding spatial lines of sight, and perform spatial intersection calculation to obtain the three-dimensional coordinates of the point to be measured.

[0013] As a preferred option: the pre-calibrated content in step (2) includes at least: global camera intrinsic and extrinsic parameters, initial external orientation parameters of each telephoto camera, intrinsic and extrinsic parameters of each corner measurement camera, geometric parameters of the cooperative marker block, installation relationship between the telephoto camera and the corner measurement camera, and baseline relationship of each digital pan-tilt camera station under a unified measurement coordinate system.

[0014] The following four transformation matrices can be obtained through calibration:

[0015] ;

[0016] In the formula, This is the extrinsic parameter transformation matrix from the global camera coordinate system to the unified measurement coordinate system. For the first When the coordinate system of the telephoto camera at each station is initially at zero position, relative to the initial calibration extrinsic parameter matrix of the unified measurement coordinate system, The extrinsic parameter transformation matrix of the cooperative marker block coordinate system relative to the angle measurement camera coordinate system. The extrinsic transformation matrix is ​​used between the angle measurement camera coordinate system and the telephoto camera coordinate system; the global camera coordinate system is... A unified measurement coordinate system is , No. The coordinate system of the telephoto local camera at each digital gimbal camera station is: The coordinate system of the angle measurement camera is The coordinate system of the cooperation marker block is ;

[0017] The digital pan-tilt camera station described in step (2) includes: a low-precision pan-tilt unit, a local field-of-view telephoto camera mounted on the pan-tilt unit, a cooperative marker block fixedly connected to the pan-tilt unit's rotating components, and one or more angle measuring cameras mounted on the pan-tilt unit's fixed bracket.

[0018] The coarse pointing control quantity in step (2) is calculated based on the pixel coordinates of the point to be measured in the global image, the global camera exterior orientation parameters, and the three-dimensional prior information or control point constraints of the survey area. The coarse pointing control quantity is only used to make the local field of view of the telephoto camera cover the target area, and is not used as the precise angle observation value in the intersection solution.

[0019] As a preferred method, the actual rotation angle or actual attitude of the gimbal rotation component in step (4) can be obtained by performing PnP solution, homography decomposition, and multi-marker feature fusion on the cooperative marker block image captured by the rotation angle measurement camera, or by using the observation data of multiple rotation angle measurement cameras for joint optimization solution.

[0020] During measurement, the angle measuring camera calculates its real-time pose relative to the angle measuring camera based on the cooperative marker block image, and recovers the real-time pose of the telephoto camera from a rigid relationship; this relationship can be expressed as:

[0021] ;

[0022] In the formula, At time t, the extrinsic transformation matrix from the coordinate system of the angle measurement camera to the coordinate system of the telephoto camera. At time t, the angle measurement camera captures and identifies the real-time pose matrix of the cooperative marker block, which is then calculated. This is the extrinsic transformation matrix between the telephoto camera and the cooperative marker block; combining the pre-calibrated transformation relationship between the angle measurement camera and the global camera, and the transformation relationship between the global camera and the unified measurement coordinate system, the real-time pose of the telephoto camera is recovered.

[0023] ;

[0024] In the formula, At time t, the first The extrinsic transformation matrix of a telephoto camera relative to a unified measurement coordinate system. This is the extrinsic transformation matrix of the global camera coordinate system relative to the unified measurement coordinate system. For the first The extrinsic transformation matrix of the angle measurement camera coordinate system relative to the global camera coordinate system. For time t, the first The extrinsic transformation matrix of the telephoto camera coordinate system relative to the angle measurement camera coordinate system. This is the rotation matrix of the telephoto camera in a unified measurement coordinate system. This is the translation vector of the telephoto camera in a unified measurement coordinate system;

[0025] The point to be measured is at the 1st The normalized image points in a telephoto camera image are represented as follows:

[0026] ;

[0027] In the formula, For the point to be measured at the th Normalized camera coordinates under a telephoto camera For the first The inverse matrix of the intrinsic parameter matrix of a telephoto camera. Let be the horizontal pixel coordinates of the point to be measured on the telephoto camera image. The vertical pixel coordinates of the point to be measured on the telephoto camera image;

[0028] Combined with the telephoto camera extrinsic transformation matrix recovered in real time in step (5) Transform it to a unified measurement coordinate system Below, the spatial line of sight can be represented as:

[0029] ;

[0030] In the formula, For depth parameters, To achieve the first under a unified measurement coordinate system, A three-dimensional unit direction vector pointing from a telephoto camera to the point to be measured. Let be the coordinates of the optical center of the i-th camera in the unified measurement coordinate system.

[0031] The unit direction vector is determined by the real-time pose of the telephoto camera and the normalized image points. For n≥2 lines of sight, the coordinates of the point to be measured can be obtained by minimizing the sum of the squared perpendicular distances from each line of sight to the point to be measured.

[0032] ;

[0033] In the formula, I is a 3×3 identity matrix;

[0034] When introducing error weights, a weight matrix can be established based on the image point extraction accuracy of the telephoto camera, the attitude estimation accuracy of the angle measurement camera, the baseline length of the station, the line-of-sight intersection angle, and the calibration residual. Different weights can be assigned to the lines of sight of different stations to improve the robustness of the intersection solution.

[0035] As a preferred embodiment, the real-time external orientation parameters of the telephoto camera in step (5) are determined by the pose of the cooperative marker block obtained by the angle measurement camera, the rigidity relationship between the cooperative marker block and the gimbal rotation component, the rigidity relationship between the telephoto camera and the gimbal rotation component, and the station external parameters.

[0036] As a preferred method, the spatial intersection solution in step (6) adopts the least squares, weighted least squares, or robust estimation method; wherein, the weight of each spatial line of sight involved in the solution is determined based on at least one of the following factors: the image point extraction accuracy of the corresponding telephoto camera, the angle measurement accuracy, the line of sight intersection angle, the station baseline length, or the calibration residual.

[0037] Preferably, the method further includes: repeatedly performing the steps of global map point selection, gimbal coarse pointing, visual measurement of actual gimbal rotation angle, telephoto local imaging, and multi-station intersection calculation, so as to achieve continuous and automated measurement of multiple test points within the global camera coverage area.

[0038] As a preferred method, by registering the image acquired by the global camera with the three-dimensional model of the object under test, the pixel coordinates of the selected test point in the global image are mapped to the surface of the three-dimensional model to obtain the coarse three-dimensional coordinates of the test point, and the coarse pointing control quantity is generated for each digital pan-tilt camera station based on the coarse three-dimensional coordinates.

[0039] Another technical solution of the present invention is the global map-guided digital gimbal camera intersection measurement device for implementing any of the foregoing methods, characterized in that it includes:

[0040] A global imaging module is used to acquire images of a large area.

[0041] At least two digital gimbal camera stations, each station including a low-precision gimbal, a local field-of-view telephoto camera mounted on the gimbal, and a cooperative marker block fixed to the rotating parts of the gimbal;

[0042] The rotation angle measurement module includes one or more close-range rotation angle measurement cameras mounted on a fixed bracket at each digital pan-tilt camera station, used to observe the cooperative marker block and output the actual rotation angle of the pan-tilt rotating component;

[0043] The calibration parameter module is used to store pre-calibrated camera parameters and rigid transformation relationships between components;

[0044] The guidance and control module is connected to the global imaging module, the calibration parameter module and the digital pan-tilt camera station, respectively. It is used to generate coarse pointing control commands based on the points to be measured in the global image and drive the low-precision pan-tilt unit to rotate.

[0045] The local image measurement module is used to control the telephoto camera to acquire local images and extract the sub-pixel level image point coordinates of the points to be measured;

[0046] The intersection solution module is connected to the calibration parameter module, the angle precision measurement module, and the local image measurement module, respectively. It is used to solve the three-dimensional coordinates of the point to be measured by spatial line intersection based on the real-time pointing parameters of at least two stations and the coordinates of local image points.

[0047] The results management module is used to store, stitch together, compare and analyze, and output the three-dimensional coordinates, quality evaluation indicators, and multi-period measurement results of multiple test points.

[0048] Preferably, the cooperative marker block is a planar checkerboard, a dot array, a coded marker, an AprilTag / ArUco type marker, a three-dimensional marker block, or a combination thereof; the angle measurement camera is one or more cameras fixedly mounted on a gimbal bracket or a component that maintains a rigid relationship with the gimbal base, used to observe the same cooperative marker block from different perspectives or to observe multiple cooperative marker blocks separately.

[0049] Preferably, the intersection solution module is also used to calculate and output the intersection residual, line-of-sight angle, coordinate covariance, or measurement reliability as measurement quality evaluation indicators; the result management module is used to store, stitch together, compare and analyze, and output the three-dimensional coordinates, quality evaluation indicators, and multi-period measurement results of multiple test points.

[0050] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0051] (1) Balancing wide-area coverage and long-distance high-resolution measurement: The global camera is responsible for large-area target discovery and guidance, while the telephoto local camera is responsible for high object surface resolution imaging, thus resolving the contradiction between the insufficient resolution of the wide-angle camera and the difficulty of searching by the telephoto camera.

[0052] (2) Reduce reliance on high-precision gimbals: Low-precision gimbals are only used for coarse coverage, and the actual pointing angle is obtained by observing cooperative marker blocks with a close-range angle measurement camera, which can significantly reduce gimbal costs and engineering deployment difficulties.

[0053] (3) Improve the accuracy of multi-station intersection: By restoring the real-time attitude of the telephoto camera through visual angle measurement and combining the spatial line-of-sight intersection of two or more digital pan-tilt cameras, the three-dimensional coordinates and quality evaluation results of the point to be measured can be obtained.

[0054] (4) Supports large-scale continuous measurement: By repeating the process of selecting points on the global map and intersecting multiple stations, multi-point, batch, automated, and high-precision measurement can be achieved within the coverage area of ​​the global camera.

[0055] (5) The system is highly scalable: the number of angle measurement cameras, the form of cooperative marker blocks, the number of telephoto cameras, the number of measurement stations, and the calculation algorithm can all be expanded according to engineering requirements.

[0056] (6) Achieving a balance between cost and accuracy: This breakthrough enables the acquisition of high-precision measurement results using low-precision hardware. In actual testing, using a low-precision gimbal costing several hundred yuan, combined with a visual angle measurement module, the final pointing accuracy can rival that of a high-precision servo gimbal costing tens of thousands of yuan, greatly reducing the barrier to system deployment.

[0057] (7) System robustness and scalability: Since the actual pointing relies on visual feedback rather than open-loop control of the gimbal, the system is insensitive to environmental changes (such as slight gimbal deformation caused by temperature) and individual gimbal differences. At the same time, the system architecture naturally supports the expansion of any number of stations. Adding stations only requires calibrating their transformation relationship with the global coordinate system, which can improve the intersection accuracy and coverage, demonstrating extremely strong system flexibility. Attached Figure Description

[0058] Figure 1 This is a schematic diagram illustrating the overall principle of digital gimbal camera intersection measurement guided by the global map of this invention. Figure 2 This is a schematic diagram of the digital gimbal camera structure of the present invention; Figure 3 This is a schematic diagram of the pre-measurement calibration process of the present invention; Figure 4 This is a flowchart of the global map-guided and multi-station intersection measurement method of the present invention; Figure 5 This is a block diagram of the functional modules of the device of the present invention.

[0059] Explanation of key component symbols: . Detailed Implementation

[0060] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0061] This invention provides a global map-guided digital pan-tilt camera intersection measurement method, which includes the following system components and implementation steps.

[0062] Device Composition: The device includes at least one global camera observation station 1, at least two digital pan-tilt camera stations 2, a unified calibration module, a global map guidance and control module, a pan-tilt angle visual measurement module, a local image measurement module, and a intersection calculation module. The global camera observation station is used to acquire images of a large area and provide coarse positioning information for the point or area to be measured. The digital pan-tilt camera station 2 includes a low-precision pan-tilt unit 5, a local field-of-view telephoto observation camera 3 mounted on the low-precision pan-tilt unit 5, a cooperative marker block 8 fixedly connected to the pan-tilt rotation mechanism 4, and one or more close-range angle measurement cameras 10 mounted on the pan-tilt bracket or fixed base 11.

[0063] The cooperative marker block 8 can be a planar checkerboard pattern, a dot array, a coded marker, an AprilTag / ArUco type marker, a three-dimensional marker block, or other reliably identifiable artificial marker. A close-range rotation angle measurement camera 10 is used to observe the image changes of the cooperative marker block 8 during the rotation of the low-precision gimbal 5, thereby determining the actual rotation angle of the gimbal or the real-time attitude of the gimbal rotation mechanism 4 relative to the fixed support 9. A local field-of-view telephoto camera 3 is used to perform high-resolution imaging of the target area indicated by the global map.

[0064] Please see Figure 2 As shown, the digital gimbal camera includes: a telephoto observation camera 3, a low-precision gimbal 5, an azimuth axis 6, a pitch axis 7, a cooperative marker block 8, a fixed bracket 9, an angle measurement camera 10, and a base 11. The base 11 is fixedly installed at the station location; the low-precision gimbal 5 is mounted on the base 11 and is equipped with an azimuth axis 6 that can rotate around a vertical axis and a pitch axis 7 that can rotate around a horizontal axis; the telephoto observation camera 3 is fixedly installed on the rotation mechanism 4 of the low-precision gimbal 5, and its optical axis changes direction with the azimuth and pitch movements of the low-precision gimbal 5; the cooperative marker block 8 is rigidly fixed to the mechanism 4 of the low-precision gimbal 5, maintaining a fixed relative pose with the telephoto observation camera 3; the fixed bracket 9 is rigidly connected to the base 11 or the fixed part of the gimbal; the angle measurement camera 10 is mounted on the fixed bracket 9, and its field of view covers the cooperative marker block 8, used to capture images of the cooperative marker block 8 in real time to calculate the actual angle or attitude of the gimbal rotation mechanism 4.

[0065] Measurement method:

[0066] Pre-measurement calibration: Before measurement, control points are set up in the common visible area of ​​the global camera observation station and each digital pan-tilt camera station 2, or a unified measurement coordinate system is established using natural feature points, artificial marker points, and total station measurement points with known three-dimensional coordinates. Through camera calibration and spatial resection, the intrinsic and extrinsic parameters of the global camera (wide-angle) 1.1 and each local telephoto camera relative to the unified measurement coordinate system in the initial state are obtained.

[0067] For each digital pan-tilt camera station 2, the rigid mounting relationship between the local field-of-view telephoto camera 3 and the pan-tilt angle measurement camera is further calibrated, the relationship between the coordinate system of the cooperative marker block and the coordinate system of the pan-tilt rotation mechanism 4 is calibrated, and the geometric relationship between the pan-tilt rotation axis system and the optical axis of the telephoto camera is determined. If multiple angle measurement cameras 10 are set up at a station, the accuracy and robustness of the marker block pose estimation are improved by joint observation of multiple cameras.

[0068] The calibration results include, but are not limited to: global camera intrinsic and extrinsic parameters, intrinsic and initial extrinsic parameters of each local telephoto camera, intrinsic parameters of the close-range rotation measurement camera and its extrinsic parameters relative to the support coordinate system, geometric model of the cooperative marker block, installation transformation between the telephoto camera and the gimbal rotation component, baseline relationship between stations, and error weight model.

[0069] Global map guidance and coarse digital PTZ pointing:

[0070] After the global camera acquires images of a large area, the operator can select specific points to be measured within the global image, or the image recognition algorithm can automatically extract structural feature points, marker points, edge intersections, or the location of defects to be measured. Based on the calibration parameters of the global camera, the pixel coordinates of the points to be measured in the global image are converted into target rays or target region constraints in a unified measurement coordinate system. Combined with the prior geometric model of the area to be measured, control point planes, structural models, or existing 3D information, the approximate direction of the target area in the coordinate system of each digital pan-tilt camera station is estimated.

[0071] Since this stage only requires the telephoto camera to cover the target area, and does not require the low-precision gimbal to directly achieve precise angle measurement accuracy, coarse pointing control commands can be sent to the low-precision gimbal based on the estimated direction, so that the local fields of view of at least two digital gimbal cameras cover the target area. This design decouples "target detection and pointing planning" from "precise angle measurement and precise intersection," thereby reducing the dependence on the mechanical control accuracy of the gimbal.

[0072] Actual gimbal rotation angle visual measurement:

[0073] After the low-precision gimbal 5 rotates to the target area, the close-range angle measurement camera 10, mounted on the fixed bracket 9, captures the image of the cooperative marker block 8, which is fixed to the gimbal rotation mechanism 4. Based on the known geometric dimensions and image coordinates of the cooperative marker block 8, the real-time pose of the cooperative marker block 8 relative to the close-range angle measurement camera 10 is obtained through PnP solving, homography decomposition, multi-marker fusion, or multi-camera joint optimization.

[0074] By utilizing the pre-calibrated rigid transformation relationship between the cooperative marker block 8, the gimbal rotation mechanism 4, the local telephoto observation camera 3, and the angle measurement camera 10, the pose of the cooperative marker block 8 is converted into the actual pose or actual pointing angle of the telephoto local camera. This actual pointing angle is no longer equivalent to the control command angle or encoding angle of the low-precision gimbal 5, but is determined by the close-range visual measurement results, which can compensate for gimbal backlash, assembly error, control error, and low-precision angle encoding error.

[0075] Local high-resolution imaging and intersection solution:

[0076] When the telephoto local fields of view of at least two digital pan-tilt cameras cover the same test point, each telephoto local camera acquires an image of the target area and extracts the high-precision pixel coordinates of the test point from the local image. These pixel coordinates can be obtained through manual point selection, template matching, sub-pixel corner point extraction, circle center extraction, edge fitting, correlation matching, or deep learning detection.

[0077] For the first A digital gimbal camera station 2 constructs the spatial line of sight corresponding to the point to be measured based on its telephoto camera intrinsic parameters, the actual attitude determined by the angle measuring camera 10, the station position, and the calibration installation relationship.

[0078] For multiple spatial lines of sight obtained from two or more stations, the three-dimensional coordinates of the point to be measured can be solved by least squares intersection, weighted least squares, robust estimation or error ellipsoid constraint method, and the intersection residual, line of sight intersection angle, coordinate covariance or quality evaluation index can be output.

[0079] By repeatedly executing the steps of global map point selection, gimbal coarse pointing, corner visual fine measurement, local high-resolution imaging, and multi-station intersection calculation, continuous high-precision three-dimensional measurement of a large area to be measured within the field of view of the global camera can be performed.

[0080] A global graph-guided digital pan-tilt-zoom (PTZ) camera intersection measurement method includes the following steps: S1. Survey area layout and coordinate system establishment: Select installation positions for global camera observation station 1 and at least two digital pan-tilt camera stations 2 around the area to be measured, ensuring that the global camera can cover a large area to be measured, and that two or more digital pan-tilt cameras observe the same target area from different baseline directions; set up control points or identifiable feature points in the common visible area of ​​the global camera and digital pan-tilt cameras, and establish a unified measurement coordinate system. S2, Installation of global camera and digital pan-tilt station: The global camera is fixedly installed at a high or long-distance observation position; the local field-of-view telephoto observation camera 3 is installed on the rotation mechanism 4 of the low-precision pan-tilt station 5, so that its optical axis can rotate with the azimuth and pitch axes of the pan-tilt station; the cooperative marker block 8 is rigidly installed on the pan-tilt station rotation mechanism 4; the angle measurement camera 10 is fixed on the installation structure that maintains a rigid relationship with the base 11 or the pan-tilt station bracket, so that the field of view of one angle measurement camera 10 can cover multiple cooperative marker blocks 8; S3, Camera Intrinsic and Extrinsic Parameter Calibration: Perform intrinsic parameter calibration on the global camera, each telephoto local camera and the corner measurement camera 10 respectively to obtain focal length, principal point, distortion parameters, etc.; use common control points or known three-dimensional points to perform spatial resection to obtain the extrinsic orientation parameters of the global camera and each digital pan-tilt station in a unified measurement coordinate system. S4, Internal rigidity calibration of digital gimbal: Calibrate the installation transformation between the telephoto local camera and the gimbal rotating parts, calibrate the fixed relationship between the coordinate system of the cooperative marker block and the coordinate system of the gimbal rotating mechanism 4, calibrate the fixed relationship between the angle measuring camera 10 and the base 11 or fixed bracket 9, and establish the geometric constraint model between the gimbal azimuth axis, pitch axis, telephoto camera optical axis and cooperative marker block 8. S5, Global Image Acquisition and Test Point Selection: The global camera acquires a large-scale image covering the test area. The operator selects the test spatial point or test area in the global image. Alternatively, the image recognition algorithm can automatically extract structural test points, marker points, edge intersections, crack endpoints, or defect feature points to obtain the pixel coordinates of the test point in the global image. S6, Coarse positioning of target area and coarse pointing planning of gimbal: Based on the global camera calibration parameters, pixel coordinates of the point to be measured, control points of the measurement area or three-dimensional prior model of the structure, estimate the coarse spatial position of the target area where the point to be measured is located in a unified measurement coordinate system; then, based on the position and initial external parameters of each digital gimbal station, calculate the coarse azimuth and coarse elevation angle of each station, and generate coarse pointing control commands for the gimbal. S7, Low-precision gimbal turning and local field of view coverage: Control at least two low-precision gimbals 5 to rotate according to coarse directional control commands so that the narrow field of view of the telephoto local camera covers the target area; this step only requires the target area to enter the field of view of the telephoto camera, and does not require the control angle of the low-precision gimbal 5 to be directly used as the precise angle observation value. S8, Visual precision measurement of actual gimbal rotation angle: After the gimbal rotates, the rotation angle measurement camera 10 captures an image of the cooperative marker block 8 and extracts the corner points, center points, coded corner points, or other identifiable features of the marker block; based on the geometric model of the cooperative marker block 8, PnP solution, homography decomposition, multi-marker fusion, or nonlinear optimization are used to obtain the actual attitude of the gimbal rotation mechanism 4 relative to the fixed support 9, thereby obtaining the actual azimuth angle and the actual pitch angle; S9, Real-time exterior orientation recovery of telephoto camera: Using the actual attitude of the gimbal obtained by the angle measurement camera 10, combined with the installation relationship between the telephoto local camera and the gimbal rotation mechanism 4 calibrated before measurement, the real-time exterior orientation parameters of the telephoto local camera in the unified measurement coordinate system are updated to obtain the actual optical center position and actual observation direction of the telephoto camera. S10, High-precision local image point extraction: Each telephoto local camera acquires high-resolution local images of the target area, and performs sub-pixel level extraction of the points to be tested in the local images; for cooperative landmark points, center fitting, corner point extraction or encoding recognition can be used; for natural structure points, template matching, edge fitting, correlation matching or deep learning detection can be used, and the image point coordinates and their quality indicators are output. S11, Multi-station spatial line of sight construction and intersection calculation: Based on the real-time exterior orientation parameters, intrinsic parameters and pixel coordinates of the point to be measured by each telephoto local camera, construct the spatial observation line of sight corresponding to the point to be measured; perform least squares intersection, weighted least squares intersection or robust estimation on the spatial lines of sight of two or more stations to obtain the three-dimensional coordinates of the point to be measured, and calculate the intersection residual, line of sight intersection angle, coordinate covariance and measurement reliability; S12, Batch Measurement Point Loop and Result Management: Repeatedly executes the following steps for multiple measurement points within the global camera's field of view: measurement point selection, gimbal coarse pointing, actual angle fine measurement, local high-resolution imaging, and multi-station intersection calculation, to form large-scale, high-resolution, and high-precision three-dimensional measurement results; The result management module stores, stitches, compares, and outputs measurement point coordinates, quality evaluation, repeated measurement data, and multi-period measurement results.

[0081] The present invention also provides a global map-guided digital gimbal camera intersection measurement device, comprising: A global imaging module for acquiring images of a large area; at least two digital pan-tilt camera stations 2, each station including a low-precision pan-tilt 5, a local field-of-view telephoto observation camera 3 mounted on the low-precision pan-tilt 5, and a cooperative marker block 8 fixedly connected to the pan-tilt rotation mechanism 4. The rotation angle measurement module includes one or more close-range rotation angle measurement cameras 10 installed on the fixed bracket 9 of each digital pan-tilt camera station 2, used to observe the cooperative marker block 8 and output the actual rotation angle of the pan-tilt rotation mechanism 4; the calibration parameter module is used to store the pre-calibrated parameters of each camera and the local image measurement module is used to control the telephoto camera to acquire local images and extract the sub-pixel level image point coordinates of the point to be measured; The intersection solution module is connected to the calibration parameter module, the angle precision measurement module, and the local image measurement module, respectively. It is used to solve the three-dimensional coordinates of the point to be measured by spatial line intersection based on the real-time pointing parameters of at least two stations and the coordinates of local image points. The results management module is used to store, stitch together, compare and analyze, and output the three-dimensional coordinates, quality evaluation indicators, and multi-period measurement results of multiple test points; each module can be composed of rigid transformation relationships between computers, embedded processors, edge computing devices, industrial cameras, lenses, pan-tilt units, brackets, cooperative marker blocks, communication units, and measurement components. The guidance and control module is connected to the global imaging module, the calibration parameter module and the digital pan-tilt camera station, respectively. It is used to generate coarse pointing control commands based on the points to be measured in the global image and drive the low-precision pan-tilt unit to rotate. Each module is implemented by a computer, embedded processor, edge computing device, industrial camera, lens, gimbal, bracket, cooperative marker block, communication unit and measurement software.

[0082] Example 1: Long-distance deformation measurement of fixed infrastructure.

[0083] This embodiment provides a specific application of the method of the present invention in bridge deformation monitoring.

[0084] The global camera is installed at a high position outside the survey area, covering the entire area of ​​bridges, slopes, tunnel entrances, or large structures; two or more digital pan-tilt telephoto cameras are deployed at different baseline positions; the operator selects the measurement points to be measured in the global image in sequence, and the system automatically guides multiple telephoto observation cameras to cover the measurement points and complete spatial intersection, thereby obtaining the three-dimensional displacement of multiple measurement points.

[0085] First, the survey area is set up. A global camera (wide-angle) 1.1 is set up in a stable area on one side of the bridge. This camera is equipped with a wide-angle lens, and its field of view can cover the entire bridge. On the stable bedrock on both sides of the bridge, a first digital pan-tilt camera station 2 and a second digital pan-tilt camera station 2 are set up respectively. Each station includes a low-precision, low-cost two-axis pan-tilt head, on which a long-focus observation camera 3 (e.g., 50mm focal length or above) is mounted. A planar checkerboard marker block is rigidly connected to the pan-tilt head rotation mechanism 4. A close-range angle measurement camera 10 is mounted on the fixed bracket 9 of the pan-tilt head, and the field of view of this camera is directly facing the checkerboard marker block. Secondly, pre-measurement calibration is performed; the three-dimensional coordinates of multiple control points around the bridge are measured using a total station to establish a unified measurement coordinate system; the control points are also clearly imaged by the global camera and two telephoto observation cameras 3; through resection, the extrinsic parameters of the global camera and the two telephoto observation cameras 3 at the initial zero position are solved respectively; at the same time, the internal relationships of each station are calibrated: namely, the fixed extrinsic parameters of the angle measuring camera 10 relative to the fixed support 9, the physical dimensions of the checkerboard marker block, and the installation angle deviation between the telephoto observation camera 10 and the pan-tilt rotation mechanism 4; this calibration process only needs to be performed once, or periodically; At the start of the measurement, the operator clicks on the top point of a pier to be measured in the panoramic view of the bridge captured by the global camera. The global map guidance control module calculates the coarse ray direction of the point to be measured in the unified measurement coordinate system according to the calibration parameters of the global camera. Combined with the coarse three-dimensional model of the bridge, it estimates the three-dimensional spatial range of the point. Then, it calculates the coarse pointing angle for the two digital pan-tilt stations respectively and drives their respective low-precision pan-tilt 5 to rotate. After the gimbal rotates to the correct position, the angle measuring camera 10 at each station immediately captures an image of the checkerboard marker block; the PnP algorithm is used to solve the precise pose of the checkerboard relative to the angle measuring camera 10; based on the pre-calibrated rigidity relationship, the system deduces the actual optical center position and optical axis pointing of the telephoto observation camera 3, and this pointing accuracy is much higher than the encoder reading of the gimbal itself. Subsequently, two telephoto cameras 3 simultaneously acquire high-resolution images of the top of the bridge pier. In the images, the image coordinates of the point to be measured are accurately extracted using a sub-pixel corner extraction algorithm. The intersection solution module constructs two spatial lines of sight based on the real-time extrinsic parameters and accurate image coordinates of the two telephoto cameras 3, and calculates their intersection using the least squares method, thus obtaining the three-dimensional coordinates of the point at the top of the bridge pier. The system automatically records these coordinates and compares them with historical data to calculate the displacement at the millimeter level. By repeating the above process, deformation monitoring of dozens of measuring points on the entire bridge can be carried out automatically and efficiently.

[0086] In this embodiment, the low-precision gimbal 5 only achieves "coarse pointing". Its error of tens of arcminutes is effectively compensated by the visual angle measurement module. The final intersection measurement accuracy is entirely determined by the visual angle measurement accuracy and the resolution of the telephoto observation camera 3, thus achieving the technical effect of completing high-precision measurement tasks with low-cost hardware.

[0087] Example 2: Joint angle measurement using multiple angle measuring cameras.

[0088] Two or three close-range angle measurement cameras 10 are deployed on each gimbal mounting bracket 9 to observe the same cooperative marker block 8 from different perspectives, or to observe multiple cooperative marker blocks 8 separately; the attitude of the gimbal rotation mechanism 4 is solved by joint optimization of multiple cameras to improve the reliability of angle measurement under occlusion conditions. To address the issue of a single angle measuring camera 10 failing due to complex lighting or occlusion, this embodiment is an enhancement of embodiment 1. Specifically, two or three close-range angle measuring cameras 10 are deployed on the fixed bracket 9 of each digital pan-tilt camera station 2. These cameras simultaneously observe the same cooperative marker block 8 (e.g., a three-dimensional corner prism marker block) from different perspectives, or separately observe multiple planar marker blocks installed at different positions on the pan-tilt rotation mechanism 4.

[0089] After the gimbal rotates, all angle measurement cameras 10 simultaneously acquire images. Due to the existence of multiple observation angles, even if a part of the features of a certain marker block is obscured by the gimbal structure itself or other objects, the system can still use clear images from other cameras to calculate the pose. Through bundle adjustment optimized by multiple cameras, the robustness and accuracy of solving the pose of cooperative marker blocks can be significantly improved. Experimental data show that the attitude angle measurement accuracy using dual-camera observation can be improved by about 40% compared to single-camera observation, especially when the gimbal pitch angle is large, effectively overcoming the problems of limited field of view and measurement ambiguity of single cameras. This embodiment enhances the adaptability and reliability of the system in harsh environments.

[0090] Example 3: Global guidance combined with the three-dimensional structural model.

[0091] When the object to be measured has a known 3D model or BIM model, the selected points in the global camera image are matched with the surface of the 3D model to obtain the rough 3D position of the point to be measured, and then the 3D position is converted into the coarse pointing angle of each pan-tilt station. This embodiment is applied to measurement objects with known 3D models or BIM models, such as large industrial equipment, dams, or ship hulls. During the calibration phase, the extrinsic parameters of the global camera are precisely registered to the CAD coordinate system where the model is located. When the operator clicks on a point to be measured (e.g., the center of a bolt hole) on the global camera image, the system does not simply convert the point into a spatial ray, but instead performs an intersection calculation between the ray and the known CAD model surface to directly obtain the approximate 3D coordinates (X, Y, Z) of the point to be measured in the CAD coordinate system. Subsequently, based on this rough three-dimensional coordinate and the position of each digital gimbal station, the system accurately calculates the theoretical pointing angle required for each telephoto observation camera 1, which serves as a coarse pointing control command. This method completely eliminates the reliance on real-time spatial forward intersection of ground control points, and is especially suitable for scenarios where the surface of the target lacks natural texture or artificial markings. Through model guidance, the gimbal can point to the target in one step, further improving measurement efficiency and automation.

[0092] Example 4: Alternative Example.

[0093] Alternative Implementation Example 1: Long-distance deformation measurement of fixed infrastructure; a global camera is installed at a high position outside the measurement area to cover the entire area of ​​large structures such as bridges and slopes; two or more digital pan-tilt telephoto observation cameras 1 are deployed at different baseline positions; the operator selects the measurement points to be measured in the global image in sequence, and the system automatically guides multiple telephoto observation cameras to cover the measurement points and complete spatial intersection to obtain the three-dimensional displacement of multiple measurement points.

[0094] Alternative embodiment 2: Multiple angle measuring cameras 10 jointly measure angles; two or three close-range angle measuring cameras 10 are arranged on each gimbal fixing bracket 9 to observe the same cooperative marker block 8 from different perspectives, or to observe multiple cooperative marker blocks 8 respectively; by jointly optimizing the attitude of the gimbal rotation mechanism 4 through multiple cameras, the reliability of angle measurement under occlusion conditions is improved.

[0095] Alternative Implementation Example 3: Combining global guidance from the structural 3D model, when the object to be measured has a known 3D model or BIM model, the selected points in the global camera image are matched with the surface of the 3D model to obtain the rough 3D position of the point to be measured, and then the 3D position is converted into the coarse pointing angle of each pan-tilt station.

[0096] Alternative Example 4: Automatic scanning and batch measurement of measurement points. The system generates a sequence of measurement points on the global map in advance or detects multiple candidate targets. It completes gimbal guidance, local imaging and intersection calculation in sequence according to the priority of measurement points to realize large-scale multi-point automated measurement.

[0097] Definitions of terms and symbols:

[0098]

[0099] Creative explanation of the overall technical solution:

[0100] Key innovations of this invention:

[0101] (1) A composite measurement architecture of “global map coarse guidance + digital pan-tilt local high-resolution imaging + multi-station intersection” is proposed, which organically combines large-scale target search with precise intersection measurement.

[0102] (2) A digital gimbal angle measurement method is proposed, in which a marker block is fixedly connected to the low-precision gimbal rotation component, and the marker block is observed by a close-range camera on the gimbal bracket to accurately restore the actual rotation angle of the gimbal.

[0103] (3) A measurement mechanism is proposed to decouple the low-precision gimbal control angle from the actual pointing angle of the telephoto camera. The low-precision gimbal only ensures that the target area enters the local field of view, while the precision line of sight is determined by the visual angle measurement result.

[0104] (4) A unified calibration and coordinate transfer method is proposed among global cameras, multiple digital pan-tilt stations, local telephoto cameras, angle measurement cameras, and cooperative marker blocks.

[0105] (5) A large-scale, high-precision measurement method is proposed, which guides two or more digital pan-tilt cameras to point to the same target area based on global image point selection, and uses the actual attitude of each station to construct the spatial line of sight for intersection calculation.

[0106] The "Global Map-Guided Digital Gimbal Camera Intersection Measurement Method and Apparatus" proposed in this invention demonstrates a high degree of inventiveness in its overall technical solution in the following aspects:

[0107] The uniqueness of the technical problem identification:

[0108] Existing technologies face a core contradiction: high-precision measurement requires the use of high-resolution telephoto lenses, but telephoto lenses have extremely narrow fields of view, making it extremely difficult to quickly and accurately search for and lock onto targets in large-scale scenes. Conventional solutions either rely on expensive high-precision servo gimbals or sacrifice measurement efficiency for blind searching. This invention accurately identifies that the essence of the problem lies not in the mechanical precision of the gimbal itself, but in the deep coupling between the two aspects of "coarse pointing" and "fine measurement." Decoupling these two aspects reveals a key path to a low-cost solution for high-precision, large-scale measurement.

[0109] The "non-obviousness" of technical solutions

[0110] The "global coarse guidance + local visual fine angle measurement + multi-station intersection" architecture proposed in this invention is not a simple superposition of existing technologies. Its non-obviousness is mainly reflected in:

[0111] Functional Module Reorganization and Coordination: In existing technologies, the gimbal's role is "pointing + angle measurement," and its accuracy determines the system's final accuracy. This invention breaks this mindset, reducing the gimbal's function to "only responsible for coarse coverage," while separating the "high-precision angle measurement" function and assigning it to a newly constructed "visual angle measurement unit." This unit consists of "cooperative marker blocks fixed on the gimbal's rotating components" and "close-range cameras fixed on the support." This functional reorganization requires cross-disciplinary insight, cleverly applying close-range visual positioning technology from photogrammetry to online compensation for gimbal rotation errors, rather than simply assembling components.

[0112] Reconstruction of the Error Propagation Chain: In traditional methods, gimbal mechanical errors (hysteresis, coding errors, assembly errors) directly contribute to the final measurement result. This invention constructs a novel error propagation chain: global detection (low precision) → gimbal coarse movement (low precision) → visual angle measurement (high precision) → telephoto imaging (high precision) → intersection calculation (high precision). Errors in the low-precision stage are effectively isolated and compensated by the subsequent high-precision visual measurement stage. This approach of error decoupling and reconstruction is not a standard solution for the conventional problem of "large-scale high-precision measurement" for those skilled in the art, and therefore possesses significant substantive characteristics.

[0113] A creative application of cooperative marker blocks: By fixing cooperative marker blocks to the rotating components of the pan-tilt unit and observing them with a close-up camera on the same base, this seemingly simple design is actually ingenious. It creates a local, stable, and high-precision angle measurement reference, which is known relative to the pan-tilt unit base, thus achieving an angular resolution far exceeding that of the pan-tilt unit encoder at extremely low cost. Compared to adding expensive photoelectric encoders to each pan-tilt unit, this solution offers significant cost advantages and flexible deployment, solving the economic challenges of multi-station deployment.

[0114] Summary of advantages compared to existing technologies:

[0115]

[0116] In summary, this invention, through its original system architecture of "functional decoupling and visual compensation," cleverly solves the industry challenge of large-scale, long-distance, and high-precision 3D measurement in a non-obvious way, and has resulted in significant technological advancements.

[0117] The above description is only a preferred embodiment of the present invention. All equivalent changes and modifications made within the scope of the claims of the present invention should be covered by the claims of the present invention.

Claims

1. A global map-guided digital pan-tilt camera intersection measurement method, characterized in that, Includes the following steps: (1) Use a global camera to acquire images of a large area and determine the points or areas to be measured; (2) Generate coarse pointing control quantities for each digital pan-tilt camera station based on the pre-calibrated coordinate relationship between the global camera and at least two digital pan-tilt camera stations. The pre-calibrated content in step (2) includes at least: global camera intrinsic and extrinsic parameters, initial external orientation parameters of each telephoto camera, intrinsic and extrinsic parameters of each corner measurement camera, geometric parameters of cooperative marker blocks, installation relationship between telephoto cameras and corner measurement cameras, and baseline relationship of each digital pan-tilt camera station under a unified measurement coordinate system. The following four transformation matrices can be obtained through calibration: ； In the formula, This is the extrinsic parameter transformation matrix from the global camera coordinate system to the unified measurement coordinate system. For the first When the coordinate system of the telephoto camera at each station is initially at zero position, relative to the initial calibration extrinsic parameter matrix of the unified measurement coordinate system, The extrinsic parameter transformation matrix of the cooperative marker block coordinate system relative to the angle measurement camera coordinate system. The extrinsic transformation matrix is ​​used between the angle measurement camera coordinate system and the telephoto camera coordinate system; the global camera coordinate system is... A unified measurement coordinate system is , No. The coordinate system of the telephoto local camera at each digital gimbal camera station is: The coordinate system of the angle measurement camera is The coordinate system of the cooperation marker block is ; The digital pan-tilt camera station described in step (2) includes: a low-precision pan-tilt unit, a local field-of-view telephoto camera mounted on the pan-tilt unit, a cooperative marker block fixedly connected to the pan-tilt unit's rotating components, and one or more angle measuring cameras mounted on the pan-tilt unit's fixed bracket. The coarse pointing control quantity in step (2) is calculated based on the pixel coordinates of the point to be measured in the global image, the global camera exterior orientation parameters, and the three-dimensional prior information of the survey area or control point constraints. The coarse pointing control quantity is only used to make the local field of view of the telephoto camera cover the target area, and is not used as the precise angle observation value in the intersection solution. (3) According to the coarse pointing control amount, control the low-precision gimbal of each digital gimbal camera station to turn so that the local field of view telephoto camera installed on the gimbal covers the point to be measured or the area to be measured. (4) Use an angle measuring camera mounted on the gimbal fixed bracket to observe the cooperative marker block fixed to the gimbal rotating component, and determine the actual rotation angle or actual attitude of the gimbal rotating component based on the observed image. (5) Based on the actual rotation angle or actual attitude, and in conjunction with the pre-calibrated rigid transformation relationship, recover the real-time exterior orientation parameters of the telephoto camera; (6) At least two telephoto cameras with recovered real-time exterior orientation parameters acquire local images containing the point to be measured, extract the pixel coordinates of the point to be measured, construct the corresponding spatial lines of sight, and perform spatial intersection calculation to obtain the three-dimensional coordinates of the point to be measured.

2. The global map-guided digital pan-tilt camera intersection measurement method according to claim 1, characterized in that, The method for obtaining the actual rotation angle or actual attitude of the gimbal rotation component in step (4) is as follows: perform PnP solution, homography decomposition, and multi-marker feature fusion on the cooperative marker block image captured by the rotation angle measurement camera, or use the observation data of multiple rotation angle measurement cameras for joint optimization solution; During measurement, the angle measuring camera calculates its real-time pose relative to the angle measuring camera based on the cooperative marker block image, and recovers the real-time pose of the telephoto camera from a rigid relationship; the relationship is expressed as: ； In the formula, At time t, the extrinsic transformation matrix from the coordinate system of the angle measurement camera to the coordinate system of the telephoto camera. At time t, the angle measurement camera captures and identifies the real-time pose matrix of the cooperative marker block, which is then calculated. This is the extrinsic transformation matrix between the telephoto camera and the cooperative marker block; By combining the pre-calibrated transformation relationship between the angle measurement camera and the global camera, and the transformation relationship between the global camera and the unified measurement coordinate system, the real-time pose of the telephoto camera is recovered: ； In the formula, At time t, the first The extrinsic transformation matrix of a telephoto camera relative to a unified measurement coordinate system. This is the extrinsic transformation matrix of the global camera coordinate system relative to the unified measurement coordinate system. For the first The extrinsic transformation matrix of the angle measurement camera coordinate system relative to the global camera coordinate system. For time t, the first The extrinsic transformation matrix of the telephoto camera coordinate system relative to the angle measurement camera coordinate system. This is the rotation matrix of the telephoto camera in a unified measurement coordinate system. This is the translation vector of the telephoto camera in a unified measurement coordinate system; The point to be measured is at the 1st The normalized image points in a telephoto camera image are represented as follows: ； In the formula, For the point to be measured at the th Normalized camera coordinates under a telephoto camera For the first The inverse matrix of the intrinsic parameter matrix of a telephoto camera. Let be the horizontal pixel coordinates of the point to be measured on the telephoto camera image. The vertical pixel coordinates of the point to be measured on the telephoto camera image; Combined with the telephoto camera extrinsic transformation matrix recovered in real time in step (5) Transform it to a unified measurement coordinate system Below, the spatial line of sight is represented as: ; In the formula, For depth parameters, To achieve the first under a unified measurement coordinate system, A three-dimensional unit direction vector pointing from a telephoto camera to the point to be measured. For the first The coordinates of the optical center of each camera in a unified measurement coordinate system. The unit direction vector is determined by the real-time pose of the telephoto camera and the normalized image points; for n≥2 lines of sight, the coordinates of the point to be measured are obtained by minimizing the sum of the squared perpendicular distances from each line of sight to the point to be measured. ； In the formula, I is a 3×3 identity matrix; When introducing error weights, a weight matrix is ​​established based on the image point extraction accuracy of the telephoto camera, the attitude estimation accuracy of the angle measurement camera, the baseline length of the station, the line-of-sight intersection angle, and the calibration residual. Different weights are assigned to the lines of sight of different stations to improve the robustness of the intersection solution.

3. The global map-guided digital pan-tilt camera intersection measurement method according to claim 1, characterized in that, The real-time external orientation parameters of the telephoto camera in step (5) are determined by the pose of the cooperative marker block obtained by the angle measurement camera, the rigidity relationship between the cooperative marker block and the gimbal rotation component, the rigidity relationship between the telephoto camera and the gimbal rotation mechanism, and the station external parameters.

4. The global map-guided digital pan-tilt camera intersection measurement method according to claim 1, characterized in that, The spatial intersection solution described in step 6 uses least squares, weighted least squares, or robust estimation methods; wherein, the weight of each spatial line of sight involved in the solution is determined based on at least one of the following factors: the image point extraction accuracy of the corresponding telephoto camera, the angle measurement accuracy, the line of sight intersection angle, the station baseline length, or the calibration residual.

5. The global map-guided digital pan-tilt camera intersection measurement method according to claim 1, characterized in that, The method further includes: repeatedly performing the steps of global map point selection, gimbal coarse pointing, visual measurement of actual gimbal rotation angle, telephoto local imaging, and multi-station intersection calculation, so as to achieve continuous measurement of multiple test points within the global camera coverage area.

6. The global map-guided digital pan-tilt camera intersection measurement method according to claim 1, characterized in that, By registering the images acquired by the global camera with the three-dimensional model of the object under test, the pixel coordinates of the selected test point in the global image are mapped to the surface of the three-dimensional model to obtain the coarse three-dimensional coordinates of the test point, and the coarse pointing control quantity is generated for each digital pan-tilt camera station based on the coarse three-dimensional coordinates.

7. A globally map-guided digital gimbal camera intersection measurement device for implementing the method of any one of claims 1 to 5, characterized in that, include: A global imaging module is used to acquire images of a large area. At least two digital gimbal camera stations, each station including a low-precision gimbal, a local field-of-view telephoto camera mounted on the gimbal, and a cooperative marker block fixed to the rotating parts of the gimbal; The rotation angle measurement module includes one or more close-range rotation angle measurement cameras mounted on a fixed bracket at each digital pan-tilt camera station, used to observe the cooperative marker block and output the actual rotation angle of the pan-tilt rotating component; The calibration parameter module is used to store pre-calibrated camera parameters and rigid transformation relationships between components; The guidance and control module is connected to the global imaging module, the calibration parameter module and the digital pan-tilt camera station, respectively. It is used to generate coarse pointing control commands based on the points to be measured in the global image and drive the low-precision pan-tilt unit to rotate. The local image measurement module is used to control the telephoto camera to acquire local images and extract the sub-pixel level image point coordinates of the points to be measured; The intersection solution module is connected to the calibration parameter module, the angle precision measurement module, and the local image measurement module, respectively. It is used to solve the three-dimensional coordinates of the point to be measured by spatial line intersection based on the real-time pointing parameters of at least two stations and the coordinates of local image points. The results management module is used to store, stitch together, compare and analyze, and output the three-dimensional coordinates, quality evaluation indicators, and multi-period measurement results of multiple test points.

8. The global map-guided digital pan-tilt camera intersection measurement device according to claim 7, characterized in that, The cooperative marker block is a planar checkerboard, a dot array, a coded marker, an AprilTag / ArUco type marker, a three-dimensional marker block, or a combination thereof; the angle measurement camera is one or more cameras fixedly mounted on a gimbal bracket or a component that maintains a rigid relationship with the gimbal base, used to observe the same cooperative marker block from different perspectives or to observe multiple cooperative marker blocks separately.

9. The global map-guided digital pan-tilt camera intersection measurement device according to claim 7, characterized in that, The intersection solution module is also used to calculate and output the intersection residual, line-of-sight angle, coordinate covariance, or measurement reliability as measurement quality evaluation indicators; the result management module is used to store, stitch together, compare and analyze, and output the three-dimensional coordinates, quality evaluation indicators, and multi-period measurement results of multiple test points.

Images