A fixed camera and survey point joint optimization layout method and system

Abstract

The application provides a fixed camera and survey point joint optimization layout method and system, and belongs to the technical field of engineering surveying and mapping. The method comprises the following steps: generating a candidate fixed camera point set and a candidate survey point set; establishing a visibility coupling relationship between the candidate fixed camera points and the candidate survey points, and generating coupling constraints based on the visibility coupling relationship; calculating the prediction error of the layout stage based on the visibility coupling relationship and the spatial geometric relationship of the candidate fixed camera points, and jointly solving the layout scheme of the fixed camera and the survey point by taking the coupling constraints as the constraint conditions and the prediction error as the evaluation index; deploying the fixed camera and the survey point according to the layout scheme, completing camera calibration based on the control points, independently evaluating the external accuracy based on the check points, and performing periodic pose drift checking based on the long-term stable points. The application solves the problem that the existing technology lacks joint configuration optimization between the fixed camera point layout and the survey point layout, which leads to the problem that the layout scheme is difficult to balance the reconstruction accuracy and the reliability of external verification.

Description

Technical Field

[0001] This invention relates to the field of engineering surveying technology, and in particular to a method and system for optimizing the joint deployment of fixed cameras and survey points. Background Technology

[0002] In the field of engineering surveying and 3D reconstruction, with the continuous advancement of infrastructure construction and underground space development, the demand for acquiring 3D geometric information in scenarios such as foundation pit engineering, road construction, and yard management is increasing. These scenarios are typically characterized by long operation cycles, complex environmental conditions, and high frequency of repeated measurements. Traditional methods such as UAV oblique photography and manual mobile photography are subject to many limitations in terms of airspace management, personnel availability, and safety conditions, making it difficult to meet the needs of long-term, fixed-point, and high-frequency monitoring. Fixed cameras, due to their advantages of long-term deployment, fixed-point data acquisition, and repeated shooting from the same perspective, are gradually becoming an important technical means for 3D reconstruction of engineering sites.

[0003] Establishing engineering coordinate benchmarks using fixed cameras and on-site control points can reduce the reliance on reorientation in each reconstruction phase and improve the geometric consistency between multiple phases of results. However, in fixed-camera mapping systems, the layout of camera installation points and the setting of control and checkpoints are key factors affecting the quality of the final results. In existing technologies, camera layout optimization, control point layout optimization, and 3D reconstruction algorithms often focus on a single objective, such as increasing image coverage, optimizing flight path planning, or increasing the number of control points. However, in practical engineering applications with long-term fixed camera deployment, optimizing camera locations or control point distribution alone often fails to meet the comprehensive needs of multiple aspects, including 3D reconstruction accuracy verification, long-term pose stability verification, and engineering measurement accuracy requirements. Especially in complex scenarios such as foundation pit excavation with large-angle oblique views, low-texture ground, and occluded areas, even increasing the number of image acquisitions or adjusting control point positions may still result in insufficient model connectivity and large external verification errors, making it difficult for fixed-camera mapping results to support engineering measurement conclusions with centimeter-level accuracy. Summary of the Invention

[0004] Based on this, the purpose of the present invention is to provide a method and system for joint optimization of fixed camera and measuring point deployment, which aims to solve the problem that the lack of joint configuration optimization between fixed camera locations and measuring point deployment in the prior art makes it difficult for the deployment scheme to balance reconstruction accuracy and external verification reliability.

[0005] According to an embodiment of the present invention, a method for joint optimization of fixed camera and measurement point deployment includes: Generate a set of candidate fixed camera points and a set of candidate measurement points. The candidate measurement points can be selected as control points, check points, or long-term stable points. Establish the visibility coupling relationship between the candidate fixed camera point and the candidate measurement point, and generate coupling constraints based on the visibility coupling relationship. The coupling constraints include control point constraints, check point constraints, and long-term stable point constraints. Based on the visibility coupling relationship and the spatial geometric relationship of the candidate fixed camera points, the prediction error in the deployment stage is calculated. Using the coupling constraint as the constraint condition and the prediction error as the evaluation index, the deployment scheme of fixed camera installation points, control points, check points and long-term stable points is jointly solved. Deploy fixed cameras and measurement points according to the deployment scheme, complete camera calibration based on control points, independently evaluate external accuracy based on check points, and perform periodic pose drift verification based on long-term stable points.

[0006] In addition, the fixed camera and measuring point joint optimization deployment method according to the above embodiments of the present invention may also have the following additional technical features: Furthermore, the steps for generating the candidate fixed camera point set and the candidate measurement point set include: The spatial information of the engineering area to be measured is obtained, the engineering area to be measured is discretized into a set of spatial points to be measured, and divided into key areas and ordinary areas according to the importance of measurement, and different weight coefficients are assigned to the spatial points to be measured in different areas. A set of candidate fixed camera points is generated based on the on-site installation conditions. The candidate fixed camera points include installation coordinates, installation height, orientation angle, elevation angle, field of view, stability level, and construction accessibility information. A set of candidate measurement points is generated based on the available deployment conditions on site. The candidate measurement points contain attribute information that can be selected as control points, checkpoints, or long-term stable points, and the selection status of each candidate point is characterized by binary decision variables.

[0007] Furthermore, the step of establishing the visibility coupling relationship between the candidate fixed camera point and the candidate measurement point, and generating coupling constraints based on the visibility coupling relationship, includes: Based on the installation coordinates, attitude parameters, field of view parameters of the candidate fixed camera points and the spatial coordinates of the spatial point to be measured, the line of sight occlusion state, effective field of view state and observation quality state are calculated, and the visibility coupling relationship between the candidate fixed camera points and the candidate measurement points is determined. Based on the visibility coupling relationship, the control point constraint, the checkpoint constraint, and the long-term stable point constraint are generated respectively. Based on the visibility state, the intersection angle between any two selected fixed cameras and the same spatial point to be measured is calculated. The multi-view observation geometric quality of each spatial point to be measured is evaluated based on the intersection angle, and the intersection angle is within a preset effective angle range. The control point constraints include control point visibility quantity constraints, control point image distribution constraints, and control point spatial distribution constraints; the checkpoint constraints include mutual exclusion constraints between checkpoints and control points and checkpoint representativeness constraints; and the long-term stable point constraints include long-term stable point visibility quantity constraints and long-term stable point stability score constraints.

[0008] Furthermore, based on the visibility coupling relationship and the spatial geometric relationship of the candidate fixed camera points, the steps for calculating the prediction error during the deployment phase include: Based on the visibility coupling relationship, the effective baseline length of the spatial point to be measured is determined according to the spatial distance between any two selected fixed camera points that simultaneously observe the same spatial point to be measured. Based on the effective baseline length, the depth scale of the spatial point to be measured, the camera focal length, and the pixel positioning error, the 3D reconstruction prediction error of each spatial point to be measured is calculated. The 3D reconstruction prediction error increases as the effective baseline length decreases. Based on the 3D reconstruction prediction error and the elevation estimation error at adjacent times, the depth difference prediction error of each spatial point to be measured is calculated. Based on the depth difference prediction error and the preset grid area, the volume prediction error of each construction zone is calculated to generate the engineering measurement prediction error.

[0009] Furthermore, the steps for jointly solving the layout scheme of fixed camera mounting points, control points, checkpoints, and long-term stable points, using the coupling constraints as constraints and the prediction error as evaluation index, include: A joint optimization objective function is constructed, wherein the coverage quality of the survey area, the geometric quality of multi-view observation, and the satisfaction of coupling constraints are used as positive optimization terms, and the prediction error and deployment cost of the deployment stage are used as negative penalty terms. The terms are weighted and integrated by preset weight coefficients. Using the coupling constraint as a hard constraint and the prediction error meeting the preset accuracy requirement as the evaluation criterion, the feasibility of the binary decision variable combination of candidate fixed camera points and candidate measurement points is screened. Before performing the combined search, infeasible candidate fixed camera points are eliminated based on preset feasibility conditions. Then, an optimization algorithm is used to search and solve the combination of the remaining candidate fixed camera points and all candidate measurement points. Under the premise of satisfying all the coupling constraints, the joint optimization objective function achieves the optimal combination of binary decision variable values, and outputs the layout scheme of fixed camera installation points, control points, check points and long-term stable points.

[0010] Furthermore, the steps of deploying fixed cameras and measurement points according to the deployment scheme, completing camera calibration based on control points, independently evaluating external accuracy based on check points, and performing periodic pose drift verification based on long-term stable points include: Install fixed cameras according to the fixed camera installation points in the deployment scheme, deploy corresponding targets according to the control point set, check point set and long-term stable point set, and measure the three-dimensional spatial coordinates of each target. Based on the three-dimensional spatial coordinates of the control points and their pixel coordinates in the calibration image, a projection relationship is established. The extrinsic parameters of each fixed camera are solved by minimizing the reprojection error, and the camera calibration and engineering coordinate registration are completed. Using a calibrated fixed camera to acquire images, the external accuracy is independently evaluated based on the deviation between the 3D spatial coordinates of the checkpoints and the 3D coordinates obtained from multi-view reconstruction. During each data acquisition cycle, the reprojection error is calculated using the image coordinates of long-term stable points. When the reprojection error exceeds a preset threshold, it is determined that the pose of the corresponding fixed camera has shifted and recalibration is triggered.

[0011] Furthermore, after completing the camera calibration and engineering coordinate registration steps, the following steps are also included: Periodic image acquisition is performed using a calibrated fixed camera. Under known camera pose conditions, the three-dimensional information of the area to be measured is reconstructed, an elevation map and a reconstructed confidence map are generated, and areas with confidence scores below a preset threshold are marked as invalid. Differential calculations are performed based on elevation maps of adjacent periods to extract engineering parameters within the effective area. When the proportion of invalid areas in a construction zone exceeds a preset threshold, the engineering parameters of that zone are marked as low confidence.

[0012] Another objective of this invention is to provide a fixed camera and measuring point joint optimization deployment system for implementing the aforementioned fixed camera and measuring point joint optimization deployment method, the system comprising: The point set determination module is used to generate a candidate fixed camera point set and a candidate measurement point set. The candidate measurement points can be selected as control points, check points, or long-term stable points. The coupling constraint determination module is used to establish the visibility coupling relationship between the candidate fixed camera point and the candidate measurement point, and generate coupling constraints based on the visibility coupling relationship. The coupling constraints include control point constraints, check point constraints and long-term stable point constraints. The deployment scheme determination module is used to calculate the prediction error in the deployment stage based on the visibility coupling relationship and the spatial geometric relationship of the candidate fixed camera points, and to jointly solve the deployment scheme of fixed camera installation points, control points, check points and long-term stable points using the coupling constraint as the constraint condition and the prediction error as the evaluation index. The verification and evaluation module is used to deploy fixed cameras and measurement points according to the deployment plan, complete camera calibration based on control points, independently evaluate external accuracy based on check points, and perform periodic pose drift verification based on long-term stable points.

[0013] Another objective of this invention is to provide a storage medium storing a computer program that, when executed by a processor, implements the steps of the aforementioned method for joint optimization of fixed camera and measurement point deployment.

[0014] Another objective of this invention is to provide an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the above-described method for joint optimization of fixed camera and measurement point deployment.

[0015] This invention generates a set of candidate fixed camera points and a set of candidate measurement points, and establishes a visibility coupling relationship between them. Based on this visibility coupling relationship, it generates coupling constraints, including control point constraints, checkpoint constraints, and long-term stable point constraints. This allows the selection of camera installation points, control points, checkpoints, and long-term stable points to mutually constrain and collaboratively optimize within a unified framework, avoiding the local optima problem caused by independently optimizing camera point locations and control point distributions in traditional methods, and improving the overall coordination of the deployment scheme. Simultaneously, based on the visibility coupling relationship and the spatial geometric relationship of the candidate fixed camera points, the prediction error during the deployment phase is calculated. Using coupling constraints as constraints and prediction error as the evaluation index, the deployment scheme is jointly solved, allowing the 3D reconstruction accuracy to be estimated during the deployment phase. This eliminates the need for accuracy verification and scheme adjustments after on-site deployment and image acquisition, reducing trial-and-error costs. Furthermore, excluding checkpoints from extrinsic parameter solving and independently evaluating external accuracy, as well as utilizing long-term stable points to perform periodic pose drift verification, enables continuous monitoring and timely correction of measurement results while ensuring camera calibration accuracy. This effectively improves the reliability and stability of fixed-camera mapping systems under long-term deployment conditions. Therefore, this invention solves the problem in existing technologies where the lack of joint configuration optimization between fixed camera locations and measurement point layout makes it difficult to balance reconstruction accuracy and external verification reliability in the deployment scheme. Attached Figure Description

[0016] Figure 1 This is a flowchart of a method for optimizing the joint deployment of a fixed camera and measuring points according to the first embodiment of the present invention; Figure 2 This is a schematic diagram of the fixed camera and measuring point joint optimization deployment system in the second embodiment of the present invention; Figure 3 This is a schematic diagram of the structure of the electronic device in the third embodiment of the present invention; Figure 4 This is a schematic diagram of the optimized layout of fixed cameras and measuring points in the foundation pit project according to the first embodiment of the present invention; The following detailed description, in conjunction with the accompanying drawings, will further illustrate the present invention. Detailed Implementation

[0017] To facilitate understanding of the present invention, a more complete description will be given below with reference to the accompanying drawings. Several embodiments of the invention are illustrated in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete.

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0019] Example 1 Please see Figure 1 The figure shows a method for optimizing the deployment of a fixed camera and measuring points in the first embodiment of the present invention. The method specifically includes steps S01-S04.

[0020] S01, Generate a set of candidate fixed camera points and a set of candidate measurement points. The candidate measurement points can be selected as control points, check points or long-term stable points.

[0021] Specifically, the spatial information of the project area to be measured is acquired, and the area is discretized into a set of spatial points to be measured. These points are then divided into key areas and ordinary areas based on measurement importance, with different weighting coefficients assigned to the spatial points in different areas. A set of candidate fixed camera points is generated based on on-site installation conditions. These candidate fixed camera points include installation coordinates, installation height, orientation angle, elevation angle, field of view, stability level, and construction accessibility information. A set of candidate measurement points is also generated based on on-site deployment conditions. These candidate measurement points contain attribute information indicating whether they can be selected as control points, checkpoints, or long-term stable points, and the selection status of each candidate point is represented by a binary decision variable. In practical implementation, by acquiring and discretizing the spatial information of the project area to be measured, continuous actual site data can be transformed into discrete data that can be processed by a computer. Furthermore, by distinguishing between key and ordinary areas and assigning different weighting coefficients, critical areas such as the bottom of the foundation pit, the toe of the slope, the top of the slope, and the excavation boundary receive higher attention during optimization, thereby ensuring that the reconstruction accuracy of these areas prioritizes meeting project requirements. Furthermore, the generation of the candidate fixed camera point set fully considers the actual installable locations and equipment parameters of the site, such as enclosures, capping beams, supports, towers, and fixed walls. The generation of the candidate measuring point set considers deployable locations such as capping beams, the top of the retaining structure, the hardened ground outside the pit, and dedicated target supports, assigning them attributes that allow them to be selected as control points, checkpoints, or long-term stable points. This enables the subsequent optimization model to be solved under realistic physical constraints, avoiding the problem of theoretically optimal solutions failing due to on-site impracticality. The definition of binary decision variables transforms the selection of each candidate point into a mathematical variable of 0 or 1, allowing the originally discrete combinational selection problem to be expressed in a unified mathematical form.

[0022] S02, establish the visibility coupling relationship between the candidate fixed camera point and the candidate measurement point, and generate coupling constraints based on the visibility coupling relationship. The coupling constraints include control point constraints, check point constraints and long-term stable point constraints.

[0023] In practical implementation, by establishing a visibility coupling relationship between candidate fixed camera points and candidate measurement points, and unifying the constraints of control points, check points, and long-term stable points into the same coupling constraint system, the deployment of four different functional points can mutually constrain and synergistically optimize within a unified mathematical framework. The establishment of the visibility coupling relationship quantifies the core geometric relationship of which cameras can see which measurement points. Check point constraints exclude check points from extrinsic parameter solutions, ensuring the independence of check point accuracy verification and avoiding overfitting caused by verification data and solution data originating from the same source. Long-term stable point constraints ensure that the deployment scheme has continuous self-checking capabilities over time, enabling timely detection and correction of pose drift.

[0024] Specifically, based on the installation coordinates, attitude parameters, and field of view parameters of the candidate fixed camera points and the spatial coordinates of the spatial point to be measured, the line-of-sight occlusion state, effective field of view state, and observation quality state are calculated to determine the visibility coupling relationship between the candidate fixed camera points and the candidate measurement points. Based on the visibility coupling relationship, the control point constraints, check point constraints, and long-term stable point constraints are generated respectively. Based on the visibility state, the intersection angle between any two selected fixed cameras and the same spatial point to be measured is calculated to evaluate the multi-view observation geometric quality of each spatial point to be measured based on the intersection angle, which is within a preset effective angle range. Among them, the control point constraints include control point visibility quantity constraints, control point image distribution constraints, and control point spatial distribution constraints; the check point constraints include mutual exclusion constraints between check points and control points and check point representativeness constraints; the long-term stable point constraints include long-term stable point visibility quantity constraints and long-term stable point stability score constraints.

[0025] In practical implementation, the calculation of line-of-sight occlusion can automatically exclude observation pairs obstructed by enclosures, supporting structures, construction machinery, or pit walls, effectively avoiding the problem of insufficient actual observation quantity due to occlusion in the deployment scheme. The determination of effective field of view status utilizes camera orientation angle, pitch angle, and field of view range parameters to ensure that the spatial points to be measured truly fall within the camera's field of view. The dual determination of observation distance and incident angle in the observation quality status ensures that effective observations not only have visibility but also sufficient spatial resolution and observation geometry, avoiding the misidentification of distant low-resolution or large-tilt-angle observations as effective coverage. Control point visibility quantity constraints ensure sufficient observation redundancy between each camera and control point; control point image distribution constraints and spatial distribution constraints avoid the problem of increased calibration extrapolation errors caused by control points being concentrated in local image or spatial regions; mutual exclusion constraints between checkpoints and control points ensure the independence of checkpoints, enabling them to objectively reflect the true reconstruction error of areas not involved in calibration; long-term stable point visibility quantity constraints and stability scoring constraints ensure the continuous effectiveness of long-term pose verification. The introduction of the intersection angle calculation and evaluation mechanism enables the optimization process to quantitatively evaluate the multi-view observation geometry of each spatial point to be measured. The setting of the effective angle range avoids the problem of insufficient depth accuracy due to too small an intersection angle or difficulty in image matching due to too large an intersection angle, thereby ensuring the depth accuracy of 3D reconstruction.

[0026] S03. Based on the visibility coupling relationship and the spatial geometric relationship of the candidate fixed camera points, calculate the prediction error in the deployment stage, and use the coupling constraint as the constraint condition and the prediction error as the evaluation index to jointly solve the deployment scheme of fixed camera installation points, control points, check points and long-term stable points.

[0027] In practical implementation, by calculating the 3D reconstruction prediction error during the deployment phase, a direct mathematical correlation is established between the accuracy prediction of the deployment scheme and the actual 3D reconstruction accuracy. This eliminates the need for a trial-and-error process involving deployment, data acquisition, verification, and adjustment to obtain a deployment scheme with guaranteed accuracy. Using prediction error as an evaluation metric, rather than relying solely on intermediate geometric metrics such as coverage and intersection angles, directly aligns the optimization objective with the final mapping requirements, avoiding the problem of meeting intermediate metrics but ultimately failing to achieve sufficient accuracy. Coupling constraints as hard constraints ensure the basic feasibility of the deployment scheme, while using prediction error as an evaluation metric guarantees that the scheme meets accuracy standards.

[0028] Specifically, based on the visibility coupling relationship, the effective baseline length of the measured spatial point is determined according to the spatial distance between any two selected fixed camera points simultaneously observing the same measured spatial point. Based on the effective baseline length, the depth scale of the measured spatial point, the camera focal length, and the pixel positioning error, the 3D reconstruction prediction error of each measured spatial point is calculated, and the 3D reconstruction prediction error increases as the effective baseline length decreases. Based on the 3D reconstruction prediction error and the elevation estimation error at adjacent times, the depth difference prediction error of each measured spatial point is calculated, and based on the depth difference prediction error and the preset grid area, the volume prediction error of each construction zone is calculated to generate the engineering measurement prediction error.

[0029] In practical implementation, the determination of the effective baseline length is directly based on the visibility coupling relationship, ensuring that baseline calculations are performed only on actually effective observation pairs. This avoids the error underestimation problem caused by including invisible or substandard camera combinations in baseline statistics. The constraint that the 3D reconstruction prediction error increases with the shortening of the effective baseline length effectively penalizes short baseline observation schemes, thus proactively avoiding insufficient depth accuracy caused by observations from distant, same-side cameras during the optimization process. Depth difference prediction error allows the deployment scheme to predict its performance in temporal difference applications, while zonal volume prediction error further links the deployment scheme directly to key measurement tasks in engineering scenarios such as foundation pit excavation, ensuring that the optimization results meet the accuracy requirements of actual engineering projects such as earthwork volume statistics.

[0030] Furthermore, a joint optimization objective function is constructed. This function uses the survey area coverage quality, multi-view observation geometric quality, and coupling constraint satisfaction as positive optimization terms, and the prediction error and deployment cost during the deployment phase as negative penalty terms. These terms are weighted and integrated using preset weighting coefficients. The coupling constraints are used as hard constraints, and the prediction error meeting preset accuracy requirements is used as the evaluation criterion. Feasibility screening is performed on the combination of binary decision variables for candidate fixed camera points and candidate measurement points. Before performing the combination search, infeasible candidate fixed camera points are first eliminated based on preset feasibility conditions. Then, an optimization algorithm is used to search and solve for the combination of the remaining candidate fixed camera points and all candidate measurement points. Under the premise of satisfying all the coupling constraints, the joint optimization objective function achieves the optimal combination of binary decision variable values, outputting the deployment scheme for fixed camera installation points, control points, checkpoints, and long-term stable points.

[0031] In practical implementation, the joint optimization objective function uses the survey area coverage quality, multi-view observation geometric quality, and coupling constraint satisfaction as positive optimization terms, and prediction error and deployment cost as negative penalty terms. Weighted synthesis is performed using preset weight coefficients, making the trade-off between coverage integrity, reconstruction accuracy, and engineering cost clear and controllable. Users can adjust the weight coefficients according to specific engineering needs to obtain deployment schemes with different tendencies. Using coupling constraints as hard constraints ensures the basic feasibility of the deployment scheme, while using prediction error as the evaluation criterion guarantees the accuracy of the scheme. The strategy of first eliminating infeasible candidate points and then performing a fine search on the remaining combinations filters the candidate fixed camera points from the original set to a subset that meets the requirements of construction accessibility, occlusion conditions, and ground sampling distance thresholds. This significantly reduces the spatial complexity of subsequent combination searches and improves solution efficiency. Through the above mechanism, this step can obtain a joint deployment scheme for fixed cameras and survey points that balances reconstruction accuracy, verification reliability, and cost-effectiveness during the deployment phase.

[0032] S04. Deploy fixed cameras and measurement points according to the deployment scheme, complete camera calibration based on control points, independently evaluate external accuracy based on check points, and perform periodic pose drift verification based on long-term stable points.

[0033] Specifically, fixed cameras are installed according to the fixed camera mounting points in the deployment scheme. Corresponding targets are deployed according to the control point set, check point set, and long-term stable point set, and the three-dimensional spatial coordinates of each target are measured. Projection relationships are established based on the three-dimensional spatial coordinates of the control points and their pixel coordinates in the calibration image. The extrinsic parameters of each fixed camera are solved by minimizing the reprojection error, and the camera calibration and engineering coordinate registration are completed. Images are acquired using the calibrated fixed cameras, and the external accuracy is independently evaluated based on the deviation between the three-dimensional spatial coordinates of the check points and the three-dimensional coordinates obtained by multi-view reconstruction. In each data acquisition cycle, the reprojection error is calculated using the image coordinates of the long-term stable points. When the reprojection error exceeds a preset threshold, it is determined that the corresponding fixed camera has experienced pose drift and recalibration is triggered.

[0034] In practical implementation, the deployment plan is transformed into actual physical installation and coordinate measurement for concrete application. During the calibration phase, the camera extrinsic parameters are solved and engineering coordinates are registered by minimizing the reprojection error of control points, providing accurate camera pose information for subsequent 3D reconstruction. Checkpoints do not participate in the extrinsic parameter solution; the deviation between the 3D coordinates of checkpoints and the reconstructed coordinates from multiple views is used to independently evaluate external accuracy, effectively avoiding overfitting during calibration and ensuring that the accuracy evaluation results truly reflect the reconstruction error of areas not involved in calibration. Furthermore, during each data acquisition cycle, the reprojection error is calculated using the image coordinates of long-term stable points and compared with a preset threshold, enabling automatic detection of camera pose drift and triggering of recalibration, allowing the deployment plan to maintain its design accuracy under long-term deployment conditions.

[0035] Furthermore, based on the calibrated fixed camera, periodic image acquisition is performed. Under known camera pose conditions, the 3D information of the area to be measured is reconstructed, generating elevation maps and reconstruction confidence maps. Areas with confidence levels below a preset threshold are marked as invalid. Differential calculations are performed on elevation maps from adjacent periods to extract engineering parameters within the valid area. When the proportion of invalid areas in a construction zone exceeds a preset threshold, the engineering parameters for that zone are marked as low confidence. In practical implementation, periodic image acquisition and 3D reconstruction under known pose conditions avoid the cumbersome process of reorientation required for each reconstruction in traditional motion-based structure reconstruction methods, significantly improving the efficiency and consistency of multi-period monitoring results. Marking areas with confidence levels below the preset threshold as invalid allows low-texture areas, occluded areas, and unreliable reconstruction areas to be automatically excluded in subsequent calculations, preventing unreliable data from contaminating the engineering parameter extraction results. Differential calculations of elevation maps from adjacent periods directly reflect surface changes, providing quantitative basis for engineering needs such as foundation pit excavation calculations and settlement monitoring. When the proportion of invalid areas in a construction zone exceeds a preset threshold, the engineering parameters of that zone are marked as low confidence, effectively preventing erroneous conclusions based on insufficient data from misleading engineering decisions and ensuring the credibility of monitoring results.

[0036] The following describes a specific implementation of the present invention using a fixed-camera 3D reconstruction and rapid excavation estimation task at a foundation pit construction site. This embodiment illustrates how to determine the fixed camera installation point, control point, check point, and long-term stable point based on the survey area, candidate camera points, candidate measurement points, coverage, ground sampling distance (GSD), intersection angle, control point geometry, check point representativeness, and excavation estimation error. Furthermore, it completes fixed camera calibration, 3D height field reconstruction, digital elevation model (DEM) differencing, zonal earthwork volume calculation, and quality evaluation.

[0037] The preliminary verification of this embodiment is based on a fixed-camera parking lot large-angle measurement experiment. In this experiment, four fixed camera points were used in conjunction with eight central moving control points. Even after increasing the number of images to approximately 100, the sparse reconstruction by COLMAP software still resulted in four independent sub-models, failing to form a unified full-field model. This result indicates that under conditions of low-texture ground, repeating white lines, and large-angle oblique viewing, simply increasing the number of photos or the position of moving control points is insufficient to solve the problem of insufficient co-view between adjacent fixed points. Further employing 6 camera points and 8 registration points, local orthorectification results can be generated. However, in the selected five-control-point scheme, the average residual of the full-point affine transformation (Affine) is approximately 0.1018 m, and the maximum residual is approximately 0.1543 m. The average error of the leave-one-out-of-cross (LOOCV) external validation is approximately 0.281 m, and the maximum error is approximately 0.383 m. This indicates that using only the internal residuals of the control points involved in fitting to evaluate the layout scheme will underestimate the true external error. Based on the above results, this embodiment incorporates camera points, control points, check points, and long-term stable points into the optimized layout model, and uses the external validation of check points as an important criterion for judging the layout results.

[0038] In this embodiment, the excavation pit is defined as having a length of 46 m and a width of 18 m, with a maximum designed excavation depth of 8 m. An engineering local coordinate system is used on-site, with the known construction control point near the southwest corner of the pit as the origin. The long side of the pit is designated as the X-axis, the short side as the Y-axis, and the elevation as the Z-axis. The pit bottom, slope toe, slope top, inner boundary of the retaining structure, construction zone boundary, and designed excavation surface are defined as the key excavation estimation areas. The pit area is discretized into a digital elevation model (DEM) grid with a planar grid resolution of 0.05 m, resulting in the set of points to be measured.

[0039] in, Indicates the first There are 1 grid points to be tested, among which , , These represent the x-axis, y-axis, and z-axis coordinates of the j-th grid point to be measured, respectively. The pit bottom, slope toe, slope top, excavation boundary, and the zoned earthwork volume statistical area constitute the key areas. The remaining areas constitute the ordinary area. .right The grid points are assigned high weights for subsequent calculations of coverage quality, intersection quality, error propagation, and earthwork volume error.

[0040] Subsequently, a set of candidate fixed camera points was generated based on the site enclosure, capping beam, supports, tower, fixed wall surface, and hardened area outside the pit. :

[0041] in, These are candidate camera points. Four candidate camera points are placed on each side of the long side of the excavation pit, denoted as... , Two candidate camera points are placed at each end of the short side, denoted as follows: , , , A total of 12 candidate camera points were selected. For each candidate camera point, the installation coordinates, installation height, orientation angle, elevation angle, lens focal length, field of view, accessibility, and support stability were recorded. A set of candidate measurement points was generated based on the capping beam, the top of the retaining structure, the hardened ground outside the pit, the fixed wall, and the dedicated target support. :

[0042] in, The 18 candidate measurement points are arranged along the perimeter of the excavation pit and the transition area in the central field of view, avoiding concentration at the ends or a single boundary. Each candidate measurement point can be selected as a control point, check point, or long-term stable point. A binary variable is defined. Indicates whether it has been selected as a fixed camera mounting point. For specific candidate camera point representation, Indicates whether it has been selected as a control point. For specific candidate measurement points, Indicates whether it has been selected as a checkpoint. Indicates whether it has been selected as a long-term stable point:

[0043]

[0044]

[0045]

[0046] In this embodiment, the checkpoints do not participate in the external parameter solution and engineering coordinate registration; therefore, the following applies to each candidate measurement point:

[0047] in, Indicates whether it has been selected as a control point. This indicates whether a checkpoint has been selected. This constraint ensures that checkpoints can independently evaluate external accuracy. Previous parking lot experiments have shown that the residuals of the control points used for fitting cannot replace external inspection errors; therefore, this embodiment treats checkpoints as fixed output objects, rather than extracting them temporarily afterward.

[0048] After candidate points are generated, calculate any candidate camera point. With the point to be measured The visibility between them is represented as :

[0049] in, Indicate whether the view is obstructed by barriers, supports, machinery, or pit walls; This indicates whether the point to be measured is within the effective field of view of the camera; This indicates whether the observation distance, incident angle, and image edge margin meet the requirements. When all three are met, ,otherwise .

[0050] The effective number of observation cameras at the points to be measured is:

[0051] in, Indicates whether it has been selected as a fixed camera mounting point. This indicates the visibility between the candidate camera point and the point to be measured.

[0052] This embodiment requires a normal area. satisfy:

[0053] Key areas satisfy:

[0054] This setup stems directly from the needs of fixed-camera experiments and foundation pit monitoring. The fixed-camera experiments showed model splitting at four camera locations, indicating that insufficient cross-point common view disrupts the connectivity of the entire model. Foundation pit monitoring also requires multi-view coverage of the pit bottom, slopes, and corner areas to reduce occlusion and elevation errors. The foundation pit technical solution also proposes sufficient overlap between adjacent cameras, with at least two cameras observing the same area, and priority given to using three cameras for key areas.

[0055] Then, the ground sampling distance for each valid observation is calculated, denoted as: :

[0056] in, For candidate cameras To the test point distance, For pixel size, For camera focal length, This is the angle between the line of sight and the normal to the measured surface. Applying this angle to the effective observation:

[0057] In this embodiment, the ground sampling distance of the target in the ordinary area Use 20 mm / pixel, and 10 mm / pixel for key areas. For the pit bottom, slope toe, slope top, and excavation boundary, if a candidate camera point causes the far-end ground sampling distance GSD to exceed the threshold, the observation will not be counted in the valid observation count. This step is to avoid far-distance, low-resolution observations being mistakenly considered as valid coverage.

[0058] Further calculations for any two cameras , To measure the point Photogrammetric intersection angle :

[0059] In this embodiment, the effective intersection angle range is set as follows: to Priority selection in key areas to The intersection angle scoring function is denoted as... When the intersection corner falls within the preferred range, a higher value is used; otherwise, a lower value is used. The intersection quality of the points to be measured. Defined as:

[0060] in, , Indicates whether it has been selected as a fixed camera mounting point. , The respective and the Visibility between candidate camera points and the point j to be measured To prevent extremely small positive numbers with a denominator of zero, if a candidate deployment scheme satisfies the field of view coverage but has limitations in most key areas... A low score indicates that the camera is mainly coming from the same direction, resulting in insufficient depth accuracy, and the solution will be given a lower score.

[0061] For candidate measurement points Calculate its candidate camera Visibility of observation :

[0062] in, Indicate whether the view is obstructed by barriers, supports, machinery, or pit walls; This indicates whether the candidate measurement point is within the effective field of view of the camera; This indicates whether the observation distance, incident angle, and image edge margin meet the requirements. When all three are met, ,otherwise . No. Number of control points visible to selected cameras for:

[0063] in, Indicates whether it has been selected as a control point. This indicates the visibility between candidate camera points and measurement points. This embodiment requires:

[0064] This means that each selected camera must observe at least 6 control points. Each selected control point must be observed by at least 2 cameras, with key control points prioritized for observation by 3 or more cameras.

[0065] in, Indicates whether it has been selected as a fixed camera mounting point. Indicates candidate camera points With measuring points Visibility between them This indicates whether a point has been selected as a control point. The distribution of control points in each camera image is represented by a matrix. evaluate:

[0066]

[0067] in, Indicates whether it has been selected as a control point. Indicates candidate camera points With measuring points Visibility between them It assembles the normalized image coordinates into a quadratic basis vector containing constants, first-order and second-order cross terms. , These are the normalized x and y coordinates of the control points in the camera image. If the control points are concentrated near image edges, local corners, or the same straight line, then... The condition number increases, therefore the following applies:

[0068] in, Represents the distribution matrix of control points in the image. The condition number of express Upper limit of matrix condition number.

[0069] The spatial distribution of control points in the engineering coordinate system is determined by a matrix. evaluate:

[0070] in, Indicates whether it has been selected as a control point. As candidate measurement points, The coordinate centroid of all selected control points is given, and:

[0071] in, Represents the spatial distribution matrix of control points in the engineering coordinate system. The condition number of Represents the spatial distribution matrix The corresponding upper limit of the number of conditions.

[0072] This step is used to avoid control points being distributed only at the ends or on one side of the boundary. In the previous parking lot experiment, the Affine residual of the full-point affine transformation with five control points was approximately 10–15 cm, but the external error increased to 0.28–0.38 m, indicating that insufficient spatial distribution of control points leads to increased errors in unknown areas. This embodiment uses... and Simultaneously constraining the image distribution and spatial distribution, the control points simultaneously cover the upper, lower, left, right, and central regions of the image, and in the engineering coordinate system, they cover the boundaries, center, and periphery of the key excavation areas of the foundation pit.

[0073] The representativeness of checkpoints is evaluated using the following formula:

[0074]

[0075] in, For regional weights, For the point to be measured, As candidate measurement points, Representative of the points to be measured The nearest Euclidean distance to the set of candidate measurement points. This indicates that candidate measuring point k has been selected as a control point. The statistical zoning includes the pit bottom, slope toe, slope top, excavation boundary, and earthwork volume areas. The higher elevation allows checkpoints to be located closer to key survey areas. This embodiment requires at least three checkpoints to be selected, covering the vicinity of the pit bottom, the slope transition zone, and the boundary area, respectively. This arrangement is to prevent checkpoints from being located only in easily surveyable areas, which could result in quality assessments that do not represent the key excavation areas.

[0076] The long-term stable point quality is defined as:

[0077] in, This represents the long-term stability score of the structure where the candidate test points are located. This indicates whether the site has been selected as a long-term stable point. Crown beams, fixed walls, the top of retaining structures, and dedicated target supports receive higher scores, while soil surfaces to be excavated or affected by construction disturbance receive lower scores. Requirements for each selected camera:

[0078] in, Indicates whether it has been selected as a long-term stable point. Indicates candidate camera points With measuring points Visibility between them. That is, each camera can observe at least two long-term stable points, which are used for camera pose drift verification after each acquisition.

[0079] For 3D reconstruction error This embodiment uses the photogrammetric error propagation approximation formula:

[0080] in, Indicates the target distance or depth scale. Indicates focal length. Indicates pixel positioning error. Indicates the effective baseline length for observing this point:

[0081] in, , Indicates whether it has been selected as a fixed camera mounting point. , The respective and the Visibility between candidate camera points and the point j to be measured , This represents the specific candidate camera points. This formula is used in optimization to penalize deployment schemes with long distances, short baselines, and intersection angle differences. If a candidate scheme results in a certain area at the bottom of the pit being observed only by a long-distance camera on the same side, then... Small As the value increases, the candidate solution's score in the objective function decreases.

[0082] For rapid excavation estimation of foundation pits, the difference error in excavation depth between adjacent time steps Defined as:

[0083] in, Represents the difference in excavation depth at time t. Let represent the difference in excavation depth at time t-1. Volumetric error of each construction zone Approximately:

[0084] in, The error target for key areas of the foundation pit is:

[0085] in, Represented as weight values, This indicates the differential error in excavation depth. For volume error, This represents the balancing weighting coefficient for the volumetric error term. The objective is to prioritize reducing the depth difference error in the earthwork volume statistics areas at the pit bottom, slope toe, slope top, excavation boundary, and construction zone.

[0086] Based on the above indicators, a joint optimization objective function is constructed. :

[0087] in, For the intersection quality of the points to be measured, For the quality of the selected control points, This represents the quality of the selected checkpoint. This is expressed as the long-term steady-point mass. For 3D reconstruction error, This refers to the error value for key areas of the foundation pit. to These are weighting coefficients. Coverage quality. Defined as:

[0088] in, The effective number of observation cameras at the point to be measured. Test point The required minimum number of observations threshold, when At the time of its establishment, =1; when the condition is not met. =0, This represents the total number of test points. Equipment and maintenance costs. Defined as:

[0089] in, Indicates whether it has been selected as a fixed camera mounting point. Indicates whether it has been selected as a control point. Indicates whether it has been selected as a checkpoint. Indicates whether it was selected as a long-term stable point. This embodiment employs a two-stage solution. The first stage eliminates candidate camera points that are inaccessible due to construction limitations, severe obstruction, exceed the GSD ground sampling distance limit, or have insufficient coverage in key areas. The second stage uses a genetic algorithm or particle swarm optimization algorithm to solve for the remaining candidate point combinations. The maximum value is obtained. In this embodiment, the optimized output is as follows: 8 fixed camera installation points are selected from 12 candidate camera points, with 3 on each side of the long side of the foundation pit and 1 at each end of the short side; 8 control points, 3 check points, and 2 long-term stable points are selected from 18 candidate measurement points. The 8 control points cover the area around the foundation pit, the middle of the long sides on both sides, and the area near the toe of the slope; the 3 check points are located near the bottom of the pit, the slope transition area, and the boundary area, respectively; the 2 long-term stable points are placed on the capping beam and the fixed wall. The optimized general area requires at least two-view observation, and key areas such as the bottom of the pit and the excavation boundary require at least three-view observation. The overlap of the fields of view of adjacent cameras is controlled within the range of 50%–70%, and the intersection angle of key areas is mainly located in the range of 20°–45°.

[0090] After optimization, install and fix the camera and target according to the output results, as shown in the deployment plan. Figure 4 As shown. The camera is fixed with a rigid bracket, and the focus, exposure, white balance, and resolution are locked after installation. Control points, check points, and long-term stable points are obtained using a total station or RTK to acquire their 3D coordinates, forming a control_points.csv file with fields including point number, X, Y, Z, and point type. Calibration images are acquired for each camera to obtain its intrinsic parameters. Distortion parameters External reference .

[0091] In practice, the initial values ​​are first obtained using the n-point perspective method (PnP), and then optimized using multi-camera joint adjustment. If the maximum reprojection error of a control point is greater than 15 pixels, the image labeling of that point is reviewed or abnormal observations are removed. If the average reprojection error of a camera is greater than 8 pixels, the camera installation status, focal length, and target visibility are reviewed. This threshold is consistent with the quality evaluation system in the previous parking lot experiment, that is, an average reprojection error of less than 3 pixels is better, 3–5 pixels is acceptable, greater than 8 pixels requires inspection, and a maximum reprojection error greater than 15 pixels usually requires manual review.

[0092] After calibration, the fixed camera enters the periodic data acquisition phase. Each measurement time... Eight fixed cameras simultaneously or quasi-synchronously capture images. If excavators, vehicles, or personnel are moving on-site, hardware triggering or timestamp filtering is used, and masking is applied to areas with moving objects. After image acquisition, distortion correction, brightness normalization, sharpness detection, control point identification, and long-term stable point reprojection verification are performed. If the long-term stable point reprojection error exceeds a threshold, it indicates that the camera may have pose drifted, requiring recalibration or a check of the support frame.

[0093] In this embodiment, the 3D reconstruction of the foundation pit uses height field reconstruction under the condition of a fixed camera with known pose. The elevation Z of the foundation pit surface is represented as:

[0094] in, Represents the XY coordinates of a plane. (·) represents the height field function. An XY grid is established within the ROI of the foundation pit, with a resolution of 0.05 m; the elevation search range is set to... Before the first phase of excavation, the search area is determined based on the designed ground level and maximum excavation depth. In subsequent phases, the elevation from the previous moment is used to narrow down the search interval. For any candidate point:

[0095] Projecting this onto all visible camera images, local image patches are extracted, and multi-view photometric consistency costs are calculated. The first implementation uses zero-mean normalized cross-correlation (ZNCC) cost, combined with grayscale conversion, distortion correction, and local brightness normalization. The elevation with the lowest cost is selected for each grid point.

[0096] in, (·) is the zero-mean normalized cross-correlation cost function, which measures the cost of cross-correlation when... Assuming an elevation of Z, the similarity between image patches obtained from different camera projections is calculated. This represents the value of z when searching for the minimum cost function. Simultaneously, the reconstruction confidence is calculated.

[0097] in, This represents the number of visible cameras at that point. The interval between the suboptimal cost and the optimal cost. For local texture intensity, This is the incident angle index.

[0098] The height field reconstruction outputs the current-time elevation map (DEM), confidence map, and effective mask:

[0099]

[0100]

[0101] Where t represents a specific construction time. For time t The surface elevation obtained by grid point reconstruction is a digital elevation model (DEM). For time t Grid reconstruction confidence; This is the effective mask at time t. When the number of visible cameras at a certain grid point is less than 2, the confidence level is less than 0.6, the local texture is too weak, or it is mechanically occluded, this point is considered to be in... The marked areas are invalid and not included in the calculation of excavation depth and earthwork volume. After obtaining the digital elevation model (DEM), median filtering, bilateral filtering, or confidence-based local plane fitting are performed on the elevation map to suppress isolated noise.

[0102] Then, a 3D point cloud and mesh model are generated from the Digital Elevation Model (DEM). For each valid mesh center... Elevation Combined into three-dimensional points:

[0103] Point cloud attributes include RGB, confidence level, number of visible cameras, and valid markers. For a regular digital elevation model (DEM), four adjacent valid grid points are split into two triangles to generate a grid. When a grid point has low confidence, is missing, or the height difference between adjacent grid points exceeds a threshold, the relevant triangle is skipped to avoid generating erroneous patches in occluded or low-confidence areas. Output files include DEM_t.tif, confidence_t.png, valid_mask_t.png, pit_point_cloud_t.ply, pit_mesh_t.obj, pit_model_t.glb, and model_metadata_t.json. The pit design also explicitly uses DEM, confidence level, valid mask, point cloud, grid, binary 3D transport model (glTF Binary, GLB), and temporal difference results as the main output links.

[0104] After obtaining the DEM at adjacent time points, calculate the excavation depth difference:

[0105] when When this occurs, it indicates that the surface at that location has subsided, typically corresponding to excavation, settlement, or material removal; when When this occurs, it indicates a surface rise, which may correspond to overloading, backfilling, mechanical obstruction, or measurement noise. Statistics only consider... and The grid point calculation is performed. The minimum detectable change threshold is determined based on static repeated experiments; initially set at 0.03 m, it is corrected to 2–3 times the system elevation noise after repeated static data collection on site. The recommended threshold range for the foundation pit scheme is 2–5 cm, and it is required to be determined through static benchmark experiments.

[0106] For the The current stage of excavation volume change in each construction zone. for:

[0107] in, The area of ​​a single XY grid. Represents grid points Excavation depth difference, Let r be the set of all XY grid points within the r-th construction zone.

[0108] Quality assessment is divided into four categories: layout quality, calibration quality, reconstruction quality, and excavation estimation quality. Layout quality assesses coverage of general areas, coverage of key areas, three-view ratio of key areas, average ground sampling distance (GSD), intersection angle distribution, and representativeness of checkpoints. Calibration quality assesses reprojection error of control points, three-dimensional error of checkpoints, and reprojection error of long-term stable points. Reconstruction quality assesses effective coverage, average confidence level, proportion of low-confidence areas, proportion of local elevation abrupt changes, and static repeatability of the DEM. Excavation estimation quality assesses the difference error between adjacent time points, zonal earthwork volume error, stability of over-excavation and under-excavation identification, and computation time. In the foundation pit technical scheme, control point reprojection error, checkpoint three-dimensional error, static DEM repeatability, root mean square error (RMSE) of DEM elevation, difference error between adjacent time points, effective coverage, and computation time are also recommended evaluation indicators.

[0109] In this embodiment, the deployment scheme judgment criteria are as follows: the effective coverage rate of ordinary areas is not less than 90%, and the effective coverage rate of key areas is not less than 95%; the proportion of at least three-view observations in key areas is not less than 85%; the number of visible control points per camera is not less than 6; the number of visible long-term stable points per camera is not less than 2; the average reprojection error of control points is less than 5px, and the maximum reprojection error is less than 15px; the three-dimensional error of checkpoints is less than the engineering allowable threshold; the static repeatability of DEM is less than half of the minimum excavation identification threshold; the excavation depth map is only output in the effective mask and high confidence areas. If any of the above indicators do not meet the requirements, the system performs corresponding processing according to the problem type: when the coverage is insufficient, add candidate cameras or adjust the camera attitude; when the number of control point conditions is too large, add control points in the middle or elevation direction; when the representativeness of checkpoints is insufficient, adjust the position of checkpoints; when the number of long-term stable points is insufficient, add targets on the cap beam or fixed wall; when the excavation estimation error is too large, increase the weight of key areas and resolve the deployment scheme.

[0110] In summary, this invention generates a set of candidate fixed camera points and a set of candidate measurement points, and establishes a visibility coupling relationship between them. Based on this visibility coupling relationship, it generates coupling constraints, including control point constraints, checkpoint constraints, and long-term stable point constraints. This allows the selection of camera installation points, control points, checkpoints, and long-term stable points to mutually constrain and collaboratively optimize within a unified framework, avoiding the local optima problem caused by independently optimizing the distribution of camera points and control points in traditional methods, and improving the overall coordination of the deployment scheme. Simultaneously, based on the visibility coupling relationship and the spatial geometric relationship of the candidate fixed camera points, the prediction error during the deployment stage is calculated. Using coupling constraints as constraints and prediction error as the evaluation index, the deployment scheme is jointly solved, allowing the 3D reconstruction accuracy of the deployment results to be estimated during the deployment stage. This eliminates the need for accuracy verification and scheme adjustments after on-site deployment and image acquisition, reducing trial-and-error costs. Furthermore, excluding checkpoints from extrinsic parameter solving and independently evaluating external accuracy, as well as utilizing long-term stable points to perform periodic pose drift verification, enables continuous monitoring and timely correction of measurement results while ensuring camera calibration accuracy. This effectively improves the reliability and stability of fixed-camera mapping systems under long-term deployment conditions. Therefore, this invention solves the problem in existing technologies where the lack of joint configuration optimization between fixed camera locations and measurement point layout makes it difficult to balance reconstruction accuracy and external verification reliability in the deployment scheme.

[0111] Example 2 Please see Figure 2 The diagram shown is a structural block diagram of the fixed camera and measurement point joint optimization deployment system proposed in the second embodiment of the present invention. The fixed camera and measurement point joint optimization deployment system 200 includes: a point set determination module 21, a coupling constraint determination module 22, a deployment scheme determination module 23, and a verification and evaluation module 24, wherein: The point set determination module 21 is used to generate a candidate fixed camera point set and a candidate measurement point set, wherein the candidate measurement points can be selected as control points, check points or long-term stable points; The coupling constraint determination module 22 is used to establish the visibility coupling relationship between the candidate fixed camera point and the candidate measurement point, and generate coupling constraints based on the visibility coupling relationship. The coupling constraints include control point constraints, check point constraints and long-term stable point constraints. The deployment scheme determination module 23 is used to calculate the prediction error in the deployment stage based on the visibility coupling relationship and the spatial geometric relationship of the candidate fixed camera points, and to jointly solve the deployment scheme of fixed camera installation points, control points, check points and long-term stable points using the coupling constraint as the constraint condition and the prediction error as the evaluation index. The verification and evaluation module 24 is used to deploy fixed cameras and measurement points according to the deployment scheme, complete camera calibration based on control points, independently evaluate external accuracy based on check points, and perform periodic pose drift verification based on long-term stable points.

[0112] Example 3 In another aspect, the present invention also proposes an electronic device, please refer to [link to relevant documentation]. Figure 3 The diagram shows an electronic device according to the third embodiment of the present invention, including a memory 20, a processor 10, and a computer program 30 stored in the memory and executable on the processor. When the processor 10 executes the computer program 30, it implements a fixed camera and measuring point joint optimization layout method as described above.

[0113] In some embodiments, the processor 10 may be a central processing unit (CPU), controller, microcontroller, microprocessor or other data processing chip, used to run program code stored in memory 20 or process data, such as executing access restriction programs.

[0114] The memory 20 includes at least one type of readable storage medium, such as flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 20 can be an internal storage unit of an electronic device, such as the hard disk of the electronic device. In other embodiments, the memory 20 can also be an external storage device of the electronic device, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc. Furthermore, the memory 20 can include both internal and external storage units of the electronic device. The memory 20 can be used not only to store application software and various types of data of the electronic device, but also to temporarily store data that has been output or will be output.

[0115] It should be pointed out that, Figure 3 The structure shown does not constitute a limitation on the electronic device. In other embodiments, the electronic device may include fewer or more components than shown, or combine certain components, or have different component arrangements.

[0116] This invention also proposes a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method for the joint optimization of fixed camera and measurement point deployment.

[0117] Those skilled in the art will understand that the logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as a ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can mean any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0118] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0119] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0120] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0121] The above embodiments merely illustrate several implementation methods of the present invention, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this patent should be determined by the appended claims.

Claims

1. A method for jointly optimizing the deployment of fixed cameras and measuring points, characterized in that, The method includes: Generate a set of candidate fixed camera points and a set of candidate measurement points. The candidate measurement points can be selected as control points, check points, or long-term stable points. Establish the visibility coupling relationship between the candidate fixed camera point and the candidate measurement point, and generate coupling constraints based on the visibility coupling relationship. The coupling constraints include control point constraints, check point constraints, and long-term stable point constraints. Based on the visibility coupling relationship and the spatial geometric relationship of the candidate fixed camera points, the prediction error in the deployment stage is calculated. Using the coupling constraint as the constraint condition and the prediction error as the evaluation index, the deployment scheme of fixed camera installation points, control points, check points and long-term stable points is jointly solved. Deploy fixed cameras and measurement points according to the deployment scheme, complete camera calibration based on control points, independently evaluate external accuracy based on check points, and perform periodic pose drift verification based on long-term stable points.

2. The method for joint optimization of fixed camera and measuring point layout according to claim 1, characterized in that, The steps for generating the candidate fixed camera point set and the candidate measurement point set include: The spatial information of the engineering area to be measured is obtained, the engineering area to be measured is discretized into a set of spatial points to be measured, and divided into key areas and ordinary areas according to the importance of measurement, and different weight coefficients are assigned to the spatial points to be measured in different areas. A set of candidate fixed camera points is generated based on the on-site installation conditions. The candidate fixed camera points include installation coordinates, installation height, orientation angle, elevation angle, field of view, stability level, and construction accessibility information. A set of candidate measurement points is generated based on the available deployment conditions on site. The candidate measurement points contain attribute information that can be selected as control points, checkpoints, or long-term stable points, and the selection status of each candidate point is characterized by binary decision variables.

3. The method for jointly optimizing the deployment of fixed cameras and measuring points according to claim 1, characterized in that, The steps of establishing the visibility coupling relationship between the candidate fixed camera point and the candidate measurement point, and generating coupling constraints based on the visibility coupling relationship, include: Based on the installation coordinates, attitude parameters, field of view parameters of the candidate fixed camera points and the spatial coordinates of the spatial point to be measured, the line of sight occlusion state, effective field of view state and observation quality state are calculated, and the visibility coupling relationship between the candidate fixed camera points and the candidate measurement points is determined. Based on the visibility coupling relationship, the control point constraint, the checkpoint constraint, and the long-term stable point constraint are generated respectively. Based on the visibility state, the intersection angle between any two selected fixed cameras and the same spatial point to be measured is calculated. The multi-view observation geometric quality of each spatial point to be measured is evaluated based on the intersection angle, and the intersection angle is within a preset effective angle range. The control point constraints include control point visibility quantity constraints, control point image distribution constraints, and control point spatial distribution constraints; the checkpoint constraints include mutual exclusion constraints between checkpoints and control points and checkpoint representativeness constraints; and the long-term stable point constraints include long-term stable point visibility quantity constraints and long-term stable point stability score constraints.

4. The method for joint optimization of fixed camera and measuring point layout according to claim 1, characterized in that, Based on the visibility coupling relationship and the spatial geometric relationship of the candidate fixed camera points, the steps for calculating the prediction error during the deployment phase include: Based on the visibility coupling relationship, the effective baseline length of the spatial point to be measured is determined according to the spatial distance between any two selected fixed camera points that simultaneously observe the same spatial point to be measured. Based on the effective baseline length, the depth scale of the spatial point to be measured, the camera focal length, and the pixel positioning error, the 3D reconstruction prediction error of each spatial point to be measured is calculated. The 3D reconstruction prediction error increases as the effective baseline length decreases. Based on the 3D reconstruction prediction error and the elevation estimation error at adjacent times, the depth difference prediction error of each spatial point to be measured is calculated. Based on the depth difference prediction error and the preset grid area, the volume prediction error of each construction zone is calculated to generate the engineering measurement prediction error.

5. The method for joint optimization of fixed camera and measuring point layout according to claim 1, characterized in that, Using the coupling constraints as constraints and the prediction error as the evaluation index, the steps for jointly solving the layout scheme of fixed camera mounting points, control points, checkpoints, and long-term stable points include: A joint optimization objective function is constructed, wherein the coverage quality of the survey area, the geometric quality of multi-view observation, and the satisfaction of coupling constraints are used as positive optimization terms, and the prediction error and deployment cost of the deployment stage are used as negative penalty terms. The terms are weighted and integrated by preset weight coefficients. Using the coupling constraint as a hard constraint and the prediction error meeting the preset accuracy requirement as the evaluation criterion, the feasibility of the binary decision variable combination of candidate fixed camera points and candidate measurement points is screened. Before performing the combined search, infeasible candidate fixed camera points are eliminated based on preset feasibility conditions. Then, an optimization algorithm is used to search and solve the combination of the remaining candidate fixed camera points and all candidate measurement points. Under the premise of satisfying all the coupling constraints, the joint optimization objective function achieves the optimal combination of binary decision variable values, and outputs the layout scheme of fixed camera installation points, control points, check points and long-term stable points.

6. The method for jointly optimizing the deployment of fixed cameras and measuring points according to claim 1, characterized in that, The steps include: deploying fixed cameras and measurement points according to the deployment scheme; completing camera calibration based on control points; independently evaluating external accuracy based on check points; and performing periodic pose drift verification based on long-term stable points. Install fixed cameras according to the fixed camera installation points in the deployment scheme, deploy corresponding targets according to the control point set, check point set and long-term stable point set, and measure the three-dimensional spatial coordinates of each target. Based on the three-dimensional spatial coordinates of the control points and their pixel coordinates in the calibration image, a projection relationship is established. The extrinsic parameters of each fixed camera are solved by minimizing the reprojection error, and the camera calibration and engineering coordinate registration are completed. Using a calibrated fixed camera to acquire images, the external accuracy is independently evaluated based on the deviation between the 3D spatial coordinates of the checkpoints and the 3D coordinates obtained from multi-view reconstruction. During each data acquisition cycle, the reprojection error is calculated using the image coordinates of long-term stable points. When the reprojection error exceeds a preset threshold, it is determined that the pose of the corresponding fixed camera has shifted and recalibration is triggered.

7. The method for joint optimization of fixed camera and measuring point layout according to claim 6, characterized in that, After completing the camera calibration and engineering coordinate registration steps, the following steps are also included: Periodic image acquisition is performed using a calibrated fixed camera. Under known camera pose conditions, the three-dimensional information of the area to be measured is reconstructed, an elevation map and a reconstructed confidence map are generated, and areas with confidence scores below a preset threshold are marked as invalid. Differential calculations are performed based on elevation maps of adjacent periods to extract engineering parameters within the effective area. When the proportion of invalid areas in a construction zone exceeds a preset threshold, the engineering parameters of that zone are marked as low confidence.

8. A fixed camera and measuring point joint optimization deployment system, characterized in that, The system, used to implement the method for joint optimization of fixed camera and measuring point deployment as described in any one of claims 1 to 7, comprises: The point set determination module is used to generate a candidate fixed camera point set and a candidate measurement point set. The candidate measurement points can be selected as control points, check points, or long-term stable points. The coupling constraint determination module is used to establish the visibility coupling relationship between the candidate fixed camera point and the candidate measurement point, and generate coupling constraints based on the visibility coupling relationship. The coupling constraints include control point constraints, check point constraints and long-term stable point constraints. The deployment scheme determination module is used to calculate the prediction error in the deployment stage based on the visibility coupling relationship and the spatial geometric relationship of the candidate fixed camera points, and to jointly solve the deployment scheme of fixed camera installation points, control points, check points and long-term stable points using the coupling constraint as the constraint condition and the prediction error as the evaluation index. The verification and evaluation module is used to deploy fixed cameras and measurement points according to the deployment plan, complete camera calibration based on control points, independently evaluate external accuracy based on check points, and perform periodic pose drift verification based on long-term stable points.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps of the fixed camera and measuring point joint optimization layout method as described in any one of claims 1 to 7.

10. An electronic device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement a method for joint optimization of fixed camera and measuring point deployment as described in any one of claims 1 to 7.

Images