Method and device for optimizing scanning parameters of 3D laser scanner, and electronic equipment

Abstract

The application provides a 3D laser scanner scanning parameter optimization method and device and electronic equipment, and relates to the technical field of laser scanning. The method comprises the following steps: determining an average density threshold value and a coefficient of variation threshold value; taking the average density threshold value and the coefficient of variation threshold value as constraint condition boundaries, and constructing a multi-objective optimization model with the maximum average density and the minimum coefficient of variation as targets; using a GA-PSO hybrid algorithm combined with a Pareto front screening strategy to solve the multi-objective optimization model, and obtaining a Pareto non-inferior solution set; and screening the Pareto non-inferior solution set based on an application scenario to obtain target scanning parameters. The GA-PSO hybrid algorithm combined with the Pareto front screening strategy can quickly obtain an optimal solution cluster that meets double-target requirements; and the secondary screening of the Pareto non-inferior solution set based on the application scenario can ensure that the finally output scanning parameters are highly matched with actual application requirements, and the stability of point cloud data quality is ensured.

Description

Technical Field

[0001] This invention relates to the field of laser scanning technology, and in particular to a method, apparatus, and electronic device for optimizing scanning parameters of a 3D laser scanner. Background Technology

[0002] As modern civil engineering structures continue to evolve towards larger spans, higher piers, and greater complexity, engineering surveying places increasingly stringent demands on data accuracy, efficiency, and intelligence. Three-dimensional laser scanning technology, with its non-contact, high-precision, and high-density data acquisition characteristics, has been widely applied in numerous fields, such as bridge and tunnel health monitoring, structural deformation analysis, construction quality control, and digital archiving of historical buildings. The point cloud data generated by scanning can accurately reconstruct target surfaces, achieving millimeter-level morphological modeling and measurement, providing crucial data support for structural safety assessment and digital twin modeling. However, the stability of point cloud data quality is key to the successful implementation of 3D scanning engineering applications. The density and spatial uniformity of the point cloud directly affect the reliability of subsequent model reconstruction, crack identification, and deformation analysis.

[0003] In existing technologies, point cloud density is affected by multiple factors such as scanning distance, resolution, incident angle, and target material, exhibiting obvious nonlinear coupling characteristics. These parameters are usually set by operators based on experience, leading to unpredictable data density, large quality fluctuations, and problems such as "overly dense redundancy" or "sparse distortion," making it impossible to guarantee the stability of point cloud data quality. Summary of the Invention

[0004] This invention provides a method, apparatus, and electronic device for optimizing scanning parameters of a 3D laser scanner, in order to solve the problem that in the prior art, scanning parameters are based on experience and are manually set, which is not accurate enough and leads to unstable point cloud data quality.

[0005] In a first aspect, embodiments of the present invention provide a method for optimizing scanning parameters of a 3D laser scanner, including: Determine the average density threshold and the coefficient of variation threshold; By using the average density threshold and the coefficient of variation threshold as the boundary conditions, a multi-objective optimization model is constructed with the goal of maximizing the average density and minimizing the coefficient of variation. A GA-PSO hybrid algorithm combined with a Pareto front screening strategy is used to solve a multi-objective optimization model and obtain a set of Pareto non-dominated solutions. The target scanning parameters are obtained by filtering the Pareto nondominated solution set based on the application scenario.

[0006] Secondly, embodiments of the present invention provide a 3D laser scanner scanning parameter optimization device, comprising: The threshold determination module is used to determine the average density threshold and the coefficient of variation threshold. The model building module is used to construct a multi-objective optimization model with the average density threshold and the coefficient of variation threshold as the boundary conditions, aiming to maximize the average density and minimize the coefficient of variation. The model solving module is used to solve multi-objective optimization models using the GA-PSO hybrid algorithm combined with the Pareto front screening strategy, and obtain the Pareto non-dominated solution set. The parameter output module is used to filter the Pareto nondominated solution set based on the application scenario to obtain the target scanning parameters.

[0007] Thirdly, embodiments of the present invention provide an electronic device, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the 3D laser scanner scanning parameter optimization method as described in the first aspect or any possible implementation of the first aspect.

[0008] This invention provides a method, apparatus, and electronic device for optimizing scanning parameters of a 3D laser scanner. The method includes: determining an average density threshold and a coefficient of variation threshold; using the average density threshold and coefficient of variation threshold as boundary conditions to construct a multi-objective optimization model with the objectives of maximizing average density and minimizing coefficient of variation; solving the multi-objective optimization model using a GA-PSO hybrid algorithm combined with a Pareto front screening strategy to obtain a set of Pareto non-dominated solutions; and screening the Pareto non-dominated solution set based on the application scenario to obtain the target scanning parameters. This application utilizes a GA-PSO hybrid algorithm combined with a Pareto front screening strategy to quickly obtain the optimal solution cluster that meets both objectives; the secondary screening of the Pareto non-dominated solution set based on the application scenario ensures that the final output scanning parameters highly match the actual application requirements, without relying on human experience settings, resulting in more accurate scanning parameters and guaranteeing the stability of point cloud data quality. Attached Figure Description

[0009] Figure 1 This is a flowchart illustrating the implementation of a 3D laser scanner scanning parameter optimization method provided in an embodiment of the present invention. Figure 2 This is a schematic diagram of the scanning experiment setup provided in an embodiment of the present invention; Figure 3 This is a flowchart illustrating a GA-PSO hybrid algorithm provided in an embodiment of the present invention; Figure 4 This is a 3D Pareto front surface target relationship diagram provided in an embodiment of the present invention; Figure 5 This is a radar chart of the main parameter combination scoring provided in the embodiments of the present invention; Figure 6This is a schematic diagram of the structure of the 3D laser scanner scanning parameter optimization device provided in an embodiment of the present invention; Figure 7 This is a schematic diagram of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0010] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0011] Figure 1 A flowchart illustrating the implementation of a 3D laser scanner scanning parameter optimization method provided in this embodiment of the invention is shown below. Details are as follows: refer to Figure 1 This invention provides a method for optimizing scanning parameters of a 3D laser scanner, comprising: S101: Determine the average density threshold and the coefficient of variation threshold; Average density directly reflects the core quality level of the scan results and is the basis for judging whether the scan data has practical application value. Too low a density will lead to invalid data, so it needs to be "maximized" as a target, while setting a threshold to ensure the minimum effective level.

[0012] The coefficient of variation accurately reflects the dispersion (i.e., consistency) of the scan results. An excessively large coefficient of variation means that the scan results at different locations and in different batches under the same scan parameters fluctuate drastically, which directly affects the reliability of the application. Therefore, it should be taken as a "minimization" target, and a threshold should be set to control the maximum fluctuation range.

[0013] The average density and the coefficient of variation often have a restrictive relationship (e.g., increasing the density may lead to increased fluctuations), which precisely constitutes the core contradiction of multi-objective optimization; by combining the threshold constraints of the two with the objective optimization, an "effective and stable" scanning effect can be achieved.

[0014] The average density threshold and the coefficient of variation threshold are the constraints for building the optimization model. Their determination needs to be combined with the correlation between the measurement error of the sample data and the scanning parameters to ensure the rationality and applicability of the thresholds.

[0015] Scanning parameters directly determine the average density and coefficient of variation (core characteristics) of the scan results, and these core characteristics directly affect the magnitude of measurement error (e.g., too low an average density will result in a weak signal, increasing measurement error; too high a coefficient of variation means discrete data, reducing measurement accuracy). Based on this, the average density threshold and coefficient of variation threshold can be derived in reverse from the measurement error and scanning parameters of the sample.

[0016] In one possible implementation, S101 may include: S1011: Obtain the measured and actual values ​​of each sample, and calculate the measurement error of each sample; the scanning parameters are different for each sample. S1012: Based on the measurement error of each sample and the scanning parameters of each sample, the average density threshold and the coefficient of variation threshold are obtained.

[0017] This application links the threshold to the measurement error that directly reflects the scanning quality, avoiding the problem of "the threshold being qualified but the actual scanning error exceeding the standard", ensuring that the constraints of the subsequent optimization model have practical significance, and improving the usability of the optimization results from the source.

[0018] In one possible implementation, S1012 includes: 1. Based on the scanning parameters of each sample, the predicted average density value and predicted coefficient of variation value of each sample are predicted. Based on the scanning parameters of each sample, a pre-defined prediction model (such as a machine learning model) is used to predict the predicted average density value and the predicted coefficient of variation value for each sample.

[0019] Tests showed that the LightGBM model performed best in terms of mean absolute error (MAE), root mean square error (RMSE), and goodness of fit (R²), demonstrating strong fitting accuracy and generalization ability. Therefore, the prediction model in this application can be the LightGBM model.

[0020] refer to Figure 2 This application employs a Leica P40 3D laser scanner. To reveal the characteristics of point cloud density influenced by multiple coupled factors, a full-factor combination design strategy is adopted, controlling variables from three dimensions: distance, angle, and resolution. 1) Scanning distance: Six typical values ​​of 2m, 5m, 8m, 11m, 14m, and 17m are set, covering near-field, mid-field, and far-field conditions; 2) Incident angle: The tilt angle of the acrylic plate is adjusted from 0° to 60° using a flatbed carriage to simulate different inclinations of the structural surface; 3) Scanning resolution: Seven levels are selected sequentially: 50mm, 25mm, 12.5mm, 6.3mm, 3.1mm, 1.6mm, and 0.8mm. Through the above 3D combination, a total of 279 sets of effective original point cloud data are obtained, providing sufficient data volume and broad coverage.

[0021] The original point cloud data is first subjected to noise removal and ROI region clipping to ensure that only valid points on the acrylic sheet surface are retained. To avoid the error amplification problem in the boundary region of the traditional neighborhood radius method, this application adopts a density estimation method based on neighbor point statistics. For any point Pi in the point cloud, a point set Ni is obtained in its neighborhood radius r, and the polygon area Ai projected onto the XOZ plane of these points is calculated. Then, the local density Di is expressed as:

[0022] The coefficient of variation (CoV) is expressed as:

[0023] Where μ is the average density of the region, and σ is the standard deviation. The smaller the CoV value, the more uniform the point cloud distribution. The average density and CoV together constitute a dual indicator system for scan quality.

[0024] From all 279 preprocessed point clouds, the system extracts the spatial 3D coordinates (x, y, z), scan resolution, and local density values ​​of all points, forming a total of approximately 96.57 million sample data points. To balance model training efficiency and data representativeness, equidistant sampling (taking 1 out of every 482 data points) is used to construct a balanced dataset of approximately 200,000 samples. Missing value imputation, outlier removal, and Z-score normalization are then performed, and the training and test sets are divided in an 8:2 ratio. These are directly used as input features for the LightGBM regression model, demonstrating good scalability and reusability.

[0025] 2. Based on the measurement error of each sample, the predicted average density value of each sample, and the predicted coefficient of variation value of each sample, the correspondence between average density and error and the correspondence between coefficient of variation and error are fitted. In the error fitting stage, the difference between the actual measured height and the theoretical size (measurement error) is extracted by fitting the target whiteboard boundary and using an equidistant mesh method to support model error construction. The processing flow is as follows: (1) Boundary fitting and corner point extraction: Fit the boundary of the target rectangular whiteboard, extract four key corner points A, B, C, and D, calculate the side length and divide the AB side into 249 segments to determine the standard step size; (2) Sliding window type panel subdivision: Starting from the right boundary, the 0~124 step size area is removed in sequence, and the corner points and heights are extracted from the remaining area to generate a total of 125 sets of height data; (3) Symmetrical sliding to generate left-side height data: Similarly, slide inward from the left boundary to generate the height of 125 regions on the left; (4) Merge to generate a complete elevation error dataset: Each point cloud can generate 250 h values ​​as the actual measured height.

[0026] Each measurement record ultimately includes: measurement height h; spatial coordinates (x, y, z); resolution R. The runtime t of the entire measurement process is also calculated, recorded, and output to a performance analysis results file.

[0027] Based on the actual measurement results of the target board (design size: height 0.9m), the height of the board was measured under different scanning parameters, and the measurement error corresponding to each set of scanning parameters was calculated. This measurement error was then paired with the predicted average density value and predicted coefficient of variation value obtained by the model under the corresponding scanning conditions to construct two sets of data pairs: "predicted average density - measurement error" and "predicted coefficient of variation - measurement error". Data fitting methods were used to fit the two sets of data pairs respectively to obtain the correspondence between average density and error, and the correspondence between coefficient of variation and error. During the fitting process, the reliability of the correspondence needs to be verified using a goodness-of-fit index (such as R², R² ≥ 0.85 is recommended).

[0028] 3. Based on the correspondence between average density and error and the maximum measurement error, the average density threshold is obtained; 4. Based on the correlation between the coefficient of variation and the error, and the maximum measurement error, the threshold of the coefficient of variation is obtained.

[0029] Based on the quality requirements of actual application scenarios, the maximum permissible measurement error is determined by substituting the average density-error correspondence and the coefficient of variation-error correspondence, and the average density threshold and coefficient of variation threshold are obtained respectively.

[0030] This application establishes a quantitative relationship between core characteristics and errors through fitting, and uses reverse derivation logic to ensure that the threshold directly matches the upper limit of the error, significantly reducing the risk of "threshold being qualified but error exceeding the standard". Furthermore, the threshold can be flexibly adjusted based on the maximum measurement error in a given scenario, improving scenario adaptability.

[0031] In one possible implementation, obtaining the average density threshold based on the average density-error correspondence and the maximum measurement error may include: (1) Substitute the maximum measurement error into the average density-error correspondence to obtain the average density threshold; The average density-error correspondence includes:

[0032] in, The measurement error is represented by ρ, which is the average density. , , These are the fitting coefficients; Based on the correlation between the coefficient of variation and the error, and the maximum measurement error, the threshold for the coefficient of variation can be obtained, which may include: (1) Substitute the maximum measurement error into the coefficient of variation-error correspondence to obtain the coefficient of variation threshold; The coefficient of variation-error correspondence includes:

[0033] in, The coefficient of variation is 1. , , represents the fitting coefficient.

[0034] Experiments show that the measurement error decreases exponentially with increasing average density, and the goodness of fit R0 2 =0.89, indicating that the increase in average density has a significant impact on reducing measurement error, clearly revealing the law that "measurement error decreases exponentially with increasing average density." Furthermore, the measurement error exhibits an approximately quadratic growth relationship with the coefficient of variation, and the goodness of fit R0 is... 2 =0.91.

[0035] For example, using a maximum measurement error of 1% (corresponding to an absolute error of 0.8 mm) as the control standard, and substituting this into the above formula, the minimum average density threshold to ensure accuracy is approximately 190.83 points / cm². To enhance robustness and system fault tolerance, a 10% safety margin is introduced, raising the recommended threshold to 210 points / cm². Simultaneously, the maximum permissible coefficient of variation threshold is calculated to be 0.6.

[0036] S102: Using the average density threshold and the coefficient of variation threshold as the boundary conditions, construct a multi-objective optimization model with the goal of maximizing the average density and minimizing the coefficient of variation. This application uses the average density threshold and the coefficient of variation threshold as the constraint boundary conditions (i.e., average density ≥ 1). points / cm²; coefficient of variation ≤ (0.6), construct a multi-objective optimization model with "maximum average density" and "minimum coefficient of variation" as the core objectives to achieve dual optimization of the effectiveness and stability of the scanning results. The decision variable is the combination of scanning parameters.

[0037] For example, the objective function can be:

[0038] in, This represents the combination of four-dimensional scanning parameters, which sets the resolution. Configure the average density of 81 subgrids under these scan parameters. This represents the standard deviation of density. (Function) Characterizes the density level of point cloud acquisition. Characterizes the stability of spatial distribution.

[0039] S103: The GA-PSO hybrid algorithm combined with the Pareto front screening strategy is used to solve the multi-objective optimization model and obtain the Pareto non-dominated solution set. The above-mentioned optimization of scanning parameters is a typical problem of finding feasible solutions and further optimizing them in a four-dimensional parameter space (x, y, z, resolution).

[0040] To ensure that the multi-objective optimization model has broad adaptability and comprehensive solution space exploration capabilities, this application defines the search space as follows: Spatial variable range: x, y, z ∈ [1.0, 10.0], step size 0.6, a total of 16 equally spaced values; Resolution options: r∈{50, 25, 12.5, 6.3, 3.1, 1.6, 0.8} The search space size is: Group parameter combination Since theoretical enumeration is not feasible, evolutionary algorithms are needed for intelligent sampling and focused search.

[0041] Given the complexity and nonlinearity of the parameter space, this application adopts a GA-PSO hybrid algorithm optimization framework that combines genetic algorithm and particle swarm optimization algorithm. For details, please refer to [link / reference]. Figure 3 .

[0042] The GA-PSO hybrid algorithm combines the global search capability of Genetic Algorithm (GA) with the local convergence advantage of Particle Swarm Optimization (PSO). Its core is to achieve synergistic optimization through operator fusion, combining global exploration with local refinement, thus fully leveraging the global exploration capabilities of GA and the local fine-grained search advantages of PSO. Furthermore, this application incorporates a Pareto front screening strategy in each iteration, comparing the objective function vectors of all particles and eliminating dominated solutions. Non-dominated solutions are stored in an external archive set, and the size of the archive set is maintained through crowding calculations. After the iteration ends, the solutions in the external archive set constitute the Pareto non-dominated solution set.

[0043] In one possible implementation, the fitness function of the GA-PSO hybrid algorithm It can be:

[0044]

[0045]

[0046] in, and These are the weighting coefficients. and These are the normalized density score and the normalized coefficient of variation penalty, respectively. This represents the actual density of the current point cloud. and These represent the maximum and minimum point cloud densities, respectively. The coefficient of variation of the current point cloud. This represents the maximum value of the coefficient of variation.

[0047] When a solution does not meet the above two threshold conditions, the fitness function is directly set to zero, or a penalty factor is added to eliminate it during iteration. To verify the stability and convergence ability of the constructed GA-PSO hybrid algorithm in search spaces with different parameter combinations, this application conducted five independent and repeated experiments on the algorithm. Each round used the same initial population size, maximum number of iterations, and fitness function, and recorded the fitness value change curve of the global optimum, the average fitness change trend, and the error and CV level of the final solution in each round. All five rounds of experiments showed good performance.

[0048] S104: Based on the application scenario, the Pareto non-dominated solution set is filtered to obtain the target scanning parameters.

[0049] The Pareto nondominated solution set provides multiple optimal parameter schemes, which need to be further filtered according to the specific application scenario to adapt to the specific application scenario and finally determine the target scanning parameters.

[0050] In one possible implementation, S104 may include: S1041: According to the preset performance classification criteria, classify and filter the solutions in the Pareto non-dominated solution set to obtain three types of target solution sets; According to the preset performance classification criteria, the solutions in the Pareto non-dominated solution set are classified. The core of the classification criteria is set around the balance between "mean density" and "coefficient of variation". The three target solution sets can be Type-H solutions, Type-U solutions, and Type-B solutions. The specific classification criteria are shown in Table 1.

[0051] Table 1. Three Types of Optimal Solution Selection Mechanisms

[0052] S1042: Based on the application scenario, select one of the three target solution sets as the target solution set; For example, (1) if the scenario is material density testing, medical image enhancement, etc., "high density solution set" should be selected first; (2) if the scenario is batch industrial product testing, long-term environmental monitoring, etc., "high stability solution set" should be selected first; (3) if the scenario is general scanning, multiple needs to be considered, etc., "balanced solution set" should be selected.

[0053] S1043: Based on the analytic hierarchy process (AHP), determine the performance score of each solution in the target solution set; The core of the Analytic Hierarchy Process (AHP) is to transform qualitative requirements into quantitative weights and calculate scores by combining the core performance indicators of the scanning parameters.

[0054] In one possible implementation, S1043 may include: 1. For any target scanning parameter, predict the predicted average density value and the predicted coefficient of variation value of the target scanning parameter; calculate the prediction error of the target scanning parameter based on the predicted average density value and the correspondence between average density and error; and obtain the performance score of the target scanning parameter based on the predicted average density value, the predicted coefficient of variation value, and the prediction error.

[0055] For any target scanning parameter, using a prediction model consistent with S1012 (e.g., the LightGBM model), the predicted average density value and predicted coefficient of variation of that parameter are predicted. Then, combining the "average density-error correspondence" fitted in S1012, the predicted average density value is substituted into the correspondence to calculate the prediction error of that parameter. In one possible implementation, the performance score of the target scanning parameter is obtained based on the predicted average density value, predicted coefficient of variation value, and prediction error, which may include: (1) Based on the predicted average density value, predicted coefficient of variation value and prediction error of the target scanning parameters, the performance score of the target scanning parameters is obtained by combining the first formula; The first formula may include:

[0056] in, The performance score for the scanning parameters of the i-th target; , , These are the predicted average density value, predicted coefficient of variation value, and prediction error of the scanning parameters for the i-th target, respectively. and To predict the minimum and maximum values ​​of the average density; and To predict the minimum and maximum values ​​of the coefficient of variation; and These represent the minimum and maximum values ​​of the prediction error; , and These are the weighting coefficients.

[0057] For example, , and The values ​​can be 0.45, 0.35, and 0.2 respectively.

[0058] S1044: Sort the solutions from largest to smallest according to their performance scores, and select the first preset number of solutions as the target scanning parameters.

[0059] Finally, a preset number of solutions are selected as the target scan parameters. For example, the preset number can be 3, selecting the first three solutions for user selection; the preset number can also be 1, outputting only the first one as the final scan parameter. The specific setting can be determined according to the actual application requirements.

[0060] The above method will be described in detail below with reference to specific embodiments.

[0061] 1. Initial screening of samples After optimizing a total of 23,296 possible parameter combinations, and under the dual quality constraints (average density ≥ 210 points / cm² and CoV ≤ 0.6), 1,389 lattice solutions were finally identified, accounting for 5.96% of all combinations. Tables 2 and 3 list the typical value ranges and statistical results of the qualified solution parameters under 0.8 mm and 1.6 mm resolutions. The 0.8 mm resolution shows a significant advantage, with an effective solution rate as high as 78.5%, and the solutions are spatially continuous and have a wide range (e.g., x: 2.0-6.5 m, y: 3.1-7.8 m), offering high operational flexibility. In contrast, the 1.6 mm resolution has an effective solution rate of only 7.3%, a narrow parameter range, and more stringent requirements for location configuration; even slight deviations can lead to density decreases or CoV exceeding the limit, making it unsuitable for large-scale engineering applications. Therefore, it can be seen that the low-resolution configuration has a larger solution space capacity and distribution flexibility while ensuring quality. In high-precision measurement scenarios, 0.8mm should be selected as the baseline option for scanning resolution, and the spatial station deployment strategy should be dynamically adjusted based on the above value range.

[0062] Table 2. Statistical distribution of characteristic parameters of qualified solutions under different resolution conditions

[0063] Table 3. Statistical distribution of initial screening samples under different resolution conditions

[0064] 2. Model Solving - Pareto Non-dominated Solution Set From 3897 combinatorial lattice solutions, 189 Pareto nondominated solutions were selected, forming a Pareto nondominated solution set. This solution set has the following characteristics: average density range: above 500 points / cm²; CoV range: 0.45~0.50; most solutions are located in the high-density-high-stability transition region, such as... Figure 4 As shown, it exhibits good balance.

[0065] 3. Performance-based classification and filtering According to the preset performance classification standard, the 189 non-inferior solutions were classified by attribute, and finally identified as: 4 Type-H solutions, 6 Type-U solutions, and 12 Type-B solutions.

[0066] To demonstrate the differences in core indicators among various recommended scanning parameters, representative optimal solutions were selected from each category and included in the comparative analysis, as shown in Table 4. All recommended solutions were configuration combinations with resolution=0.8mm, verifying the overall conclusion that "lower resolution is better".

[0067] Table 4 Comparison of representative parameter combinations and performance indicators for three types of optimal solutions

[0068] From the above, we can see that (1) The Type-H solution has advantages in detail precision control, but the CoV is too high and it is not robust for applications with strict requirements for point cloud uniformity. (2) Type-U solutions can effectively reduce prediction error fluctuations, but their point cloud density is slightly low, making them unsuitable for complex boundary or surface reconstruction. (3) The average density of the 3Type-B solution and CoV are both in the optimization range, showing good balance and adaptability, and are recommended as the mainstream scanning configuration benchmark.

[0069] Therefore, the specific solution type can be selected based on the actual application scenario.

[0070] 4. Screening based on the Analytic Hierarchy Process (AHP) Based on hierarchical analysis, the top three solutions were selected as the primary scanning parameter configurations, as detailed in Table 5. All these primary configurations were based on a 0.8 mm resolution, and their prediction errors were all below 0.9%. The results are as follows... Figure 5 As shown, the three main scanning parameter configuration schemes achieve a good balance among the indicators and have wide adaptability and ease of operation.

[0071] Table 5. Comparison of Scores and Indicators for the Main Scanning Parameter Configuration Scheme

[0072] 5. Experimental verification Using 12 sets of Type-B solutions from the three types of representative Pareto non-dominated solutions (Type-H, Type-U, and Type-B) as experimental samples, simulation verification and error analysis were carried out.

[0073] For the 12 optimal solutions, the model predicted an average density range of 549-659 points / cm², with CoV values ​​ranging from 0.45 to 0.5, and all prediction errors were below 1%. Among these, 7 solutions had prediction errors controlled below 0.8%, demonstrating high stability. Statistical results are detailed in Table 6.

[0074] Table 6. Statistical Results of Experimental Verification of Multi-Objective Optimal Solution

[0075] As shown above, all scanning parameters under the 0.8mm configuration exhibit superior performance in terms of average density and CoV; the lowest prediction error group (E10) has a rate of 0.79%, indicating that the optimization strategy has strong error compression capabilities. The statistical verification above leads to the conclusion that the selected optimal scanning parameter solution possesses high predictive stability and error controllability, meeting the measurement quality requirements of practical engineering applications.

[0076] This application takes different inputs based on the different requirements of the maximum measurement error, and then reverse-engineers different thresholds to obtain recommended scanning schemes for 3D laser scanning under different accuracy requirements. No human intervention is required, the scanning parameters are more accurate, and the stability of point cloud data quality is guaranteed.

[0077] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0078] The following are device embodiments of the present invention. For details not described in detail, please refer to the corresponding method embodiments described above.

[0079] Figure 6 A schematic diagram of the 3D laser scanner scanning parameter optimization device provided in an embodiment of the present invention is shown. For ease of explanation, only the parts related to the embodiment of the present invention are shown, and are described in detail below: like Figure 6 As shown, the 3D laser scanner scanning parameter optimization device includes: Threshold determination module 21 is used to determine the average density threshold and the coefficient of variation threshold; The model building module 22 is used to construct a multi-objective optimization model with the average density threshold and the coefficient of variation threshold as the constraint boundary conditions, aiming at maximizing the average density and minimizing the coefficient of variation. The model solving module 23 is used to solve the multi-objective optimization model by using the GA-PSO hybrid algorithm combined with the Pareto front screening strategy to obtain the Pareto non-dominated solution set. The parameter output module 24 is used to filter the Pareto non-dominated solution set to obtain the target scanning parameters.

[0080] In one possible implementation, the threshold determination module 21 may include: The error calculation unit is used to acquire the measured and actual values ​​of each sample and calculate the measurement error of each sample; the scanning parameters are different for each sample. The threshold output unit is used to obtain the average density threshold and the coefficient of variation threshold based on the measurement error and scanning parameters of each sample.

[0081] In one possible implementation, the threshold output unit may include: The prediction subunit is used to predict the predicted average density value and the predicted coefficient of variation value of each sample based on the scanning parameters of each sample. The correspondence is established into a sub-unit, which is used to fit the average density-error correspondence and the coefficient of variation-error correspondence based on the measurement error of each sample, the predicted average density value of each sample, and the predicted coefficient of variation value of each sample. The first threshold calculation subunit is used to obtain the average density threshold based on the average density-error correspondence and the maximum measurement error; The second threshold calculation subunit is used to obtain the coefficient of variation threshold based on the correlation between the coefficient of variation and the error and the maximum measurement error.

[0082] In one possible implementation, the first threshold calculation subunit can be specifically used to: substitute the maximum measurement error into the average density-error correspondence to obtain the average density threshold. The average density-error correspondence can include:

[0083] in, The measurement error is represented by ρ, which is the average density. , , These are the fitting coefficients; The second threshold calculation subunit can be specifically used to: substitute the maximum measurement error into the coefficient of variation-error correspondence to obtain the coefficient of variation threshold; The coefficient of variation-error correspondence can include:

[0084] in, The coefficient of variation is 1. , , represents the fitting coefficient.

[0085] In one possible implementation, the parameter output module may include: The first screening unit is used to classify and screen each solution in the Pareto non-dominated solution set according to the preset performance classification criteria to obtain three types of target solution sets; The second filtering unit is used to select one of the three target solution sets as the target solution set according to the application scenario; The scoring unit is used to determine the performance score of each solution in the target solution set based on the analytic hierarchy process (AHP). The third filtering unit is used to sort the solutions from largest to smallest according to their performance scores, and select the first preset number of solutions as target scanning parameters.

[0086] In one possible implementation, the scoring unit can be specifically used for: 1. For any target scanning parameter, predict the predicted average density value and the predicted coefficient of variation value of the target scanning parameter; calculate the prediction error of the target scanning parameter based on the predicted average density value and the correspondence between average density and error; and obtain the performance score of the target scanning parameter based on the predicted average density value, the predicted coefficient of variation value, and the prediction error.

[0087] In one possible implementation, the performance score of the target scanning parameter is obtained based on the predicted average density value, predicted coefficient of variation value, and prediction error, which may include: (1) Based on the predicted average density value, predicted coefficient of variation value and prediction error of the target scanning parameters, the performance score of the target scanning parameters is obtained by combining the first formula; The first formula may include:

[0088] in, The performance score for the scanning parameters of the i-th target; , , These are the predicted average density value, predicted coefficient of variation value, and prediction error of the scanning parameters for the i-th target, respectively. and To predict the minimum and maximum values ​​of the average density; and To predict the minimum and maximum values ​​of the coefficient of variation; and These represent the minimum and maximum values ​​of the prediction error; , and These are the weighting coefficients.

[0089] In one possible implementation, the fitness function of the GA-PSO hybrid algorithm It can be:

[0090]

[0091]

[0092] in, and These are the weighting coefficients. and These are the normalized density score and the normalized coefficient of variation penalty, respectively. This represents the actual density of the current point cloud. and These represent the maximum and minimum point cloud densities, respectively. The coefficient of variation of the current point cloud. This represents the maximum value of the coefficient of variation.

[0093] Figure 7 This is a schematic diagram of the electronic device 3 provided in an embodiment of the present invention. Figure 7 As shown, the electronic device 3 of this embodiment includes a processor 30 and a memory 31. The memory 31 stores a computer program 32. When the processor 30 executes the computer program 32, it implements the steps in the various method embodiments described above. Alternatively, when the processor 30 executes the computer program 32, it implements the functions of each module / unit in the various device embodiments described above.

[0094] For example, computer program 32 may be divided into one or more modules / units, which are stored in memory 31 and executed by processor 30 to complete the present invention. The one or more modules / units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of computer program 32 in electronic device 3.

[0095] Electronic device 3 may include, but is not limited to, processor 30 and memory 31. Those skilled in the art will understand that... Figure 7 This is merely an example of electronic device 3 and does not constitute a limitation on electronic device 3. It may include more or fewer components than shown, or combine certain components, or different components. For example, electronic device 3 may also include input / output devices, network access devices, buses, etc.

[0096] The processor 30 can be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor.

[0097] The memory 31 can be an internal storage unit of the electronic device 3, such as a hard disk or memory of the electronic device 3. The memory 31 can also be an external storage device of the electronic device 3, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the electronic device 3. Furthermore, the memory 31 can include both internal and external storage units of the electronic device 3. The memory 31 is used to store the computer program 32 and other programs and data required by the electronic device 3. The memory 31 can also be used to temporarily store data that has been output or will be output.

[0098] For the sake of simplicity and clarity, only the above-described functional modules / units are used as examples. In practical applications, the functions described above can be assigned to different functional modules / units as needed. These modules / units can be implemented in hardware, software, or a combination of both.

[0099] This invention also provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the methods described in the above-described method embodiments.

[0100] This invention also provides a computer program product, including a computer program. When the computer program is executed by a processor, it implements the methods described in the above-described method embodiments.

[0101] Computer programs include computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. Computer-readable media can include: any entity or device capable of carrying computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc.

[0102] In the above embodiments, the descriptions of each embodiment have their own emphasis. Parts not detailed or described in a particular embodiment can be referred to in the relevant descriptions of other embodiments. Unless otherwise specified or in conflict with logic, the terminology and / or descriptions between different embodiments are consistent and can be referenced interchangeably. Technical features in different embodiments can be combined to form new embodiments based on their inherent logical relationships.

[0103] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A method for optimizing scanning parameters of a 3D laser scanner, characterized in that, include: Determine the average density threshold and the coefficient of variation threshold; Using the average density threshold and the coefficient of variation threshold as constraint boundaries, a multi-objective optimization model is constructed with the objectives of maximizing the average density and minimizing the coefficient of variation. The multi-objective optimization model is solved by using the GA-PSO hybrid algorithm combined with the Pareto front screening strategy to obtain the Pareto non-dominated solution set. The Pareto nondominated solution set is filtered based on the application scenario to obtain the target scanning parameters.

2. The 3D laser scanner scanning parameter optimization method according to claim 1, characterized in that, The determination of the average density threshold and the coefficient of variation threshold includes: The measured and actual values ​​of each sample are obtained, and the measurement error of each sample is calculated; the scanning parameters of each sample are different. The average density threshold and the coefficient of variation threshold are obtained based on the measurement error and scanning parameters of each sample.

3. The method for optimizing scanning parameters of a 3D laser scanner according to claim 2, characterized in that, The step of obtaining the average density threshold and the coefficient of variation threshold based on the measurement error and scanning parameters of each sample includes: Based on the scanning parameters of each sample, the predicted average density value and predicted coefficient of variation value of each sample are predicted. Based on the measurement error of each sample, the predicted average density value of each sample, and the predicted coefficient of variation value of each sample, the correspondence between average density and error and the correspondence between coefficient of variation and error are fitted. The average density threshold is obtained based on the average density-error correspondence and the maximum measurement error. The coefficient of variation threshold is obtained based on the coefficient of variation-error correspondence and the maximum measurement error.

4. The 3D laser scanner scanning parameter optimization method according to claim 3, characterized in that, The step of obtaining the average density threshold based on the average density-error correspondence and the maximum measurement error includes: Substituting the maximum measurement error into the average density-error correspondence, the average density threshold is obtained; The average density-error correspondence includes: in, The measurement error is represented by ρ, which is the average density. , , These are the fitting coefficients; The step of obtaining the coefficient of variation threshold based on the coefficient of variation-error correspondence and the maximum measurement error includes: Substituting the maximum measurement error into the coefficient of variation-error correspondence, we obtain the coefficient of variation threshold; The coefficient of variation-error correspondence includes: in, The coefficient of variation is 1. , , represents the fitting coefficient.

5. The method for optimizing scanning parameters of a 3D laser scanner according to any one of claims 1 to 3, characterized in that, The process of filtering the Pareto non-dominated solution set based on the application scenario to obtain the target scanning parameters includes: According to the preset performance classification criteria, each solution in the Pareto non-dominated solution set is classified and screened to obtain three types of target solution sets; Based on the application scenario, one of the three target solution sets is selected as the target solution set; The performance score of each solution in the target solution set is determined based on the analytic hierarchy process (AHP). The solutions are sorted from largest to smallest according to the performance score, and the first preset number of solutions are selected as the target scanning parameters.

6. The 3D laser scanner scanning parameter optimization method according to claim 5, characterized in that, The method of determining the performance score of each solution in the target solution set based on the analytic hierarchy process includes: For any target scanning parameter, the predicted average density value and the predicted coefficient of variation value of the target scanning parameter are predicted; based on the predicted average density value of the target scanning parameter and the correspondence between the average density and the error, the prediction error of the target scanning parameter is calculated; based on the predicted average density value, the predicted coefficient of variation value, and the prediction error of the target scanning parameter, the performance score of the target scanning parameter is obtained.

7. The 3D laser scanner scanning parameter optimization method according to claim 6, characterized in that, The process of obtaining a performance score for the target scanning parameters based on the predicted average density value, predicted coefficient of variation value, and prediction error includes: Based on the predicted average density value, predicted coefficient of variation value and prediction error of the target scanning parameters, the performance score of the target scanning parameters is obtained by combining the first formula; The first formula includes: in, The performance score for the scanning parameters of the i-th target; , , These are the predicted average density value, predicted coefficient of variation value, and prediction error of the scanning parameters for the i-th target, respectively. and To predict the minimum and maximum values ​​of the average density; and To predict the minimum and maximum values ​​of the coefficient of variation; and These represent the minimum and maximum values ​​of the prediction error; , and These are the weighting coefficients.

8. The method for optimizing scanning parameters of a 3D laser scanner according to claim 5, characterized in that, The fitness function of the GA-PSO hybrid algorithm for: in, and These are the weighting coefficients. and These are the normalized density score and the normalized coefficient of variation penalty, respectively. This represents the actual density of the current point cloud. and These represent the maximum and minimum point cloud densities, respectively. The coefficient of variation of the current point cloud. This represents the maximum value of the coefficient of variation.

9. A 3D laser scanner scanning parameter optimization device, characterized in that, include: The threshold determination module is used to determine the average density threshold and the coefficient of variation threshold. The model building module is used to construct a multi-objective optimization model with the average density threshold and the coefficient of variation threshold as constraint boundary conditions, aiming at maximizing the average density and minimizing the coefficient of variation. The model solving module is used to solve the multi-objective optimization model by employing the GA-PSO hybrid algorithm combined with the Pareto front screening strategy to obtain the Pareto non-dominated solution set. The parameter output module is used to filter the Pareto non-dominated solution set based on the application scenario to obtain the target scanning parameters.

10. An electronic device, characterized in that, It includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the 3D laser scanner scanning parameter optimization method as described in any one of claims 1 to 8.

Images