A 3D Measurement Viewpoint Planning Method for Aircraft Overall Shape Inspection

Abstract

This invention relates to the field of aviation digital measurement technology, solving the technical problems of low quality, excessive redundancy, and insufficient global optimization capabilities in existing 3D measurement viewpoint generation technologies. Specifically, it relates to a 3D measurement viewpoint planning method for aircraft overall shape inspection. The method includes discretizing the aircraft point cloud to generate an initial candidate viewpoint set; constructing a comprehensive evaluation function for viewpoint observation quality to perform multi-dimensional comprehensive evaluation of the initial candidate viewpoint set; using an adaptive non-maximum suppression algorithm to filter the initial candidate viewpoint set; and employing an improved alternating evolutionary genetic algorithm to perform multi-objective optimization of the viewpoint set, ultimately obtaining an optimal viewpoint set that combines coverage integrity and superior quality. This invention balances observation quality and viewpoint compactness while ensuring coverage, significantly improving the automation and efficiency of viewpoint generation, and facilitating its application to actual measurement tasks on different aircraft models and engineering sites.

Description

Technical Field

[0001] This invention relates to the field of aviation digital measurement technology, and in particular to a three-dimensional measurement viewpoint planning method for the inspection of the overall shape of an aircraft. Background Technology

[0002] As the core product of the modern aviation industry, the dimensions and geometric accuracy of aircraft directly affect the assembly accuracy, aerodynamic performance, and flight safety of the entire aircraft. Measuring the overall shape of an aircraft not only requires maintaining high measurement accuracy at large scales, but also necessitates full coverage and high-quality observation of various complex surfaces, gaps, and small components. Frequent geometric inspections and deviation checks in the manufacturing, assembly, and maintenance stages make efficient and reliable measurement processes essential.

[0003] Traditional viewpoint planning methods often rely on manual experience, regularized grids, or simple geometric heuristics, revealing several limitations in practical engineering: First, they are inefficient, with manual adjustments being time-consuming and non-repeatable; second, viewpoint redundancy is high, increasing data acquisition and processing costs; third, they lack quantification and optimization of observation quality; fourth, they struggle to adaptively adjust to different camera models and complex local shapes, exhibiting poor scalability; and fifth, they are difficult to balance between global optimization and local quality, easily falling into suboptimal solutions. Therefore, how to ensure coverage while simultaneously considering observation quality, suppressing redundancy, and performing global multi-objective optimization remains a key bottleneck in practical applications. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a three-dimensional measurement viewpoint planning method for aircraft overall shape inspection, which solves the technical problems of low quality, excessive redundant viewpoints, and insufficient global optimization capabilities in the generation of three-dimensional measurement viewpoints in existing technologies.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a three-dimensional measurement viewpoint planning method for aircraft overall shape inspection, the method comprising the following steps:

[0006] S1. Obtain the CAD model of the aircraft's overall shape and preprocess the CAD model to generate an initial candidate viewpoint set that is offset along the normal direction under the sensor's depth constraint.

[0007] S2. Construct a comprehensive evaluation function for the observation quality of viewpoints in the initial candidate viewpoint set to evaluate them in multiple dimensions, and calculate the comprehensive observation quality score of all viewpoints based on the comprehensive evaluation function for the observation quality of viewpoints.

[0008] S3. Based on the comprehensive observation quality score, the initial candidate viewpoint set is screened using the adaptive nonmaximum suppression algorithm to obtain a high-quality viewpoint subset with reasonable spatial distribution.

[0009] S4. An improved alternating evolutionary genetic algorithm is used to perform multi-objective optimization on the viewpoint subset, alternately minimizing the number of viewpoints and maximizing the viewpoint quality, and finally obtaining the optimal viewpoint set that has both coverage integrity and superior quality.

[0010] S5. Arrange measurement viewpoints according to the optimal viewpoint set, and perform shape measurement on the entire aircraft.

[0011] This invention achieves efficient selection and balanced optimization of viewpoint sets by introducing quality-driven comprehensive evaluation, redundancy suppression mechanism and global evolutionary optimization; and significantly reduces viewpoint redundancy while ensuring measurement coverage, thereby improving the compactness and reliability of planning results.

[0012] This invention integrates multiple techniques such as quality modeling, heuristic screening, and multi-objective evolutionary optimization, effectively improving the automation and intelligence level of measurement viewpoint planning.

[0013] Furthermore, in step S1, the specific process includes the following steps:

[0014] S11. Discretely sample the acquired CAD model to obtain the aircraft point cloud P, and divide the aircraft point cloud P into... There are several clusters, namely:

[0015] ;

[0016] in, The first A cluster of aircraft point clouds;

[0017] S12. Calculate the centroid of any cluster as its cluster center, that is:

[0018] ;

[0019] in, For the first The centroids of each cluster; The clustering is represented as aircraft point clouds. The point in the middle;

[0020] S13. Estimate the normal vector of the cluster center based on the cluster neighborhood point set and normalize it, that is:

[0021] ;

[0022] in, The normal vector of the cluster center; It is a three-dimensional real number space;

[0023] S14, in sensor depth of field Under the constraints of cluster centers Along the normal vector The initial viewpoint is obtained by directional offset, that is:

[0024] ;

[0025] in, This is the initial viewpoint obtained after offsetting;

[0026] S15. Repeat steps S12-S14 to obtain all initial viewpoints. And form the initial candidate viewpoint set, namely:

[0027] ;

[0028] in, This represents the initial candidate viewpoint set composed of the initial viewpoint.

[0029] Furthermore, in step S2, the specific process includes the following steps:

[0030] S21. Define evaluation functions for each viewpoint, including viewpoint measurement coverage, viewpoint measurement incident angle, viewpoint measurement distance, and viewpoint measurement redundancy.

[0031] S22. A weighted strategy is used to form a comprehensive evaluation function for viewpoint observation quality;

[0032] S23. Calculate the comprehensive observation quality score of all viewpoints in the initial candidate viewpoint set based on the comprehensive evaluation function of viewpoint observation quality.

[0033] Furthermore, the definition process of the evaluation function includes:

[0034] Define the viewpoint measurement coverage evaluation function, namely:

[0035] viewpoint The coverage rate is defined as the ratio of the number of visible points at that viewpoint to the total number of points in the input aircraft point cloud, expressed as:

[0036] ;

[0037] in, Indicate viewpoint Coverage rate; From the perspective The number of visible points; Enter the total number of points in the aircraft point cloud;

[0038] Define the viewpoint measurement incident angle evaluation function, namely:

[0039] The average of the squared cosines of the angle between the viewpoint direction and the normal vector of the visible point is used as the angle evaluation index. The angle of incidence is defined as:

[0040] ;

[0041] in, Indicate viewpoint The angle of incidence; From the perspective The set of visible points; Indicate viewpoint Observation direction and visible point The angle between the normal vectors;

[0042] Define the viewpoint measurement distance evaluation function, namely:

[0043] The observation distance term is introduced as a quality index, expressed as an exponential decay function over the distance between the viewpoint and its visible point, defined as follows:

[0044] ;

[0045] in, Indicate viewpoint Its visible point An exponentially decaying function over the distance between them; Indicate viewpoint and visible points The Euclidean distance between them; It is an adjustable parameter used to control the degree of penalty based on distance;

[0046] Define the viewpoint measurement redundancy evaluation function, namely:

[0047] A redundancy suppression term is introduced as one of the evaluation metrics to quantify the visible point set. With all other viewpoints The overlap rate between the visible point sets is defined as follows:

[0048] ;

[0049] in, Represents the set of visible points Other viewpoints The overlap rate between the visible point sets; Representation and viewpoint The corresponding set of visible points.

[0050] Furthermore, in step S22, the expression for the comprehensive evaluation function of viewpoint observation quality is:

[0051] ;

[0052] in, From the perspective The overall observation quality score; , , , These are the weighting coefficients for coverage, angle of incidence, distance, and redundancy metrics.

[0053] Furthermore, in step S3, the specific process includes the following steps:

[0054] S31. Based on the comprehensive observation quality score The viewpoints in the initial candidate viewpoint set are sorted in descending order, and a suppression radius is adaptively assigned to each viewpoint based on its overall observation quality score. ;

[0055] Each viewpoint Suppression radius The calculation formula is:

[0056] ;

[0057] ;

[0058] in , It is a predefined range of suppression radius; It is a normalized fraction; , It is the initial viewpoint set The minimum and maximum values ​​of the overall observation quality score for all viewpoints; It is a constant that avoids a denominator of zero;

[0059] S32. The initial candidate viewpoint set is filtered using an adaptive nonmaximum suppression algorithm to obtain a filtered high-quality viewpoint subset. ;

[0060] That is, candidate viewpoints are checked one by one according to the sorting order. If a candidate viewpoint is within the suppression radius of the selected viewpoint, it is eliminated; otherwise, it is retained.

[0061] Furthermore, in step S32, the specific steps are as follows:

[0062] For each candidate viewpoint after sorting The following judgments are made in sequence:

[0063] If the currently selected set is empty, then the first viewpoint is retained directly, that is: ;

[0064] Otherwise, calculate the current candidate viewpoint. With the selected viewpoint Euclidean distance ,Right now:

[0065] ;

[0066] Determine the current candidate viewpoint Does it fall into a selected viewpoint? The inhibition range S, i.e. Make , Selected viewpoint The suppression radius;

[0067] If the condition is met, suppress the current candidate viewpoint. Not added to the set ;

[0068] If the conditions are not met, the current candidate viewpoint is retained. ,Right now: ;

[0069] Repeat the above judgment process until the number of selected viewpoints reaches a preset threshold, and finally output a filtered subset of high-quality viewpoints with reasonable spatial distribution. .

[0070] Furthermore, in step S4, the specific process includes the following steps:

[0071] S41. Establish a multi-objective optimization function for viewpoint planning, used to optimize viewpoint subsets. The viewpoint set with the fewest number and highest observation quality is identified and used for viewpoint planning.

[0072] The expression for the multi-objective optimization function is:

[0073] ;

[0074] in, Represents the optimal viewpoint set; Indicates the coverage threshold; From the perspective The number of visible points; N is the total number of points in the input aircraft point cloud; From the perspective The overall observation quality score;

[0075] S42. An improved alternating evolutionary genetic algorithm is used to perform global optimization of the multi-objective optimization function to obtain the optimal viewpoint set. .

[0076] By employing the above technical solution, the present invention provides a three-dimensional measurement viewpoint planning method for the inspection of the overall shape of an aircraft, which has at least the following beneficial effects:

[0077] 1. The method proposed in this invention takes into account both observation quality and viewpoint compactness while ensuring coverage, significantly improving the automation and efficiency of viewpoint generation, and making it easy to be promoted to actual measurement tasks of different models and engineering sites.

[0078] 2. This invention can achieve efficient generation and global optimization of measurement viewpoints, solving problems such as unreasonable viewpoint layout and excessive computational burden in traditional methods, and ensuring the reliability and intelligence level of complex target surface measurement.

[0079] 3. The method proposed in this invention realizes intelligent viewpoint planning throughout the entire process from quality modeling and redundancy suppression to global optimization, which improves the automation and intelligence level of measurement viewpoint planning, thereby effectively improving the compactness, rationality and execution efficiency of the aircraft's overall measurement viewpoint. Attached Figure Description

[0080] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0081] Figure 1 This is a flowchart of the three-dimensional measurement viewpoint planning method in this invention;

[0082] Figure 2 This diagram illustrates the four key influencing factors in the comprehensive evaluation function of viewpoint observation quality in the method proposed in this invention.

[0083] Figure 3 This is a schematic diagram of the adaptive nonmaximum suppression algorithm in the method proposed in this invention. Detailed Implementation

[0084] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. This will allow for a full understanding of how the present application uses technical means to solve technical problems and achieve technical effects, and to facilitate its implementation.

[0085] This embodiment proposes a 3D measurement viewpoint planning method for aircraft overall shape inspection. Specifically, it is a multi-objective optimization viewpoint planning method that integrates a comprehensive evaluation function of viewpoint observation quality, an adaptive non-maximum suppression algorithm, and an improved alternating evolutionary genetic algorithm. This method performs quality-driven screening of the candidate viewpoint set, effectively removes redundant viewpoints using the adaptive non-maximum suppression algorithm, and achieves a balance between measurement coverage, observation quality, and the number of viewpoints within a multi-objective optimization framework, thereby obtaining the optimal measurement viewpoint set. Through these technical means, this invention can achieve efficient generation and global optimization of measurement viewpoints, solving problems such as unreasonable viewpoint layout and excessive computational burden in traditional methods, ensuring the reliability and intelligence level of complex target surface measurement. Figure 1 As shown, the method includes the following steps:

[0086] S1. Obtain the CAD model of the aircraft's overall shape and preprocess the CAD model to generate an initial candidate viewpoint set offset along the normal direction under sensor depth constraints. In this embodiment, the CAD model is discretely sampled and converted into an aircraft point cloud P. Then, region clustering is performed on the aircraft point cloud P. The initial candidate viewpoint set is obtained by calculating the distance of each cluster center from the sensor depth of field along its normal. The specific process includes the following steps:

[0087] S11. Discretely sample the acquired CAD model to obtain the aircraft point cloud P, and divide the aircraft point cloud P into... There are several clusters, namely:

[0088] ;

[0089] in, The first A cluster of aircraft point clouds;

[0090] S12. Calculate the centroid of any cluster as its cluster center, that is:

[0091] ;

[0092] in, For the first The centroids of each cluster; The clustering is represented as aircraft point clouds. The point in the middle;

[0093] S13. Estimate the normal vector of the cluster center based on the cluster neighborhood point set and normalize it, that is:

[0094] ;

[0095] in, The normal vector of the cluster center; It is a three-dimensional real number space;

[0096] S14, in sensor depth of field Under the constraints of cluster centers Along the normal vector The initial viewpoint is obtained by directional offset, that is:

[0097] ;

[0098] in, This is the initial viewpoint obtained after offsetting;

[0099] S15. Repeat steps S12-S14 to obtain all initial viewpoints. And form the initial candidate viewpoint set, namely:

[0100] ;

[0101] in, This represents the initial candidate viewpoint set composed of the initial viewpoint.

[0102] S2. Construct a comprehensive evaluation function for viewpoint observation quality to perform multi-dimensional comprehensive evaluation of viewpoints in the initial candidate viewpoint set, and calculate the comprehensive observation quality score for all viewpoints based on the comprehensive evaluation function. In a preferred embodiment, step S2 specifically includes:

[0103] S21. Define evaluation functions for each viewpoint, including viewpoint measurement coverage, viewpoint measurement incident angle, viewpoint measurement distance, and viewpoint measurement redundancy. Please refer to [reference needed]. Figure 2 This embodiment defines a viewpoint measurement coverage evaluation function, a viewpoint measurement incident angle evaluation function, a viewpoint measurement distance evaluation function, and a viewpoint measurement redundancy evaluation function, including:

[0104] (1) Define the viewpoint measurement coverage evaluation function:

[0105] Each viewpoint Each has a corresponding set of visible points. ,Should It is calculated using a ray casting method. Viewpoint The coverage rate is defined as the ratio of the number of visible points at that viewpoint to the total number of points in the input aircraft point cloud, i.e.:

[0106] ;

[0107] in, Indicate viewpoint Coverage rate; From the perspective The number of visible points; Enter the total number of points in the aircraft point cloud.

[0108] (2) Define the evaluation function for the incident angle of viewpoint measurement:

[0109] Each viewpoint Its visible point set a certain point in the middle The surface normal forms a certain angle of incidence. To evaluate the quality of the observation angle from the viewpoint, the average of the squared cosines of the angle between the viewpoint direction and the visible point normal vector is used as the angle evaluation index. Specifically, the viewpoint... The angle of incidence is defined as:

[0110] ;

[0111] in, Indicate viewpoint The angle of incidence; Indicate viewpoint Observation direction and visible point The angle between the normal vectors. A smaller angle of incidence is more conducive to obtaining high-quality measurement data.

[0112] (3) Definition of viewpoint measurement distance evaluation function: In order to evaluate the spatial suitability of the viewpoint relative to the surface to be observed, the observation distance term is introduced as a quality index. This term is expressed as an exponential decay function of the distance between the viewpoint and its visible point, and is defined as follows:

[0113] ;

[0114] in, Indicate viewpoint Its visible point An exponentially decaying function over the distance between them; Indicate viewpoint and visible points The Euclidean distance between them; This is an adjustable parameter used to control the degree of penalty based on distance. This index reflects the impact of observation distance; the closer the viewpoint, the higher the value, and the higher the measurement accuracy.

[0115] (4) Define the viewpoint measurement redundancy evaluation function: for each viewpoint Each has a corresponding set of visible points. This point set may overlap with the visible regions of other viewpoints. To suppress redundant measurements and improve observation efficiency, a redundancy suppression term is introduced as one of the evaluation metrics. This term quantifies the visible point set. With all other viewpoints The overlap rate between the visible point sets is defined as follows:

[0116] ;

[0117] in, Represents the set of visible points Other viewpoints The overlap rate between the visible point sets; Representation and viewpoint The corresponding set of visible points.

[0118] S22. A weighted strategy is used to form a comprehensive evaluation function for viewpoint observation quality. The above content systematically evaluates the observation quality of each viewpoint from four aspects: coverage, incident angle, measurement distance, and redundancy. However, since a single index cannot fully characterize the viewpoint observation quality, a weighted fusion strategy is used to construct a comprehensive evaluation function for viewpoint observation quality, balancing the contributions of each factor and enhancing the rationality and applicability of viewpoint selection. The comprehensive evaluation function for viewpoint observation quality is defined as follows:

[0119] ;

[0120] in, From the perspective The overall observation quality score; , , , These are the weighting coefficients corresponding to coverage, angle of incidence, distance, and redundancy indicators, thus achieving multi-dimensional quantification of viewpoint quality.

[0121] S23. Calculate the comprehensive observation quality score for all viewpoints in the initial candidate viewpoint set based on the viewpoint observation quality comprehensive evaluation function. Calculate the comprehensive observation quality score for each viewpoint in the initial candidate viewpoint set. The evaluated comprehensive observation quality score will be used for subsequent spatial filtering and the final optimization objective.

[0122] S3. Based on the comprehensive observation quality score, the initial candidate viewpoint set is filtered using an adaptive nonmaximum suppression algorithm to obtain a high-quality viewpoint subset with a reasonable spatial distribution. To ensure high-quality selection and uniform spatial distribution of viewpoints, a comprehensive observation quality score is calculated for all viewpoints. Then, an adaptive non-maximum suppression algorithm is used. First, the suppression radius of each viewpoint in the initial candidate viewpoint set is calculated. Second, the adaptive non-maximum suppression algorithm is used to filter the initial candidate viewpoint set. Finally, a filtered subset of high-quality viewpoints is obtained. In a preferred embodiment, step S3 specifically includes:

[0123] S31. Based on the comprehensive observation quality score The viewpoints in the initial candidate viewpoint set are sorted in descending order to ensure that high-quality viewpoints are given priority. Based on this, a suppression radius is adaptively assigned to each viewpoint according to its overall observation quality score. In other words, the higher the quality score, the larger the suppression radius, thus making high-quality viewpoints more competitive in spatial distribution. Each viewpoint... Suppression radius The calculation formula is:

[0124] ;

[0125] ;

[0126] in , It is a predefined range of suppression radius; It is a normalized fraction; , It is the initial viewpoint set The minimum and maximum values ​​of the overall observation quality score for all viewpoints; It is a constant that avoids a denominator of zero.

[0127] S32. The initial candidate viewpoint set is filtered using an adaptive nonmaximum suppression algorithm to obtain a filtered high-quality viewpoint subset. That is, candidate viewpoints are checked one by one in the sorted order. If a candidate viewpoint is within the suppression radius of the selected viewpoints, it is discarded; otherwise, it is retained. Please refer to [link / reference]. Figure 3 The specific steps are as follows:

[0128] For each candidate viewpoint after sorting The following judgments are made in sequence:

[0129] If the currently selected set is empty, then the first one (highest overall observation quality score) will be retained, i.e.: ;

[0130] Otherwise, calculate the current candidate viewpoint. With the selected viewpoint Euclidean distance ,Right now:

[0131] ;

[0132] Determine the current candidate viewpoint Does it fall into a selected viewpoint? The inhibition range S, i.e. Make , Selected viewpoint The suppression radius;

[0133] If the condition is met, suppress the current candidate viewpoint. Not added to the set ;

[0134] If the conditions are not met, the current candidate viewpoint is retained. ,Right now: ;

[0135] Repeat the above judgment process until the number of selected viewpoints reaches a preset threshold, and finally output a filtered subset of high-quality viewpoints with reasonable spatial distribution. .

[0136] This greedy selection process ensures that the final retained viewpoints possess both high observation quality and avoid overly dense spatial distribution. Through the above steps, a high-quality subset of viewpoints with excellent quality and reasonable spatial distribution can be obtained, laying the foundation for subsequent visibility analysis and optimization.

[0137] S4. An improved alternating evolutionary genetic algorithm is used for the viewpoint subset. By performing multi-objective optimization, alternately minimizing the number of viewpoints and maximizing viewpoint quality, an optimal set of viewpoints that combines coverage integrity with superior quality is ultimately obtained. This embodiment uses an improved alternating evolutionary genetic algorithm to perform multi-objective optimization to obtain the optimal viewpoint set. As a preferred implementation, step S4 specifically includes:

[0138] S41. Establish a multi-objective optimization function for viewpoint planning, used to optimize viewpoint subsets. The viewpoint set with the fewest number and highest observation quality is identified for viewpoint planning. In this step, the viewpoint planning problem is reformulated as a constrained multi-objective optimization task aimed at selecting from the viewpoint subset filtered by an adaptive nonmaximum suppression algorithm. The set of viewpoints with the fewest number and the highest observation quality was identified.

[0139] The multi-objective optimization function is defined as:

[0140] ;

[0141] in, This represents the optimal viewpoint set, i.e., the subset of viewpoints filtered by the adaptive nonmaximum suppression algorithm. The best set of viewpoints selected from; This represents the coverage threshold, typically set to 95% or 98%. From the perspective The number of visible points; N is the total number of points in the input aircraft point cloud; From the perspective The overall observation quality score;

[0142] S42. An improved alternating evolutionary genetic algorithm is used to perform global optimization of the multi-objective optimization function to obtain the optimal viewpoint set. An improved alternating evolutionary genetic algorithm is used to globally solve the established multi-objective optimization function. Specifically, the candidate viewpoint set is first encoded into a binary vector to generate an initial population; then, in each generation of evolution, the coverage rate of individuals is calculated and the coverage rate constraint is strictly enforced. Solutions that are infeasible are directly deemed invalid. In feasible solutions, Boolean variables are switched... The algorithm alternately optimizes by minimizing the number of viewpoints and maximizing overall quality, constructing a fitness function accordingly. During evolution, selection, crossover, and mutation operations are used to update the population, and an elite retention strategy is employed to maintain the optimal individuals. When a preset number of generations is reached or a coverage threshold is met, the algorithm outputs the final optimal set of viewpoints, achieving a balance between the compactness and quality of the solution under coverage constraints. Specifically:

[0143] (1) Encoding and population initialization;

[0144] Using binary set encoding, an individual is represented as:

[0145] ;

[0146] Among them, the A value of 1 indicates the selection of a viewpoint. Record the number of selected viewpoints as , where m is the total number of individuals.

[0147] Generate the initial population ,in Population size;

[0148] Initialize Elite Pool Global Optimum ;

[0149] Initialize Boolean switches Iteration counting ;

[0150] (2) Assessment of adaptability and feasibility;

[0151] For each individual ;

[0152] Compute Coverage Set ;

[0153] Calculate coverage ;

[0154] Calculate fitness ;

[0155] (3) Main loop iteration;

[0156] In the Execution in proxy, that is:

[0157] ① Selection: Select parent individuals based on fitness;

[0158] ② Crossover: Perform crossover on the parent code to generate the child code;

[0159] ③ Mutation: Flip some bits of the offspring according to probability;

[0160] ④ Evaluation: Recalculate using the formula and ;

[0161] ⑤ Elite Preservation: Preserve the best individual and update the global optimal solution;

[0162] ⑥ Alternate switching: If Then, the target parameters are switched and a portion of the population is initialized, i.e.:

[0163] ;

[0164] in, It is a modular arithmetic. For the interval of alternating evolution, For Boolean switches;

[0165] (3) Termination and output;

[0166] When the termination condition is met, that is:

[0167] ;

[0168] Output the final optimal viewpoint set ;in The maximum number of iterations;

[0169] Thus, the optimal set of viewpoints for aircraft shape measurement was obtained, which simultaneously satisfies the requirements of measurement coverage, minimization of the number of viewpoints, and maximization of observation quality.

[0170] S5. Arrange measurement viewpoints according to the optimal viewpoint set, and perform shape measurement on the entire aircraft.

[0171] This invention employs a strategy combining quality-driven and intelligent optimization. First, a comprehensive evaluation function of viewpoint observation quality is used to score candidate viewpoints across multiple dimensions, comprehensively measuring factors such as visibility, angle of incidence, distance, and redundancy. Then, an adaptive non-maximum suppression algorithm is used to filter the initial candidate set, removing redundancy and maintaining a balanced spatial distribution. Finally, under coverage constraints, an improved alternating evolutionary genetic algorithm is introduced to perform multi-objective global optimization of the viewpoint set, achieving a dynamic balance between minimizing the number of viewpoints and maximizing the overall quality of the viewpoints. The method proposed in this invention realizes intelligent viewpoint planning throughout the entire process, from quality modeling and redundancy suppression to global optimization, effectively improving the compactness, rationality, and execution efficiency of aircraft-wide measurement viewpoints.

[0172] In summary, the method proposed in this invention integrates multiple techniques, including a comprehensive evaluation function for viewpoint observation quality, an adaptive nonmaximum suppression algorithm, and a multi-objective optimization algorithm based on an improved alternating evolutionary genetic algorithm. This effectively improves the rationality and efficiency of viewpoint generation. This method not only reduces the number of viewpoints and improves viewpoint observation quality while ensuring measurement coverage, but also balances global equilibrium with local detail optimization, thereby significantly improving the efficiency and accuracy of digital measurement. It has good engineering applicability and promotional value.

[0173] Those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0174] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. Since the above embodiments are substantially similar to the method embodiments, their descriptions are relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0175] The above embodiments provide a detailed description of the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A three-dimensional measurement viewpoint planning method for aircraft overall shape inspection, characterized in that, The method includes the following steps: S1. Obtain the CAD model of the aircraft's overall shape and preprocess the CAD model to generate an initial candidate viewpoint set that is offset along the normal direction under the sensor's depth constraint. S2. Construct a comprehensive evaluation function for the observation quality of viewpoints in the initial candidate viewpoint set, and calculate the comprehensive observation quality score of all viewpoints based on the comprehensive evaluation function. The specific process includes the following steps: S21. Define evaluation functions for each viewpoint, including viewpoint measurement coverage, viewpoint measurement incident angle, viewpoint measurement distance, and viewpoint measurement redundancy, including: Define the viewpoint measurement coverage evaluation function, namely: viewpoint The coverage rate is defined as the ratio of the number of visible points at that viewpoint to the total number of points in the input aircraft point cloud, expressed as: ； in, Indicate viewpoint Coverage rate; From the perspective The number of visible points; Enter the total number of points in the aircraft point cloud; Define the viewpoint measurement incident angle evaluation function, namely: The average of the squared cosines of the angle between the viewpoint direction and the normal vector of the visible point is used as the angle evaluation index. The angle of incidence is defined as: ； in, Indicate viewpoint The angle of incidence; From the perspective The set of visible points; Indicate viewpoint Observation direction and visible point The angle between the normal vectors; Define the viewpoint measurement distance evaluation function, namely: The observation distance term is introduced as a quality index, expressed as an exponential decay function over the distance between the viewpoint and its visible point, defined as follows: ； in, Indicate viewpoint Its visible point An exponentially decaying function over the distance between them; Indicate viewpoint and visible points The Euclidean distance between them; It is an adjustable parameter used to control the degree of penalty based on distance; Define the viewpoint measurement redundancy evaluation function, namely: A redundancy suppression term is introduced as one of the evaluation metrics to quantify the visible point set. With all other viewpoints The overlap rate between the visible point sets is defined as follows: ； in, Represents the set of visible points Other viewpoints The overlap rate between the visible point sets; Representation and viewpoint The corresponding set of visible points; S22. A weighted strategy is used to form a comprehensive evaluation function for viewpoint observation quality; S23. Calculate the comprehensive observation quality score of all viewpoints in the initial candidate viewpoint set based on the comprehensive evaluation function of viewpoint observation quality. S3. Based on the comprehensive observation quality score, the initial candidate viewpoint set is screened using an adaptive nonmaximum suppression algorithm to obtain a high-quality viewpoint subset with a reasonable spatial distribution. The specific process includes the following steps: S31. Based on the comprehensive observation quality score The viewpoints in the initial candidate viewpoint set are sorted in descending order, and a suppression radius is adaptively assigned to each viewpoint based on its overall observation quality score. ; Each viewpoint Suppression radius The calculation formula is: ； ； in , It is a predefined range of suppression radius; It is a normalized fraction; , It is the initial viewpoint set The minimum and maximum values ​​of the overall observation quality score for all viewpoints; It is a constant that avoids a denominator of zero; S32. The initial candidate viewpoint set is filtered using an adaptive nonmaximum suppression algorithm to obtain a filtered high-quality viewpoint subset. ; That is, candidate viewpoints are checked one by one in the sorting order. If a candidate viewpoint is within the suppression radius of the selected viewpoints, it is eliminated; otherwise, it is retained. S4. An improved alternating evolutionary genetic algorithm is used to perform multi-objective optimization on the viewpoint subset, alternately minimizing the number of viewpoints and maximizing the viewpoint quality, and finally obtaining the optimal viewpoint set that has both coverage integrity and superior quality. S5. Arrange measurement viewpoints according to the optimal viewpoint set, and perform shape measurement on the entire aircraft.

2. The three-dimensional measurement viewpoint planning method according to claim 1, characterized in that, In step S1, the specific process includes the following steps: S11. Discretely sample the acquired CAD model to obtain the aircraft point cloud P, and divide the aircraft point cloud P into... There are several clusters, namely: ； in, The first A cluster of aircraft point clouds; S12. Calculate the centroid of any cluster as its cluster center, that is: ； in, For the first The centroids of each cluster; The clustering is represented as aircraft point clouds. The point in the middle; S13. Estimate the normal vector of the cluster center based on the cluster neighborhood point set and normalize it, that is: ； in, The normal vector of the cluster center; It is a three-dimensional real number space; S14, in sensor depth of field Under the constraints of cluster centers Along the normal vector The initial viewpoint is obtained by directional offset, that is: ； in, This is the initial viewpoint obtained after offsetting; S15. Repeat steps S12-S14 to obtain all initial viewpoints. And form the initial candidate viewpoint set, namely: ； in, This represents the initial candidate viewpoint set composed of the initial viewpoint.

3. The three-dimensional measurement viewpoint planning method according to claim 1, characterized in that, In step S22, the expression for the comprehensive evaluation function of viewpoint observation quality is: ； in, From the perspective The overall observation quality score; , , , These are the weighting coefficients for coverage, angle of incidence, distance, and redundancy metrics.

4. The three-dimensional measurement viewpoint planning method according to claim 1, characterized in that, In step S32, the specific steps are as follows: For each candidate viewpoint after sorting The following judgments are made in sequence: If the currently selected set is empty, then the first viewpoint is retained directly, that is: ; Otherwise, calculate the current candidate viewpoint. With the selected viewpoint Euclidean distance ,Right now: ； Determine the current candidate viewpoint Does it fall into a selected viewpoint? The inhibition range S, i.e. Make , Selected viewpoint The suppression radius; If the condition is met, suppress the current candidate viewpoint. Not added to the set ; If the conditions are not met, the current candidate viewpoint is retained. ,Right now: ; Repeat the above judgment process until the number of selected viewpoints reaches a preset threshold, and finally output a filtered subset of high-quality viewpoints with reasonable spatial distribution. .

5. The three-dimensional measurement viewpoint planning method according to claim 4, characterized in that, In step S4, the specific process includes the following steps: S41. Establish a multi-objective optimization function for viewpoint planning, used to optimize viewpoint subsets. The viewpoint set with the fewest number and highest observation quality is identified and used for viewpoint planning. The expression for the multi-objective optimization function is: ； in, Represents the optimal viewpoint set; Indicates the coverage threshold; From the perspective The number of visible points; N is the total number of points in the input aircraft point cloud; From the perspective The overall observation quality score; S42. An improved alternating evolutionary genetic algorithm is used to perform global optimization of the multi-objective optimization function to obtain the optimal viewpoint set. .

Images