A method and system for measuring feather arch based on 3D point cloud

Abstract

This invention discloses a method and system for measuring feather camber based on three-dimensional point clouds, belonging to the field of feather camber measurement technology. This invention overcomes dimensional limitations, achieving accurate quantification of true three-dimensional deformation: existing technologies rely on two-dimensional image projection, which cannot obtain the true curvature of the feather in space (Z-axis camber) and is easily affected by placement posture. This invention, through three-dimensional point cloud acquisition and principal component analysis (PCA), constructs a local coordinate system based on the feather using projection parameters, eliminating the interference of physical placement posture on the measurement results and accurately restoring the true spatial curvature of the feather shaft.

Description

Technical Field

[0001] This invention relates to the field of feather arch measurement technology, and in particular to a method and system for measuring feather arch based on three-dimensional point clouds. Background Technology

[0002] Badminton is a sport that demands extremely high stability from the equipment. The trajectory, spin, and landing accuracy of a badminton shuttlecock primarily depend on the aerodynamic characteristics of the skirt-like structure composed of 16 feathers. Among these, the arch of a single feather in the Z-axis direction (perpendicular to the feather's plane) is a core parameter determining the catch area and drag coefficient. Excessive or insufficient arch, or excessive dispersion in the arch of the same feather group, will cause the shuttlecock to exhibit "snake-like" wobbling, unstable spin, or yaw in the air, severely impacting the quality of the match.

[0003] Currently, the industry's testing of feather arch has mainly gone through two stages, but both have significant limitations:

[0004] I. Manual Visual Inspection Stage

[0005] Traditional feather sorting relies heavily on the personal experience of skilled workers. Workers must hold feathers under strong light, visually inspecting the curvature of the feather shafts and combining this with tactile feedback to grade them. This method has the following inherent drawbacks:

[0006] 1. High subjectivity: The lack of quantitative standards makes it easy for judgment standards to drift between different workers and at different times, resulting in extremely poor quality consistency of the same batch of products.

[0007] 2. Low efficiency: The slow speed of manual sorting has become a bottleneck process restricting the automation of badminton production lines.

[0008] 3. Physiological limits: Prolonged high-intensity visual work can easily lead to visual fatigue, resulting in high rates of missed detections and false detections.

[0009] II. Detection Stage Based on 2D Machine Vision

[0010] With the development of industrial automation, existing technologies attempt to introduce industrial cameras for non-contact measurement. A typical solution, such as Chinese invention patent CN104697458B, uses horizontal and vertical dual cameras working together to obtain the camber value by extracting the feather outline and calculating its deviation from a reference straight line. However, such existing technologies still have the following technical problems that urgently need to be solved in practical applications:

[0011] 1. Measurement distortion due to missing dimensions: Feathers are flexible organisms with complex spatial curvature. Existing two-dimensional vision technology essentially projects three-dimensional objects onto a two-dimensional plane, only acquiring two-dimensional projection information of the feather from a specific viewpoint. When the feather undergoes slight flipping, tilting, or non-planar torsion during transmission, the two-dimensional projection cannot reproduce its true three-dimensional curvature, resulting in a systematic deviation between the measurement data and the actual physical shape.

[0012] 2. Insufficient robustness due to the non-standard nature of biological materials: Feathers are natural non-standard materials, varying in shape and complex in surface texture. Existing algorithms are mostly based on fixed template matching or simple edge extraction, which are easily affected by feather surface texture, damage, stains, and changes in ambient light. In addition, the flexible deformation of feathers under airflow or mechanical vibration can also cause the center line of the feather shaft extracted by traditional algorithms to break or shift, resulting in misjudgment.

[0013] 3. Lack of closed-loop feedback mechanism: Existing testing is mostly an isolated screening process, which fails to link high-precision three-dimensional arch data with subsequent processes such as feather implantation and glue injection, and thus cannot achieve adaptive production adjustments based on feather characteristics.

[0014] In summary, existing feather arch measurement technologies are either limited by the subjective inefficiency of manual measurement or by the dimensional limitations and algorithmic fragility of two-dimensional vision, failing to meet the production demands of modern high-end badminton manufacturing for high precision, high consistency, and high robustness. Summary of the Invention

[0015] To address the technical problems existing in the measurement of feather arch, this invention provides a method and system for measuring feather arch based on three-dimensional point clouds. By using three-dimensional point clouds and layered culling, the dimensional error of two-dimensional vision is eliminated, and the measurement accuracy can reach the micrometer level.

[0016] This invention protects a method for measuring feather arch based on three-dimensional point clouds, comprising the following steps: Step S1: Obtain the three-dimensional point cloud data of the feather; Step S2: Determine the principal axis direction of the feather, perform principal axis projection on the 3D point cloud, and obtain the one-dimensional projection parameters along the principal axis direction; Step S3: Based on the one-dimensional projection parameters, the three-dimensional point cloud is divided into root point set and tip point set; Step S4: Perform edge removal processing on the root point set to obtain the root center region point set; Step S5: Fit the reference plane using the point set of the root center region; Step S6: Calculate the signed distance from each point in the cusp set to the reference plane; Step S7: Take the median of the signed distances as the camber measurement value.

[0017] In a preferred embodiment of the present invention, the principal component analysis method is used to determine the principal axis direction of the feather in step S2: the covariance matrix of the three-dimensional point cloud is calculated, the covariance matrix is ​​decomposed into eigenvalues, and the eigenvector corresponding to the largest eigenvalue is taken as the principal axis direction.

[0018] In a preferred embodiment of the present invention, step S3 further includes the following steps: Step S31: Calculate the cumulative distribution function CDF based on the one-dimensional projection parameters obtained in step S2, and take the low quantile interval of the cumulative distribution as the first region and the high quantile interval as the second region; wherein, the root point set preferably corresponds to the low quantile interval of the one-dimensional projection parameters from 0% to 30%, and the apex point set preferably corresponds to the high quantile interval of the one-dimensional projection parameters from 70% to 100%.

[0019] Step S32: Based on the prior knowledge of feather morphology (e.g., larger root cross-sectional area, concentrated point cloud distribution), determine the semantic labels of the first region and the second region as either root point set or tip point set.

[0020] As a preferred embodiment of the present invention, step S4 further includes the following steps: step S41, dividing the root point set into multiple parallel layers along the principal axis direction; step S42, for each parallel layer, calculating the projection distribution of points within that layer in the direction perpendicular to the principal axis; step S43, determining the effective boundary using a strategy based on quantiles or based on local extremum shrinkage, and removing points outside the boundary, wherein the shrinkage amplitude is a fixed number of pixels, a fixed ratio, or an adaptive value based on the point density within the layer.

[0021] In a preferred embodiment of the present invention, the fitting reference plane in step S5 adopts one of the following strategies and is configured with an automatic backoff mechanism: Strategy A, RANSAC plane fitting method: randomly sample three points to form a plane model, iteratively find the maximum set of interior points, and finally refine it using the least squares method; Strategy B, Histogram Voting Method: Construct histograms for the slope and intercept parameters of the plane equation respectively, and determine the optimal parameters through Gaussian smoothing and sub-pixel peak extraction; Strategy C, RANSAC with local back-substitution plane fitting method: First, fit the root point with a quadratic surface, then project the root point onto the surface, and finally fit the projected point with the least squares method to obtain the plane. When the current strategy fails to fit, it automatically reverts to the next strategy and continues execution. The criteria for determining failure to fit include: In Strategy A, the proportion of the largest interior set is lower than the preset threshold, or the iteration count is exhausted and the convergence is not achieved. In Strategy B, the histogram shows a multi-peak distribution with no significant main peak, or the peak confidence is lower than the preset threshold. In Strategy C, the residual of the quadratic surface fitting is greater than the preset threshold, or the projected point set cannot form an effective plane.

[0022] As a preferred embodiment of the present invention, step 7 further includes: before calculating the signed distance, performing outlier filtering on the tip point set to remove abnormal distance values ​​caused by hair curling or scanning noise.

[0023] This invention also protects a feather arch measurement system based on three-dimensional point clouds, comprising: The point cloud acquisition module is used to acquire the three-dimensional point cloud data of the feather; The principal axis calculation module is used to perform principal axis analysis on the 3D point cloud to obtain the one-dimensional projection parameters and principal axis direction vector in the principal axis direction. The layering module is used to divide a 3D point cloud into a root point set and a apex point set based on one-dimensional projection parameters. The edge culling module is used to perform edge culling on the root point set to obtain the root center region point set; The plane fitting module is used to fit a reference plane using the point set of the root center region; The camber calculation module is used to calculate the signed distance from each point in the top point set to the reference plane, and take the median of the signed distances as the camber measurement value.

[0024] The present invention also protects a computer-readable storage medium having a computer program stored thereon, characterized in that the computer program, when executed by a processor, implements the above-described feather arch measurement method.

[0025] Beneficial effects of this invention:

[0026] 1. Breaking through dimensional limitations to achieve precise quantification of true 3D deformation: Existing technologies rely on 2D image projection, which cannot obtain the true curvature (Z-axis camber) of feathers in space and is easily affected by placement posture. This invention, through 3D point cloud acquisition and principal component analysis (PCA), constructs a local coordinate system based on the feather using projection parameters, eliminating the interference of physical placement posture on measurement results and accurately restoring the true spatial curvature of the feather shaft.

[0027] 2. A pioneering edge culling mechanism combining axial slicing and adaptive shrinkage significantly improves baseline stability: Addressing the issues of laser scattering noise and microscopic barbs at the edges of point clouds at the feather root, this invention innovatively proposes an edge culling strategy based on layered slicing. Unlike traditional global filtering, this scheme independently shrinks the boundary within each micro-element layer and adaptively adjusts it in conjunction with point cloud density. This design can precisely peel away interference points like "peeling an onion," preserving the core geometric features of the feather shaft, laying a solid foundation for subsequent fitting of a high-precision reference plane.

[0028] 3. Constructing a robust defense system of "multi-level strategy + automatic backoff" to ensure zero downtime in industrial environments: Production line environments are complex and variable (dust, vibration, glare). This invention overcomes the vulnerability of single algorithms by designing a three-level defense strategy: RANSAC plane fitting, histogram voting, and surface projection correction. When the primary strategy fails due to excessive noise, the system automatically and seamlessly switches to the backup strategy. In particular, the introduced "surface RANSAC + local back-substitution" (strategy C) is the first to acknowledge and compensate for the naturally occurring micro-arcs at the base of the feather, incorporating the physical deformation of biological materials into the algorithm's considerations, greatly improving the system's survivability and measurement consistency in harsh industrial environments.

[0029] 4. Statistically based "median camber" extraction effectively filters out non-rigid interference: When calculating camber, this invention uses the median of signed distances as the final result, rather than the average. This design cleverly utilizes statistical principles, naturally immunizing against interference from extreme outliers such as slight curling of feather tufts or scanning flight points. Regardless of individual feather differences, the system always outputs a stable value representing the main bending trend of the feather quill, ensuring a high degree of consistency in camber grading within the same batch of feathers.

[0030] 5. Scale Invariance and Adaptive Partitioning for Fully Automated Unmanned Production: By introducing the cumulative distribution function (CDF) for quantile threshold segmentation and combining it with the design of transition buffers, this invention can automatically adapt to the measurement needs of different breeds (goose / duck feathers) and different sizes of feathers, eliminating the need for frequent manual parameter adjustments. Combined with an intelligent discard mechanism in the transition range (e.g., 30%~70%), it effectively avoids the ambiguous zone at the root tip junction, truly achieving a leap from "human visual inspection" to "fully adaptive intelligent machine inspection." Attached Figure Description

[0031] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0032] Figure 1 This is a schematic diagram of the feather arch measurement method based on 3D point cloud. Detailed Implementation

[0033] To make the objectives, technical solutions, and advantages of the present invention more apparent, exemplary embodiments according to the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are merely a part of the embodiments of the present invention, and not all of the embodiments of the present invention. It should be understood that the present invention is not limited to the exemplary embodiments described herein. Based on the embodiments of the present invention described herein, all other embodiments obtained by those skilled in the art without inventive effort should fall within the protection scope of the present invention.

[0034] In the following description, numerous specific details are set forth in order to provide a more thorough understanding of the invention. However, it will be apparent to those skilled in the art that the invention can be practiced without one or more of these details. In other instances, certain technical features well-known in the art have not been described in order to avoid obscuring the invention.

[0035] It should be understood that the invention can be embodied in various forms and should not be construed as being limited to the embodiments set forth herein. Rather, providing these embodiments will make the disclosure thorough and complete, and will fully convey the scope of the invention to those skilled in the art.

[0036] To fully understand this invention, a detailed structure will be presented in the following description to illustrate the technical solution proposed by this invention. Optional embodiments of the invention are described in detail below; however, in addition to these detailed descriptions, the invention may have other embodiments.

[0037] The following detailed description of some embodiments of the present invention is provided in conjunction with the accompanying drawings. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0038] This embodiment discloses a method for measuring feather arch based on three-dimensional point clouds. See [link to relevant documentation]. Figure 1 The aim is to solve the problems of poor consistency and inability to accurately measure three-dimensional deformation caused by reliance on manual visual inspection or two-dimensional vision in existing technologies.

[0039] 1. Obtain 3D point cloud data of feathers

[0040] Three-dimensional point clouds of feathers are acquired using a line laser sensor or a structured light camera. To reduce computational load, the raw point cloud can be preprocessed, such as by removing background noise and downsampling.

[0041] 2. Main axis projection

[0042] Calculate the covariance matrix of the point cloud and perform eigenvalue decomposition. Take the eigenvector corresponding to the largest eigenvalue as the principal axis direction v. Orthogonally project the 3D point cloud onto this principal axis to obtain the one-dimensional projection parameter t.

[0043] 3. Distinguish between the root and the tip.

[0044] Here, a proportional truncation algorithm based on "cumulative projection length" can be used. The specific implementation steps are as follows:

[0045] ① Calculate the total length: Obtain the maximum and minimum values ​​of the one-dimensional projection, the total length L = t max -t min .

[0046] ② Set an empirical ratio: Based on a large number of experimental statistics, the base of a badminton feather usually accounts for about 25% to 35% of the total length. Let's assume a ratio coefficient of α = 0.3.

[0047] ③ Directly calculate the tangent point: the dividing point t cut = t min +α×L.

[0048] ④ Execution classification: All projection parameters t < t cut The points are classified into the root point set; all points t ≥ t cut The points are classified into the cusp point set.

[0049] The advantage of the above algorithm is that it is extremely fast, requiring only a few subtractions and multiplications; the disadvantage is that if the production line changes to different specifications of feathers, or if a batch of feathers has an unusual growth pattern (particularly long roots), the fixed proportion will lead to misalignment of the cutting.

[0050] Therefore, this embodiment proposes a partitioning method based on CDF to calculate the cumulative distribution function (CDF) of the one-dimensional projection parameter t.

[0051] Preferred scheme: Set the low quantile to 30% and the high quantile to 70%. Define the region with t≤30% as the root point set, the region with t≥65% as the apical point set, and exclude the region with 30%~70% as the transition interval, which has a complex point cloud morphology (transition from hair stalk to hair leaf) to prevent interference with subsequent calculations.

[0052] Based on the morphological prior of feathers having "large root cross-sectional area and concentrated point cloud distribution", this embodiment also constructs an adaptive semantic determination mechanism: the system calculates the density characteristics or cross-sectional geometric dimensions of the point cloud in the low quantile interval (e.g., 0%~30%) and the high quantile interval (e.g., 70%~100%) respectively, and automatically determines the area with denser point cloud and larger cross-sectional area as the root point set, and the area with sparse point cloud and smaller cross-sectional area as the tip point set; this mechanism can effectively eliminate the problem of inconsistent physical position and projection parameter order caused by the ambiguity of point cloud orientation flipping or principal axis direction, and can achieve accurate partitioning of feathers in any posture without manual intervention.

[0053] 4. Layered edge removal

[0054] A statistical elimination algorithm based on "one-dimensional projection distance" can be used here. The specific implementation steps are as follows:

[0055] ① Calculate the one-dimensional centroid: Extract the current "root point set", calculate the average (or median) of their projection parameters on the principal axes, and define it as the one-dimensional geometric center t of the root. center .

[0056] ② Calculate the dispersion: Calculate the dispersion of all points in the root set up to point t. center Calculate the standard deviation σ or interquartile range IQR of the distances.

[0057] ③ Set confidence interval (edge ​​truncation): retain those projection distances within [t] center -k*σ,t center Points within the range of +k*σ] (usually k is 1.5 or 2.0).

[0058] ④ Output center set: The points that are filtered out are the "root center region point set" after removing the edge burrs at both ends.

[0059] The above algorithms are extremely fast, perfectly match the main logic of the early stage, and have a strong ability to resist edge interference.

[0060] However, one-dimensional statistics only focuses on the "length position" of a point on the principal axis, completely ignoring the "two-dimensional morphology" of the point on the cross-section. When the base of a feather is under pressure or has a natural curvature, its cross-section is often flat or crescent-shaped. From a one-dimensional perspective, these normal flat edge points are far from the center and are easily mistakenly deleted as "outliers." Conversely, if the edge is attached to a piece of hair or dust close to the base (closer to the center), one-dimensional statistics will classify it as an "internal valid point" and retain it.

[0061] Meanwhile, one-dimensional statistical elimination relies on the globally calculated mean and standard deviation. In actual production lines, feather roots often have irregular barbs, forks, or "flying spots" caused by laser scattering. These outliers widen the global standard deviation σ, resulting in an excessively wide filtering threshold. As a result, the burrs that should be cut are not removed, while the core feather shafts that should not be cut are partially removed, distorting the reference plane benchmark for subsequent fitting.

[0062] Therefore, based on the goal of addressing laser scattering noise at the edges of the root point cloud, this embodiment employs an axial slice shrinkage algorithm:

[0063] ① Divide the root point set into multiple parallel layers with a thickness of δ (e.g., 0.2 mm) along the principal axis v.

[0064] ② For each layer, project the points within the layer onto a two-dimensional plane perpendicular to v.

[0065] ③ Adopt a non-maximum suppression strategy based on quantiles (such as removing 5% from the beginning and end) or a local extremum shrinkage strategy to determine the effective inner boundary of the layer and remove points outside the boundary. The shrinkage range can be a fixed number of pixels or a fixed ratio, or it can be adaptively adjusted according to the point density within the layer: more shrinkage for denser points and less shrinkage for sparser points.

[0066] Based on the aforementioned edge culling mechanism of "axial slicing + adaptive shrinkage," the baseline stability is significantly improved. Addressing the issues of laser scattering noise and microscopic barbs at the edge of the feather root point cloud, this invention innovatively proposes an edge culling strategy based on layered slicing. Unlike traditional global filtering, this scheme independently shrinks the boundary within each micro-element layer and adaptively adjusts it in conjunction with the point cloud density. This design can precisely peel away interference points like "peeling an onion," preserving the core geometric features of the feather shaft, thus laying a solid foundation for subsequent fitting of a high-precision reference plane.

[0067] 5. Fitting the reference plane

[0068] The fitting of the reference plane can also be achieved based on existing technologies. In this embodiment, to improve the robustness of the system, a multi-level backtracking plane fitting is proposed. The reference plane is fitted using the point set of the root center region, and the following strategies are combined, with an automatic backtracking mechanism configured:

[0069] Strategy A (Main Strategy): RANSAC robust fitting. Randomly sample three points to construct a plane, iteratively find the largest set of interior points (threshold set is 30% of the total number of points), and finally refine using least squares. If it fails (insufficient interior points), revert to Strategy B.

[0070] Strategy B (auxiliary strategy): Histogram voting method. Construct a histogram for the plane normal vector parameters, smooth it with Gaussian, and then extract sub-pixel peaks. If it fails (multiple peaks without a dominant peak), revert to strategy C.

[0071] Strategy C (Minimum Guarantee Strategy): Surface Projection Correction Method. First, fit the root point with a quadratic surface (to compensate for the slight curvature of the biomaterial), then project the point onto the surface, and finally perform planar fitting on the projected point.

[0072] 6. Arch calculation

[0073] Calculate the signed distance d from each point in the apex set to the reference plane. To eliminate outlier interference caused by leaf curling, take the median of all distances d as the final camber measurement H.

[0074] The above description is merely a specific embodiment or illustration of the embodiments of this application. The scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. The scope of protection of this application shall be determined by the scope of the claims.

Claims

1. A method for measuring feather camber based on three-dimensional point clouds, characterized in that, Includes the following steps: Step S1: Obtain the three-dimensional point cloud data of the feather; Step S2: Determine the principal axis direction of the feather, perform principal axis projection on the 3D point cloud, and obtain the one-dimensional projection parameters along the principal axis direction; Step S3: Based on the one-dimensional projection parameters, the three-dimensional point cloud is divided into root point set and tip point set; Step S4: Perform edge removal processing on the root point set to obtain the root center region point set; Step S5: Fit the reference plane using the point set of the root center region; Step S6: Calculate the signed distance from each point in the cusp set to the reference plane; Step S7: Take the median of the signed distances as the camber measurement value.

2. The feather arch measurement method based on three-dimensional point clouds according to claim 1, characterized in that, In step S2, principal component analysis is used to determine the principal axis direction of the feather: the covariance matrix of the three-dimensional point cloud is calculated, the covariance matrix is ​​decomposed into eigenvalues, and the eigenvector corresponding to the largest eigenvalue is taken as the principal axis direction.

3. The feather arch measurement method based on three-dimensional point cloud according to claim 2, characterized in that, Step S3 further includes the following steps: Step S31: Calculate the cumulative distribution function CDF based on the one-dimensional projection parameters obtained in step S2, and take the low quantile interval of the cumulative distribution as the first region and the high quantile interval as the second region. Step S32: Based on the prior knowledge of feather morphology, determine the semantic labels of the first region and the second region as either root point set or tip point set.

4. The feather arch measurement method based on three-dimensional point clouds according to claim 3, characterized in that, The root point set corresponds to the low quantile range of 0%~30% of the one-dimensional projection parameters, while the apex point set corresponds to the high quantile range of 70%~100% of the one-dimensional projection parameters.

5. The feather arch measurement method based on three-dimensional point cloud according to claim 3, characterized in that, Step S4 further includes the following steps: Step S41, dividing the root point set into multiple parallel layers along the principal axis; Step S42, for each parallel layer, calculating the projection distribution of points within that layer in the direction perpendicular to the principal axis; Step S43, determining the effective boundary using a strategy based on quantiles or local extremum shrinkage, and removing points outside the boundary, wherein the shrinkage magnitude is a fixed number of pixels, a fixed ratio, or an adaptive value based on the point density within the layer.

6. The feather arch measurement method based on three-dimensional point cloud according to claim 1, characterized in that, The fitting reference plane in step S5 employs one of the following strategies and is configured with an automatic backoff mechanism: Strategy A, RANSAC plane fitting method: randomly sample three points to form a plane model, iteratively find the maximum set of interior points, and finally refine it using the least squares method; Strategy B, Histogram Voting Method: Construct histograms for the slope and intercept parameters of the plane equation respectively, and determine the optimal parameters through Gaussian smoothing and sub-pixel peak extraction; Strategy C, RANSAC with local back-substitution plane fitting method: First, fit the root point with a quadratic surface, then project the root point onto the surface, and finally fit the projected point with the least squares method to obtain the plane. When the current strategy fails to fit, it automatically reverts to the next strategy and continues execution.

7. The feather arch measurement method based on three-dimensional point clouds according to claim 6, characterized in that, The criteria for determining fitting failure include: In Strategy A, the proportion of the largest interior set is lower than the preset threshold, or the iteration count is exhausted and the convergence is not achieved. In Strategy B, the histogram shows a multi-peak distribution with no significant main peak, or the peak confidence is lower than the preset threshold. In Strategy C, the residual of the quadratic surface fitting is greater than the preset threshold, or the projected point set cannot form an effective plane.

8. The feather arch measurement method based on three-dimensional point cloud according to claim 1, characterized in that, Step 7 also includes: before calculating the signed distance, performing outlier filtering on the tip point set to remove abnormal distance values ​​caused by leaf curling or scanning noise.

9. A feather arch measurement system based on three-dimensional point clouds, characterized in that, include: The point cloud acquisition module is used to acquire the three-dimensional point cloud data of the feather; The principal axis calculation module is used to perform principal axis analysis on the 3D point cloud to obtain the one-dimensional projection parameters and principal axis direction vector in the principal axis direction. The layering module is used to divide a 3D point cloud into a root point set and a apex point set based on one-dimensional projection parameters. The edge culling module is used to perform edge culling on the root point set to obtain the root center region point set; The plane fitting module is used to fit a reference plane using the point set of the root center region; The camber calculation module is used to calculate the signed distance from each point in the top point set to the reference plane, and take the median of the signed distances as the camber measurement value.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the feather arch measurement method as described in any one of claims 1 to 8.

Images