A circle hole extraction method and system based on secondary region of interest division

Abstract

This application relates to the field of 3D scanning point cloud processing technology, and discloses a method and system for extracting circular holes based on secondary region of interest (ROI) segmentation. The method first performs initial segmentation and fitting on the original workpiece point cloud to obtain a first-level ROI. Then, based on the estimated initial central axis and initial radius of the first-level ROI, potential edge-distorted point clouds are identified and removed for secondary segmentation. A second-level ROI is determined from the first-level ROI, and further fitting and analysis based on the second-level ROI determine the center coordinates and radius of the target circular hole. Thus, through secondary ROI segmentation, edge-distorted point clouds can be specifically targeted and removed. This method is not a generalized data filtering technique, but a targeted cleaning mechanism for the specific physical phenomenon of hole edge distortion, which can improve the accuracy of circular hole identification and extraction.

Description

Technical Field

[0001] This invention relates to the field of three-dimensional scanning point cloud processing technology, and in particular to a method and system for extracting circular holes based on secondary region of interest division. Background Technology

[0002] In modern industrial manufacturing and quality control, 3D scanning and point cloud processing technologies play a crucial role. Especially in industries such as aerospace, automotive manufacturing, and precision electronics, components contain numerous circular holes used for connection, positioning, or conduction. The processing quality of these holes directly affects the assembly accuracy and performance of the product. Therefore, achieving high-precision, automated inspection of the size and position of circular holes has become one of the core requirements in industrial production. With the development of 3D measurement technologies, such as laser scanning and structured light photogrammetry, massive amounts of 3D point cloud data on workpiece surfaces can be rapidly acquired. Accurately extracting the geometric parameters of the circular holes (such as center coordinates and radius) from these dense and potentially noisy point clouds is a key prerequisite for subsequent tolerance analysis and quality assessment.

[0003] Currently, point cloud-based methods for extracting geometric parameters of circular holes mainly fall into two categories: direct fitting and model matching. However, these methods significantly impact measurement accuracy when dealing with circular holes formed by specific processes, especially stamped holes. The stamping process, while forming a circular hole, generates microscopic plastic deformation at the hole's edge, typically manifesting as slight collapse, curling, or concavity (i.e., "tool deflection"). This results in the point cloud data near the inner wall of the hole not being located on an ideal cylindrical surface, but rather presenting a non-standard, complex curved surface with chamfers or concavities. Traditional direct fitting methods include this distorted point cloud along with the normal hole wall point cloud in the fitting process, causing the final fitted circular contour to shift towards the distorted region. This systematically increases the fitting radius of the circular hole and leads to a deviation in the center position. This systematic error introduced by the manufacturing process itself severely affects the realism and accuracy of the measurement. Model matching methods also cannot effectively solve the fitting deviation problem caused by the distortion at the hole's edge, unless the distortion feature is accurately modeled in the CAD model, which is often uneconomical and impractical in actual engineering.

[0004] It is evident that existing methods for extracting circular hole features have low measurement accuracy when dealing with circular holes formed by specific processes. Summary of the Invention

[0005] This invention provides a method and system for extracting circular holes based on secondary region of interest (ROI) division, in order to solve the problem that existing circular hole feature extraction methods have low measurement accuracy when dealing with circular holes formed by specific processes.

[0006] Firstly, this application provides a method for extracting circular holes based on quadratic region of interest (ROI) division, including:

[0007] The initial point cloud set containing the target circular hole is segmented from the acquired original workpiece point cloud as the first-level region of interest; The initial central axis and initial radius of the target circular hole are estimated based on the first-level region of interest; Based on the initial central axis and initial radius, edge distortion point clouds in the first-level region of interest are identified and removed to determine the second-level region of interest from the first-level region of interest; A reference plane is obtained by fitting based on the second-level region of interest, and the reference plane is perpendicular to the central axis of the target circular hole; and all point clouds in the second-level region of interest are vertically projected onto the reference plane to obtain a set of two-dimensional planar point clouds; The center coordinates and radius of the target circular hole are determined based on the two-dimensional planar point cloud, and the target circular hole is extracted based on the center coordinates and radius.

[0008] Secondly, this application also provides a circular hole extraction system based on secondary region of interest division, comprising: The first unit is used to segment an initial point cloud set containing the target circular hole from the acquired original workpiece point cloud as the first-level region of interest. The second unit is used to estimate the initial central axis and initial radius of the target circular hole based on the first-level region of interest; The third unit is used to identify and remove edge distortion point clouds in the first-level region of interest based on the initial central axis and initial radius, so as to determine the second-level region of interest from the first-level region of interest; The fourth unit is used to fit a reference plane based on the second-level region of interest, the reference plane being perpendicular to the central axis of the target circular hole; and to project all point clouds in the second-level region of interest vertically onto the reference plane to obtain a set of two-dimensional planar point clouds; The fifth unit is used to determine the center coordinates and radius of the target circular hole based on the two-dimensional planar point cloud, and to extract the target circular hole according to the center coordinates and radius.

[0009] The present invention has the following beneficial effects: The circular hole extraction method based on secondary region of interest (ROI) segmentation proposed in this application firstly segments and fits the original workpiece point cloud to obtain a first-level ROI. Then, based on the first-level ROI, the initial central axis and initial radius are estimated. Potential edge-distorted point clouds are identified based on the initial central axis and initial radius, and these are removed for secondary segmentation. A second-level ROI is determined from the first-level ROI. Furthermore, the center coordinates and radius of the target circular hole are determined based on the fitting and analysis of the second-level ROI. The target circular hole is then extracted based on these center coordinates and radius. In this way, through secondary ROI segmentation, edge-distorted point clouds can be specifically targeted and removed. This method is not a generalized data filtering technique, but rather a targeted cleaning mechanism for the specific physical phenomenon of hole edge distortion, which can improve the accuracy of circular hole identification and extraction.

[0010] In addition to the objectives, features and advantages described above, the present invention has other objectives, features and advantages.

[0011] The present invention will now be described in further detail with reference to the figures. Attached Figure Description

[0012] The accompanying drawings, which form part of this application, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings: Figure 1 This is a flowchart of a circular hole extraction method based on secondary region of interest division according to a preferred embodiment of the present invention. Detailed Implementation

[0013] The technical solution of the present invention will be clearly and completely described below. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0014] Unless otherwise defined, the technical or scientific terms used in this invention shall have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Similarly, the terms "an" or "a," and similar terms, do not indicate a quantity limitation, but rather indicate the presence of at least one.

[0015] It should be understood that existing technologies have inherent accuracy limitations when dealing with geometric distortions at the edges of holes caused by processes such as stamping, and cannot meet the ever-increasing demands for precision industrial inspection. Specifically, circular holes formed by specific processes include stamped holes and drilled holes. Therefore, there is an urgent need for a new method for extracting circular holes that can effectively eliminate the influence of edge distortion regions, thereby significantly improving the accuracy and reliability of the calculation of circular hole geometric parameters.

[0016] Please see Figure 1 This application provides a method for extracting circular holes based on quadratic region of interest (ROI) division, including: The initial point cloud set containing the target circular hole is segmented from the acquired original workpiece point cloud as the first-level region of interest; The initial central axis and initial radius of the target circular hole are estimated based on the first-level region of interest; Based on the initial central axis and initial radius, edge distortion point clouds in the first-level region of interest are identified and removed to determine the second-level region of interest from the first-level region of interest; A reference plane is obtained by fitting based on the second-level region of interest, and the reference plane is perpendicular to the central axis of the target circular hole; and all point clouds in the second-level region of interest are vertically projected onto the reference plane to obtain a set of two-dimensional planar point clouds; The center coordinates and radius of the target circular hole are determined based on the two-dimensional planar point cloud, and the target circular hole is extracted based on the center coordinates and radius.

[0017] The aforementioned circular hole extraction method based on secondary region of interest (ROI) segmentation first performs initial segmentation and fitting on the original workpiece point cloud to obtain a first-level ROI. Then, based on the first-level ROI, the initial central axis and initial radius are estimated. Potential edge-distorted point clouds are identified based on the initial central axis and initial radius, and these are removed for secondary segmentation. A second-level ROI is determined from the first-level ROI, and further fitting and analysis based on the second-level ROI determine the center coordinates and radius of the target circular hole. Through secondary ROI segmentation, distorted point clouds at the hole opening edge can be removed. This method is not a generalized data filtering technique, but rather a targeted cleaning mechanism for the specific physical phenomenon of edge distortion at the opening of stamped holes, which can improve the accuracy of circular hole identification and extraction.

[0018] In one embodiment, segmenting an initial point cloud set containing the target circular hole from the acquired original workpiece point cloud as a first-level region of interest includes: An initial point cloud set containing the target circular hole is segmented from the acquired original workpiece point cloud using a preset method as the first-level region of interest. The preset methods include region growing or spatial grating. Since the original point cloud typically contains a large number of irrelevant background, stray points, or noise points, these can interfere with the accurate identification of the circular hole feature. In this embodiment, an initial point cloud set containing the target circular hole is segmented as a first-level Region of Interest (ROI) using region growing or spatial grating methods.

[0019] Specifically, the selection method for the preset mode is as follows: When the original workpiece point cloud corresponds to a planar workpiece and the difference in geometric features between the target circular hole and the surrounding area exceeds a set threshold, the preset method is region growing. Through region growing, initial region segmentation with low computational load and high real-time performance can be achieved for planar workpieces and simple scenes, solving the problem of computational redundancy in simple scenes using traditional methods.

[0020] When the size and density of the original workpiece point cloud both exceed the set threshold and the target circular hole position can be coarsely located by spatial coordinates, the preset method is spatial grid. By using spatial grid, fast downsampling and coarse segmentation can be achieved for large-scale point clouds and noisy scenes, solving the problems of low processing efficiency and large noise interference in large-scale point cloud processing.

[0021] In one embodiment, estimating the initial central axis and initial radius of the target circular hole based on the first-level region of interest includes: Preliminary planar or cylindrical surface fitting is performed on the first-level region of interest to estimate the initial central axis and initial radius of the target circular hole.

[0022] In this implementation, a fitting operation is performed on the first-level region of interest (ROI) instead of calculating the entire original point cloud. This method of local focusing combined with simple fitting can quickly obtain effective initial parameters and significantly reduce the number of iterations for subsequent secondary segmentation and precise extraction. Specifically, by performing preliminary modeling of the first-level ROI through planar or cylindrical surface fitting, the initial central axis and initial radius of the target circular hole can be accurately estimated from the geometric feature level. Using this initial central axis and initial radius as the core basis for subsequent secondary ROI segmentation can effectively narrow the range of secondary segmentation, lock the feature focusing direction, avoid extraction deviations caused by benchmark ambiguity in subsequent processing, and improve the detection accuracy of the final circular hole parameters from the source.

[0023] As a preferred embodiment, the step of identifying and removing edge distortion point clouds from the first-level region of interest based on the initial central axis and initial radius to determine the second-level region of interest from the first-level region of interest includes: Calculate the radial distance from each point in the first-level region of interest to the initial central axis, and generate a histogram of radial distance distribution; The distance distribution curve corresponding to the radial distance distribution map is determined, and the inflection point of the distance distribution curve is determined based on the initial radius. According to the inflection point of the distance distribution curve, the point cloud corresponding to the first-level region of interest is divided into inner wall point cloud and edge distortion point cloud. The edge distortion point cloud is removed to obtain the second-level region of interest.

[0024] In this implementation, the culling threshold is automatically determined by analyzing the radial distance distribution map of the point cloud from the initial central axis in the first-level ROI. Since edge-distorted point clouds typically cause a tail or a second peak in the distance distribution near the initial radius, the point cloud can be categorized into inner-wall point clouds and edge-distorted point clouds by identifying the inflection point of the distance distribution curve, thus achieving adaptive threshold setting. This method eliminates the need for manual parameter setting and is more intelligent than manual experience-based methods.

[0025] As a variation of implementation, the point cloud corresponding to the first-level region of interest can also be divided into inner wall point cloud and edge distortion point cloud by clustering algorithm, and the edge distortion point cloud can be removed to obtain the second-level region of interest.

[0026] More preferably, identifying and removing edge distortion point clouds from the first-level region of interest based on the initial central axis and initial radius to determine the second-level region of interest from the first-level region of interest includes: Set a preset safety threshold; The radius threshold is obtained by subtracting the initial radius from the safety threshold. Based on the initial central axis, a virtual cylindrical surface with a cross-sectional radius of a radius threshold is constructed; The second-level region of interest is obtained by removing the edge distortion point cloud located outside the virtual cylindrical surface in the first-level region of interest.

[0027] Specifically, in one example, an initial radius of R1 is set, and a virtual cylindrical surface with a radius of R1-Δd is constructed, where Δd is a preset safety threshold, such as 0.1mm to 0.5mm. In one example, Δd could also be 0.3mm; this is merely an example and not a limitation. The range of values ​​for Δd can be flexibly adjusted according to the specific measurement scenario; for example, a smaller value can be used for high-precision measurements, while a larger value can be used for complex working conditions. This adjustable design makes this method not limited to specific sizes or accuracies, significantly improving the applicability and versatility of the technical solution.

[0028] In this optional implementation, a virtual cylindrical surface is determined based on an initial central axis and a radius threshold. Point clouds located outside this virtual cylindrical surface are identified as potentially edge-distorted point clouds and are removed. The set of point clouds located inside this virtual cylindrical surface constitutes a second-level ROI representing the main portion of the hole wall. This isolates unreliable data near the hole opening.

[0029] Furthermore, the generated virtual cylindrical surface provides a precise model of the geometric features of the inner wall of the circular hole, serving as a core constraint for secondary region of interest (ROI) segmentation. The virtual cylindrical surface more accurately defines the spatial extent and geometric shape of the secondary region, allowing subsequent hole extraction steps to focus on cleaner and more representative point cloud data, fundamentally improving the detection accuracy of the final hole parameters.

[0030] In one embodiment, determining the center coordinates and radius of the target circular hole based on the two-dimensional planar point cloud can be achieved by using a fitting method to perform circle fitting on the two-dimensional planar point cloud to obtain the center coordinates and radius of the target circular hole. In this embodiment, the fitting method can be RANSAC, algebraic fitting, or geometric fitting, etc. The accurate parameters obtained by fitting methods such as RANSAC, algebraic fitting, or geometric fitting can provide reliable data for subsequent quality inspection, error analysis, and process optimization, helping to improve the precision of product quality control and obtain high-precision center coordinates and radius.

[0031] In a more preferred embodiment, determining the center coordinates and radius of the target circular hole based on the two-dimensional planar point cloud includes: A fitting method is used to perform a circle fitting on the two-dimensional planar point cloud to obtain the coordinates of the first center and the first radius of the first circular hole; Using the first radius as the initial iteration radius, a first virtual cylindrical surface is obtained based on the initial central axis and the initial iteration radius, and the second-level region of interest under the current iteration is determined based on the first virtual cylindrical surface; The reference plane for the current iteration is obtained by fitting the second-level region of interest under the current iteration. Project all point clouds in the second-level region of interest under the current iteration vertically onto the reference plane under the current iteration to obtain a two-dimensional planar point cloud under the current iteration. The second center coordinates and second radius of the second hole are determined based on the two-dimensional planar point cloud under the current iteration; the second radius is used as the current iteration radius for iteration until the radius change between two adjacent iterations is less than the convergence threshold, the iteration step is stopped, and the center coordinates and radius of the target hole are determined based on the output of the last iteration step.

[0032] In this preferred embodiment, an iterative process is used to gradually approach the optimal solution. Specifically: after obtaining the first radius by executing all steps for the first time, the first radius is used as the initial iteration radius. The virtual cylindrical surface is redefined as the exclusion region, and subsequent steps are repeated for a second, third, and several more fitting iterations. Each iteration further purifies the point cloud based on the better result of the previous iteration until the radius change between two adjacent fittings is less than a certain convergence threshold. At this point, the iteration stops, and the center coordinates and radius of the target circular hole are determined based on the output of the last iteration. In this way, by gradually approaching the optimal solution through multiple iterations, the ultimate accuracy can be further improved.

[0033] This application also provides a circular hole extraction system based on quadratic region of interest division, including: The first unit is used to segment an initial point cloud set containing the target circular hole from the acquired original workpiece point cloud as the first-level region of interest. The second unit is used to estimate the initial central axis and initial radius of the target circular hole based on the first-level region of interest; The third unit is used to identify and remove edge distortion point clouds in the first-level region of interest based on the initial central axis and initial radius, so as to determine the second-level region of interest from the first-level region of interest; The fourth unit is used to fit a reference plane based on the second-level region of interest, the reference plane being perpendicular to the central axis of the target circular hole; and to project all point clouds in the second-level region of interest vertically onto the reference plane to obtain a set of two-dimensional planar point clouds; The fifth unit is used to determine the center coordinates and radius of the target circular hole based on the two-dimensional planar point cloud, and to extract the target circular hole according to the center coordinates and radius.

[0034] Optionally, based on the initial central axis, edge distortion point clouds in the first-level region of interest are identified and removed to determine a second-level region of interest from the first-level region of interest, including: Calculate the distance from each point in the first-level region of interest to the initial central axis, and generate a distance distribution map; The distance distribution curve corresponding to the distance distribution map is determined, and the inflection point of the distance distribution curve is determined based on the initial radius. According to the inflection point of the distance distribution curve, the point cloud corresponding to the first-level region of interest is divided into inner wall point cloud and edge distortion point cloud. The edge distortion point cloud is removed to obtain the second-level region of interest.

[0035] In this optional implementation, the culling threshold is automatically determined by analyzing the radial distance distribution histogram of the point cloud from the initial central axis in the first-level ROI. Specifically, the distance from each point to the initial central axis is calculated, generating a distance distribution map. Since edge-distorted point clouds typically cause a tail or a second peak in the distance distribution near the initial radius, the point cloud can be divided into inner-wall point clouds and edge-distorted point clouds by finding the inflection point of the distribution curve, thereby achieving adaptive threshold setting. This method eliminates the need for manual parameter setting based on experience, making it more intelligent than manual setting methods.

[0036] Optionally, based on the initial central axis and initial radius, edge distortion point clouds in the first-level region of interest are identified and removed to determine a second-level region of interest from the first-level region of interest, including: The radius threshold is obtained by subtracting the initial radius from the safety threshold. With the initial central axis as the center, construct a virtual cylindrical surface with a cross-sectional radius of a threshold value; The second-level region of interest is obtained by removing the edge distortion point cloud located outside the virtual cylindrical surface in the first-level region of interest.

[0037] In this optional implementation, a virtual cylindrical surface is determined based on an initial central axis and an initial radius. Point clouds located outside this virtual cylindrical surface are identified as potentially edge-distorted point clouds and are removed. The set of point clouds located inside this virtual cylindrical surface constitutes a second-level ROI representing the main portion of the hole wall. This isolates unreliable data near the hole opening.

[0038] The circular hole extraction system based on quadratic region of interest division can implement all the embodiments of the circular hole extraction method based on quadratic region of interest division described above, and can achieve the same beneficial effects. Here, it will not be elaborated further.

[0039] The above description is merely a specific embodiment of this disclosure, but the scope of protection of this disclosure 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 disclosure should be included within the scope of protection of this disclosure. Therefore, the scope of protection of this disclosure should be determined by the scope of the claims.

Claims

1. A method for extracting circular holes based on quadratic region of interest (ROI) division, characterized in that, include: The initial point cloud set containing the target circular hole is segmented from the acquired original workpiece point cloud as the first-level region of interest; The initial central axis and initial radius of the target circular hole are estimated based on the first-level region of interest; Based on the initial central axis and initial radius, edge distortion point clouds in the first-level region of interest are identified and removed to determine the second-level region of interest from the first-level region of interest; A reference plane is obtained by fitting based on the second-level region of interest, and the reference plane is perpendicular to the central axis of the target circular hole; Then, all point clouds in the second-level region of interest are vertically projected onto the reference plane to obtain a set of two-dimensional planar point clouds; The center coordinates and radius of the target circular hole are determined based on the two-dimensional planar point cloud, and the target circular hole is extracted based on the center coordinates and radius.

2. The circular hole extraction method based on quadratic region of interest division according to claim 1, characterized in that, The step of segmenting an initial point cloud set containing the target circular hole from the acquired original workpiece point cloud as the first-level region of interest includes: An initial point cloud set containing the target circular hole is segmented from the acquired original workpiece point cloud using a preset method as the first-level region of interest. The preset methods include: region growing or spatial grid; and the selection method of the preset method is as follows: When the original workpiece point cloud corresponds to a planar workpiece and the difference in geometric features between the target circular hole and the surrounding area exceeds a set threshold, the preset method is region growth. When the size and density of the original workpiece point cloud both exceed a set threshold and the target circular hole position can be coarsely located using spatial coordinates, the preset method is spatial grid.

3. The circular hole extraction method based on quadratic region of interest division according to claim 1, characterized in that, The estimation of the initial central axis and initial radius of the target circular hole based on the first-level region of interest includes: The first-level region of interest is fitted with a planar or cylindrical surface to estimate the initial central axis and initial radius of the target circular hole.

4. The circular hole extraction method based on quadratic region of interest division according to claim 1, characterized in that, The step of identifying and removing edge distortion point clouds from the first-level region of interest based on the initial central axis and initial radius, in order to determine the second-level region of interest from the first-level region of interest, includes: Calculate the radial distance from each point in the first-level region of interest to the initial central axis, and generate a radial distance distribution map; The distance distribution curve corresponding to the radial distance distribution map is determined, and the inflection point of the distance distribution curve is determined based on the initial radius. According to the inflection point of the distance distribution curve, the point cloud corresponding to the first-level region of interest is divided into inner wall point cloud and edge distortion point cloud. The edge distortion point cloud is removed to obtain the second-level region of interest.

5. The circular hole extraction method based on quadratic region of interest division according to claim 1, characterized in that, Based on the initial central axis and initial radius, edge distortion point clouds in the first-level region of interest are identified and removed to determine the second-level region of interest from the first-level region of interest, including: Set a preset safety threshold; The radius threshold is obtained by subtracting the initial radius from the safety threshold. Based on the initial central axis, a virtual cylindrical surface with a cross-sectional radius of a radius threshold is constructed; The second-level region of interest is obtained by removing the edge distortion point cloud located outside the virtual cylindrical surface in the first-level region of interest.

6. The circular hole extraction method based on quadratic region of interest division according to claim 5, characterized in that, Determining the center coordinates and radius of the target circular hole based on the two-dimensional planar point cloud includes: A fitting method is used to perform a circle fitting on the two-dimensional planar point cloud to obtain the coordinates of the first center and the first radius of the first circular hole; Using the first radius as the initial iteration radius, a first virtual cylindrical surface is obtained based on the initial central axis and the initial iteration radius, and the second-level region of interest under the current iteration is determined based on the first virtual cylindrical surface; The reference plane for the current iteration is obtained by fitting the second-level region of interest under the current iteration. Project all point clouds in the second-level region of interest under the current iteration vertically onto the reference plane under the current iteration to obtain a two-dimensional planar point cloud under the current iteration. The second center coordinates and second radius of the second hole are determined based on the two-dimensional planar point cloud under the current iteration; the second radius is used as the current iteration radius for iteration until the radius change between two adjacent iterations is less than the convergence threshold, the iteration step is stopped, and the center coordinates and radius of the target hole are determined based on the output of the last iteration step.

7. A circular hole extraction system based on quadratic region of interest (ROI) division, characterized in that, include: The first unit is used to segment an initial point cloud set containing the target circular hole from the acquired original workpiece point cloud as the first-level region of interest. The second unit is used to estimate the initial central axis and initial radius of the target circular hole based on the first-level region of interest; The third unit is used to identify and remove edge distortion point clouds in the first-level region of interest based on the initial central axis and initial radius, so as to determine the second-level region of interest from the first-level region of interest; The fourth unit is used to fit a reference plane based on the second-level region of interest, the reference plane being perpendicular to the central axis of the target circular hole; Then, all point clouds in the second-level region of interest are vertically projected onto the reference plane to obtain a set of two-dimensional planar point clouds; The fifth unit is used to determine the center coordinates and radius of the target circular hole based on the two-dimensional planar point cloud, and to extract the target circular hole according to the center coordinates and radius.

Images