A method for detecting cell parameters

Abstract

The present invention discloses a method for detecting pore parameters, comprising: step 1, calculating the normal vectors of three-dimensional points in a three-dimensional point cloud of a pore; step 2, extracting three-dimensional points to fit a pore cone, and evaluating the error between the fitted pore cone and a theoretical pore cone; step 3, screening and eliminating three-dimensional points whose fitting error exceeds a standard value; step 4, for a three-dimensional point whose fitting error meets the standard, selecting other three-dimensional points within the neighborhood of the three-dimensional point; step 5, constructing a candidate set of the selected three-dimensional points, solving an optimal candidate set with the goal of maximizing the number of three-dimensional points in the candidate set and minimizing the error between the fitted pore cone fitted by the three-dimensional points in the candidate set and the theoretical pore cone; step 6, fitting the pore cone using the three-dimensional points in the optimal candidate set and solving the parameters of the pore; the present invention separates and fits the pore cone based on a three-dimensional point cloud model of the pore, and eliminates and optimizes the three-dimensional points with an out-of-tolerance value, so as to fit the pore cone more efficiently and accurately, thereby ensuring the efficiency and accuracy of subsequent extraction of pore parameters.

Description

Technical Field

[0001] The invention belongs to the technical field of socket measurement and relates to a method for detecting socket parameters. Background Art

[0002] Commonly used materials for aircraft skins are titanium alloys, aluminum alloys, and composite materials. These materials are difficult to weld and are relatively thin. Forced welding will inevitably cause bubbles and cracks, which will pose a great threat to flight safety. Therefore, aircraft skins are mostly connected with rivets. Compared with welding technology, riveted connections have high strength and good tightness. They are highly reliable when subjected to impact loads and are not easy to loosen. They are suitable for thinner connectors.

[0003] To ensure the reliability of the aircraft, the process has high requirements for the size of the holes in the aircraft skin. The traditional measurement method is to use a contact hole gauge to measure the depth of the hole, but there are several disadvantages:

[0004] 1. Each nest gauge can only measure the nest depth of one nest angle;

[0005] 2. The nest gauge can only be used offline and cannot be used for online measurement during the processing;

[0006] 3. The socket gauge can only measure the depth or diameter, but cannot measure the socket angle and coaxiality.

[0007] Therefore, there is an urgent need for a method based on three-dimensional modeling of the pore to detect pore parameters. Summary of the Invention

[0008] The purpose of the present invention is to provide a method for detecting cellular parameters, which separates and fits the cellular cone based on the three-dimensional point cloud model of the cellular cavity, and eliminates and optimizes the three-dimensional points that exceed the tolerance, so as to fit the cellular cone more efficiently and accurately, thereby ensuring the efficiency and accuracy of the subsequent extraction of cellular parameters.

[0009] The present invention is achieved through the following technical solutions:

[0010] A method for detecting fossa parameters is implemented based on a 3D point cloud model of fossa captured by a camera, comprising the following steps:

[0011] Step 1: Calculate the normal vector of the three-dimensional point in the three-dimensional point cloud of the socket and correct the normal vector;

[0012] Step 2: extracting 3D points from the 3D point cloud of the cavities to fit the cavities cone, and evaluating the error between the fitted cavities cone and the theoretical cavities cone;

[0013] Step 3: Based on the error, the three-dimensional points whose fitting error exceeds the standard value are screened and eliminated;

[0014] Step 4: For the 3D points with qualified fitting errors, select other 3D points within the neighborhood range of the 3D points;

[0015] Step 5: Construct a candidate set with the selected 3D points, and solve the optimal candidate set with the goal of maximizing the number of 3D points in the candidate set and minimizing the error between the fitting socket hole cone and the theoretical socket hole cone fitted by the 3D points in the candidate set;

[0016] Step 6: Fit the socket hole cone with the 3D points in the optimal candidate set, and separate the parameters of the socket hole through the fitted socket hole cone.

[0017] To better implement the present invention, further, extract 3D points on the fitted socket hole cone, project the current 3D point onto the theoretical socket hole cone to obtain a projection point, the error includes a distance error and an angle error, the distance error is the distance between the 3D point and the corresponding projection point, and the angle error is the angle between the normal of the 3D point and the normal of the corresponding projection point.

[0018] To better implement the present invention, further, the candidate set is:

[0019] p φ ={p i 丨p i ∈P∧d(φ,p i )<ε∧arccos(|n(p i )·n(φ,p i )|)<α};

[0020] Where: p φ represents the candidate set; p i represents the i-th 3D point in the candidate set; P represents the socket hole 3D point cloud; d(φ,p i ) represents the distance between the 3D point and the projection point; n(p i ) represents the normal of the 3D point; n(φ,p i ) represents the normal of the projection point; ε represents the distance calibration value; α represents the angle calibration value.

[0021] To better implement the present invention, further, divide the 3D points in the socket hole 3D point cloud into several non-overlapping 3D point subsets, establish a scoring function, and score the 3D point subsets based on the candidate set.

[0022] To better implement the present invention, further, the scoring function is:

[0023] σ si =-1-f(-2-|s1|,-2-|p|,-1-σ s1 ); [[ID=%]]

[0024] Where: |s1| is the number of 3D points in the first 3D point subset; |p| is the number of 3D points in the 3D point cloud of the pore; σ s1 is the score of the first 3D point subset; σ si is the score of the other i-th 3D point subset; f(-2-|s1|,-2-

[0025] |p|,-1-σ s1 ) represents the hypergeometric distribution with respect to the variable.

[0026] In order to better implement the present invention, further, in the step 2, three non-coplanar three-dimensional points are extracted from the cellular three-dimensional point cloud, the cone intersection is solved through the normal plane where the three non-coplanar three-dimensional points are located, and the cellular cone is fitted based on the three non-coplanar three-dimensional points and the cone intersection; the cone axis is solved through the three non-coplanar three-dimensional points and the cone intersection, the cone center angle is solved through the three non-coplanar three-dimensional points and the cone intersection, and the cellular cone is fitted through the cone axis and the center angle.

[0027] In order to better implement the present invention, further, in step 2, an octree structure is established for the three-dimensional points, and based on the hierarchical depth of the octree structure, the remaining three-dimensional points are selected within the neighborhood of the current three-dimensional point, and the probability that all the selected three-dimensional points belong to the same cellular cone is calculated.

[0028] In order to better implement the present invention, further, the calculation formula for the probability that the three-dimensional points belong to the same cellular cone is:

[0029]

[0030] Where: p local (n) represents the probability that the n selected 3D points belong to the same pore cone; n represents the number of selected 3D points; N represents the number of 3D points in the pore 3D point cloud; d represents the hierarchical depth of the octree structure; k d Indicates the depth of the current 3D point in the octree structure.

[0031] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0032] The present invention separates and fits the pore cone based on the three-dimensional point cloud model of the pore, and eliminates and optimizes the out-of-tolerance three-dimensional points, which can fit the pore cone more efficiently and accurately, ensuring the efficiency and accuracy of subsequent extraction of pore parameters. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 A flow chart of the steps of a method for detecting socket parameters;

[0034] Figure 2 is a schematic diagram of the socket parameters;

[0035] Figure 3 Schematic diagram of the fitting of the pore cone. DETAILED DESCRIPTION

[0036] The following detailed description is illustrative and is intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used in the present invention have the same meaning as commonly understood by those skilled in the art to which the present invention belongs.

[0037] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present invention. As used herein, unless otherwise expressly specified in the present invention, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, they indicate the presence of features, steps, operations, devices, components and / or combinations thereof.

[0038] For the convenience of description, if the words "up", "down", "left" and "right" appear in the present invention, they only indicate that they are consistent with the up, down, left and right directions of the drawings themselves, and do not limit the structure. They are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or component referred to must have a specific orientation, be constructed and operate in a specific orientation. Therefore, they should not be understood as limiting the present invention.

[0039] Explanation of terms: The terms "install", "connect", "connect", "fix" and the like in the present invention should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral whole; it can be a mechanical connection, an electrical connection, a direct connection, or an indirect connection through an intermediate medium, an internal connection between two elements, or an interaction relationship between two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to the specific circumstances.

[0040] Example 1:

[0041] A method for detecting socket parameters in this embodiment is as follows: Figure 1 As shown, the realization of the 3D point cloud model of the socket obtained based on camera photography includes the following steps:

[0042] Step 1: Calculate the normal vector of the three-dimensional point in the three-dimensional point cloud of the socket and correct the normal vector;

[0043] Step 2: extracting 3D points from the 3D point cloud of the cavities to fit the cavities cone, and evaluating the error between the fitted cavities cone and the theoretical cavities cone;

[0044] Step 3: Taking the error as the reference, screen and remove the three-dimensional points whose fitting error exceeds the standard value;

[0045] Step 4: For the three-dimensional points with qualified fitting error, select other three-dimensional points within the neighborhood range of the three-dimensional points;

[0046] Step 5: Construct a candidate set with the selected three-dimensional points, and solve the optimal candidate set with the goal of the largest number of three-dimensional points in the candidate set and the minimum error between the fitting socket hole cone and the theoretical socket hole cone fitted by the three-dimensional points in the candidate set;

[0047] Step 6: Fit the socket hole cone through the three-dimensional points in the optimal candidate set, and separate the parameters of the socket hole through the fitted socket hole cone.

[0048] Extract three-dimensional points on the fitted socket hole cone, project the current three-dimensional points onto the theoretical socket hole cone to obtain projection points. The error includes distance error and angle error. The distance error is the distance between the three-dimensional point and the corresponding projection point, and the angle error is the angle between the normal direction of the three-dimensional point and the normal direction of the corresponding projection point.

[0049] The optimal candidate set is:

[0050] p φ ={p i 丨p i ∈P∧d(φ,p i )<ε∧arccos (|n(p i )·n(φ,p i )|)<α};

[0051] Where: p φ represents the candidate set; p i represents the i-th three-dimensional point in the candidate set; P represents the socket hole three-dimensional point cloud; d(φ,p i ) represents the distance between the three-dimensional point and the projection point; n(p i ) represents the normal direction of the three-dimensional point; n(φ,p i ) represents the normal direction of the projection point; ε represents the distance calibration value; α represents the angle calibration value.

[0052] Divide the three-dimensional points in the socket hole three-dimensional point cloud into several non-overlapping three-dimensional point subsets, establish a scoring function, and score the three-dimensional point subsets based on the candidate set. The scoring function is:

[0053] σ si =-1-f(-2-|s1|,-2-|p|,-1-σ s1 );

[0054] Where: |s1| is the number of 3D points in the first 3D point subset; |p| is the number of 3D points in the 3D point cloud of the pore; σ s1 is the score of the first 3D point subset; σ si is the score of the other i-th 3D point subset; f(-2-|s1|,-2-

[0055] |p|,-1-σ s1 ) represents the hypergeometric distribution with respect to the variable.

[0056] In step 2, three non-coplanar three-dimensional points are extracted from the cellular three-dimensional point cloud, the cone intersection is solved through the normal plane where the three non-coplanar three-dimensional points are located, and the cellular cone is fitted based on the three non-coplanar three-dimensional points and the cone intersection; the cone axis is solved through the three non-coplanar three-dimensional points and the cone intersection, the cone center angle is solved through the three non-coplanar three-dimensional points and the cone intersection, and the cellular cone is fitted through the cone axis and the center angle.

[0057] In step 2, an octree structure is established for the 3D points. Based on the layer depth of the octree structure, the remaining 3D points are selected within the neighborhood of the current 3D point, and the probability that all the selected 3D points belong to the same pore cone is calculated. The calculation formula for the probability that the 3D points belong to the same pore cone is:

[0058]

[0059] Where: p local (n) represents the probability that the n selected 3D points belong to the same pore cone; n represents the number of selected 3D points; N represents the number of 3D points in the pore 3D point cloud; d represents the hierarchical depth of the octree structure; k d Indicates the depth of the current 3D point in the octree structure.

[0060] The rest of this embodiment is the same as that of embodiment 1, so it will not be described again.

[0061] Example 2:

[0062] This embodiment provides a method for detecting fossa parameters. Based on a 3D fossa point cloud, the fossa reference plane and fossa cone are separated. The normal vector of the reference plane is parallel to the camera's imaging axis. To obtain the fossa cone, the normal vector of each 3D point in the fossa point cloud is first calculated.

[0063] Specifically, first calculate the covariance matrix of each three-dimensional point. The formula is as follows:

[0064]

[0065] Where: k represents the number of three-dimensional points; p irepresents the i-th three-dimensional point; represents the average value of the three-dimensional points; T represents the transpose of the matrix.

[0066] The eigenvector is calculated by the covariance matrix, and then the normal vector of the 3D point is calculated by the eigenvector. The reference plane and the pore cone are separated according to the normal vector of the 3D point. The RANSAC algorithm is used to find the optimal candidate set, and the pore cone is obtained by fitting the 3D points in the optimal candidate set, and then the parameters of the pore are obtained, such as Figure 2 As shown, the cavity parameters include cavity depth H, cavity center angle w, cavity diameter D, etc.

[0067] Specifically:

[0068] Calculate the normal vectors of 3D points using principal component analysis. Given a set of 3D point clouds P, where P = {p1, p2, …pn}, where p1, p2, …pn represent n 3D points, establish corresponding normal vectors {l1, l2, …ln} for each of these n 3D points. Select a subset of 3D points from the 3D point cloud to form a set of 3D point sets G, where G = {G1, G2, …Gn}; G1, G2, …Gn represent randomly selected 3D point sets. Perform 3D cone fitting on the 3D points in these sets to obtain a set of 3D cones M = {M1, M2, …Mn}, where M1, M2, …Mn represent the fitted 3D cones.

[0069] Evaluate the error between the fitted cell cone and the theoretical cell cone, specifically:

[0070] The selected 3D points are used to construct a candidate set. The optimal candidate set is determined by maximizing the number of 3D points in the candidate set and minimizing the error between the fitted pore cone and the theoretical pore cone. A 3D point is extracted from the fitted pore cone and projected onto the theoretical pore cone to obtain a projection point. The error includes distance error, which is the distance between the 3D point and the corresponding projection point, and angle error, which is the angle between the normal of the 3D point and the normal of the corresponding projection point.

[0071] The candidate set is:

[0072] p φ ={p i |p i ∈P∧d(φ,p i )<ε∧arccos(|n(p i )·n(φ,p i )|)<α};

[0073] Where: p φrepresents the candidate set; p i represents the i-th 3D point in the candidate set; P represents the 3D point cloud of the pore; d(φ,p i ) represents the distance between the three-dimensional point and the projection point; n(p i ) represents the normal direction of a three-dimensional point; n(φ,p i ) represents the normal direction of the projection point; ε represents the distance calibration value; α represents the angle calibration value.

[0074] To ensure that the selected 3D points meet the optimal candidate set criteria, sampling must be performed within the neighborhood of the 3D point. Furthermore, to ensure that the error in the selected 3D points is not too large, the maximum distance between the 3D point and the projected point must not exceed ε, and the angle between the normal of the 3D point and the normal of the projected point must not exceed α.

[0075] For the pore cone, if Figure 3 As shown, three non-coplanar 3D points and their normal vectors are selected to define the pore cone. The vertex position c of the pore cone is the intersection of the cone planes where the three 3D points are located. The plane normal is obtained through the three 3D points and the vertex position:

[0076] Where c represents the position of the vertex; p1, p2, and p3 represent the positions of any three non-coplanar three-dimensional points; ||p1-c|| represents the Euclidean distance between the current three-dimensional point and the cone vertex.

[0077] Then we can get the axis γ of the pore cone and the pore center angle w of the pore cone:

[0078] where i = 1, 2, 3; p i Represents the position of the i-th 3D point.

[0079] The time complexity of the pore cone fitting process is primarily determined by the number of 3D points required for detection and the time required to evaluate the scores of the 3D points in the candidate set. To obtain the optimal pore cone, it is necessary to increase the probability that the selected 3D points belong to the same pore cone set. Since the 3D points in a pore cone set are only a subset of the entire pore 3D point cloud, the smaller the distance between two selected 3D points, the higher the probability that they belong to the same pore cone. An octree structure is constructed for each 3D point to search for other 3D points within the 3D point's neighborhood.

[0080] After selecting the first 3D point p1, the neighborhood range N of the 3D point is nei Select other n-1 N nei , and these N nei The probability of belonging to a cell cone is:

[0081]

[0082] Where: p local (n) represents the probability that the n selected 3D points belong to the same pore cone; n represents the number of selected 3D points; N represents the number of 3D points in the pore 3D point cloud; d represents the hierarchical depth of the octree structure; k d Indicates the depth of the current 3D point in the octree structure.

[0083] When evaluating the scores of the candidate set, it is necessary to calculate the distance and normal angle between each 3D point in the pore 3D point cloud set and the projection point. In order to further reduce the detection time, when evaluating the scores of the candidate set, the pore 3D point cloud set P is divided into non-overlapping subsets, that is, P = {S1, S2…S r}, each subset contains several 3D points to reduce the number of calculations required during the evaluation process, and an octree is generated for each subset.

[0084] Use the candidate set to score the subset S1, and the score of the subset S1 is σ s1 , then the scores of other subsets can be inferred by the following formula:

[0085] σ si =-1-f(-2-|s1|,-2-|p|,-1-σ s1 );

[0086] Where: |s1| is the number of 3D points in the first 3D point subset; |p| is the number of 3D points in the 3D point cloud of the pore; σ s1 is the score of the first 3D point subset; σ si is the score of the other i-th 3D point subset; f(-2-|s1|,-2-|p|,-1-σ s1 ) represents the hypergeometric distribution with respect to the variable.

[0087] The selected 3D points are used to construct a candidate set. The goal is to find the optimal candidate set, with the largest number of 3D points (i.e., the highest score) and the smallest error between the fitted cavity cone and the theoretical cavity cone. A cavity cone is fitted using the 3D points in the optimal candidate set, and the cavity parameters are calculated using the fitted cavity cone.

[0088] The rest of this embodiment is the same as that of embodiment 1, so it will not be described again.

[0089] The above are only preferred embodiments of the present invention and do not limit the present invention in any form. Any simple modifications or equivalent changes made to the above embodiments based on the technical essence of the present invention fall within the scope of protection of the present invention.

Claims

1. A method for detecting fossa parameters, based on a 3D point cloud model of fossa obtained by camera photography, characterized in that: The following steps are involved: Step 1: Calculate the normal vector of the three-dimensional point in the three-dimensional point cloud of the socket and correct the normal vector; Step 2: extracting 3D points from the 3D point cloud of the cavities to fit the cavities cone, and evaluating the error between the fitted cavities cone and the theoretical cavities cone; Step 3: Based on the error, the three-dimensional points whose fitting error exceeds the standard value are screened and eliminated; Step 4: For the 3D point whose fitting error meets the standard, select other 3D points in the neighborhood of the 3D point; Step 5: Construct a candidate set from the selected 3D points, and determine the optimal candidate set with the goal of maximizing the number of 3D points in the candidate set and minimizing the error between the fitted pore cone of the 3D points in the candidate set and the theoretical pore cone. Step 6: Fitting the cellular cone using the 3D points in the optimal candidate set, and calculating the parameters of the cellular cone using the fitted cellular cone; extracting a 3D point on the fitted cellular cone, and projecting the current 3D point onto the theoretical cellular cone to obtain a projection point. The error includes a distance error and an angle error. The distance error is the distance between the 3D point and the corresponding projection point, and the angle error is the angle between the normal of the 3D point and the normal of the corresponding projection point. The candidate set is: Where: p φ represents the candidate set; p i represents the i-th 3D point in the candidate set; P represents the 3D point cloud of the pore; d(φ,p i ) represents the distance between the three-dimensional point and the projection point; n(p i ) represents the normal direction of a three-dimensional point; n(φ,p i ) represents the normal of the projection point; ε represents the distance calibration value; α represents the angle calibration value; the 3D points in the 3D point cloud of the cellular cavity are divided into several non-intersecting 3D point subsets, a scoring function is established, and the 3D point subsets are scored based on the candidate set; The scoring function is: σ si =-1-f(-2-|s1|,-2-|p|,-1-σ s1 ); Where: |s1| is the number of 3D points in the first 3D point subset; |p| is the number of 3D points in the 3D point cloud of the pore; σ s1 is the score of the first 3D point subset; σ si is the score of the other i-th 3D point subset; f(-2-|s1|,-2-|p|,-1-σ s1 ) represents the hypergeometric distribution with respect to the variable.

2. A method for detecting socket parameters according to claim 1, characterized in that: In step 2, three non-coplanar three-dimensional points are extracted from the cellular three-dimensional point cloud, the cone intersection is solved through the normal plane where the three non-coplanar three-dimensional points are located, and the cellular cone is fitted based on the three non-coplanar three-dimensional points and the cone intersection; the cone axis is solved through the three non-coplanar three-dimensional points and the cone intersection, the cone center angle is solved through the three non-coplanar three-dimensional points and the cone intersection, and the cellular cone is fitted through the cone axis and the center angle.

3. A method for detecting socket parameters according to claim 2, characterized in that: In step 2, an octree structure is established for the three-dimensional points. Based on the hierarchical depth of the octree structure, the remaining three-dimensional points are selected within the neighborhood of the current three-dimensional point, and the probability that all the selected three-dimensional points belong to the same pore cone is calculated.

4. A method for detecting socket parameters according to claim 3, characterized in that: The calculation formula for the probability that the three-dimensional points belong to the same pore cone is: Where: p local (n) represents the probability that the n selected 3D points belong to the same pore cone; n represents the number of selected 3D points; N represents the number of 3D points in the pore 3D point cloud; d represents the hierarchical depth of the octree structure; k d Indicates the depth of the current 3D point in the octree structure.

Images