Defect detection method of anti-scaling and anti-corrosion casing based on three-dimensional point cloud data

Through the method based on three-dimensional point cloud data, local space and sphere space are constructed, density and laser reflection intensity are calculated, and artificial dependence and randomness problems in anti-scaling and anti-corrosion casing detection are solved, improving the accuracy and credibility of the detection.

CN118446974BActive Publication Date: 2024-11-15QINGDAO JIANTAI ANTI-FRICTION SEALING MATERIALS CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410523984.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-04-29
Publication Date
2024-11-15
Estimated Expiration
2044-04-29

AI Technical Summary

Technical Problem

The existing anti-scaling and anti-corrosion casing detection methods rely on manual judgment and traditional non-destructive detection, which have problems such as inefficiency in subjectivity and inability to cover all the time, and the randomness of the split value of isolated forests leads to unstable detection results.

Method used

Using a method based on three-dimensional point cloud data, by constructing local space and sphere space, the density degree of data points and laser reflection intensity are calculated, the degree of defects are obtained, and the abnormal score is calculated to determine casing defects, which improves the accuracy of isolated forest splitting values.

Benefits of technology

It improves the accuracy and credibility of the detection, reduces the instability of the result caused by random splitting values, and enhances the detection ability of cannula defects.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118446974B_ABST
    Figure CN118446974B_ABST
Patent Text Reader

Abstract

The present invention relates to the field of image data processing, and in particular to a method for detecting defects in anti-scaling and anti-corrosion casings based on three-dimensional point cloud data, comprising: obtaining three-dimensional point cloud data of the anti-scaling and anti-corrosion casings, obtaining the first degree of defects of each data point according to the number, distance and density of data points in the local space of each data point, obtaining the second degree of defects of each data point according to the laser reflection intensity of each data point at different laser detection angles, the number and distance of data points in the local space, obtaining the defect parameter index of each data point according to the first degree and second degree of defects of each data point, and then obtaining the abnormal score of each data point, and when the abnormal score of any data point is greater than a preset abnormal threshold, it is determined that there is a defect in the anti-scaling and anti-corrosion casing. The present invention improves the accuracy of detecting abnormal data by the isolated forest by obtaining the optimal splitting value of each split of the isolated forest.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of computer vision, and in particular to a method for detecting defects of anti-scaling and anti-corrosion casing based on three-dimensional point cloud data. Background Art

[0002] Detect and diagnose the defects of anti-scaling and anti-corrosion casing in the pipeline industry. At present, the pipeline industry mainly relies on manual inspections and traditional non-destructive detection methods such as ultrasonic detection and X-ray detection for the detection of anti-scaling and anti-corrosion casing. However, these methods have certain limitations, such as reliance on the subjectivity of manual judgment, low work efficiency, and inability to fully cover. During the production process, anti-scaling and anti-corrosion casings often have performance defects such as unevenness. When anomalies are detected on the three-dimensional point cloud data of the casing through isolation forests, the selection of split values ​​by isolation forests is random, which will make it overly dependent on randomness and may lead to instability in the final detection results. At the same time, the random selection of split values ​​may be sensitive to certain characteristics of the data distribution, and the model may not be able to effectively capture the true structure of the data due to improper selection of split values. This will affect the accuracy of the final detection of casing defects. Summary of the invention

[0003] The invention provides an anti-scaling and anti-corrosion casing defect detection method based on three-dimensional point cloud data to solve the existing problems.

[0004] The anti-scaling and anti-corrosion casing defect detection method based on three-dimensional point cloud data of the present invention adopts the following technical scheme:

[0005] An embodiment of the present invention provides a method for detecting defects in anti-scaling and anti-corrosion casing based on three-dimensional point cloud data, the method comprising the following steps:

[0006] Obtain three-dimensional point cloud data of anti-scaling and anti-corrosion casing;

[0007] Constructing a local space corresponding to each data point in the three-dimensional point cloud data; constructing a spherical space corresponding to each data point in the three-dimensional point cloud data, and obtaining the density of each data point according to the spherical space corresponding to each data point; obtaining the first degree of defect in the local space of each data point in the three-dimensional point cloud data according to the number of data points, the distance between the data points and the density;

[0008] Obtain the laser reflection intensity of each data point in the three-dimensional point cloud data at different laser detection angles; obtain the second degree of defect in the local space of each data point according to the laser reflection intensity of each data point in the three-dimensional point cloud data at different laser detection angles, the number of data points in the local space, and the distance between the data points;

[0009] Obtaining a defect parameter index for each data point according to a first degree and a second degree of a defect in a local space of each data point;

[0010] According to the defect parameter index of each data point in the three-dimensional point cloud data, an abnormality score of each data point in the three-dimensional point cloud data is obtained;

[0011] When the anomaly score of any data point in the three-dimensional point cloud data is greater than the preset anomaly threshold, it is determined that there is a defect in the anti-scaling and anti-corrosion casing.

[0012] Furthermore, the local space corresponding to each data point is constructed in the three-dimensional point cloud data, and the specific steps include the following:

[0013] In the three-dimensional point cloud data, a local space of size a×a×a is constructed with each data point as the center, where a is the preset first space side length.

[0014] Furthermore, a spherical space corresponding to each data point is constructed in the three-dimensional point cloud data, and the density of each data point is obtained according to the spherical space corresponding to each data point. The specific steps include the following:

[0015] In the three-dimensional point cloud data, a spherical space with a radius of r is constructed with each data point as the center, where r is the preset second space radius;

[0016] The ratio of the number of data points in the spherical space corresponding to each data point to the volume of the spherical space is recorded as the density of each data point.

[0017] Furthermore, according to the number of data points in the local space where each data point in the three-dimensional point cloud data is located, the distance between the data points and the density, the first degree of defect in the local space of each data point is obtained, and the corresponding specific calculation formula is:

[0018]

[0019] Among them, Q i represents the first degree of defect in the local space of the i-th data point; N represents the number of data points other than the i-th data point in the local space of the i-th data point; c i,n represents the density of the nth data point in the local space of the i-th data point except the i-th data point; Δd(i,n) represents the Euclidean distance between the i-th data point and the n-th data point in the local space of the i-th data point except the i-th point; Δc(i,n) represents the absolute value of the difference in density between the i-th data point and the n-th data point in the local space of the i-th data point except the i-th point; Indicates that when n belongs to the interval [1,N], all c i,n The maximum value in ; Indicates that when n belongs to the interval [1,N], all c i,n The minimum value in; N' represents the number of data points in the local space of the i-th data point; c' i,n' Indicates the density of the n'th data point in the local space of the ith data point; Indicates that when n' is in the interval [1, N'], all c' i,n' The standard deviation of ; || represents the absolute value function; norm() represents the normalization function.

[0020] Further, according to the laser reflection intensity of each data point in the three-dimensional point cloud data at different laser detection angles, the number of data points in the local space, and the distance between the data points, the second degree of defect in the local space of each data point is obtained, and the specific steps included are as follows:

[0021] In the local space of the ith data point corresponding to the jth laser detection angle, with the distance from each data point to the ith data point as the horizontal axis and the laser reflection intensity of each data point as the vertical axis, all data points are curve fitted using the least squares method to obtain the fitting curve corresponding to the ith data point corresponding to the jth laser detection angle;

[0022] Calculate the second-order derivative of the fitting curve corresponding to the ith data point corresponding to the jth laser detection angle, and obtain the extreme point in the fitting curve corresponding to the ith data point corresponding to the jth laser detection angle;

[0023] Among all laser detection angles, any two non-repeating laser detection angles form a combination to obtain several combinations;

[0024] The second degree of defect in the local space of each data point is obtained based on the laser reflection intensity of each data point at different laser detection angles, the number of data points in the local space, and the extreme points in the fitting curve corresponding to each data point corresponding to the two laser detection angles in the combination.

[0025] Furthermore, according to the laser reflection intensity of each data point at different laser detection angles, the number of data points in the local space, and the extreme value points in the fitting curve corresponding to each data point corresponding to the two laser detection angles in the combination, the second degree of defect in the local space of each data point is obtained, and the corresponding specific calculation formula is:

[0026]

[0027] Among them, W i represents the second degree of defect in the local space of the i-th data point; L represents the number of laser detection angles; σ i,lrepresents the standard deviation of the laser reflection intensity of all data points in the local space of the i-th point in the 3D point cloud data corresponding to the l-th laser detection angle; N' represents the number of data points in the local space of the i-th data point; Δq i,l,n' represents the absolute value of the difference in laser reflection intensity between the n'th data point and its nearest data point in the local space of the ith point in the 3D point cloud data corresponding to the lth laser detection angle; B represents the number of combinations; G b,i represents the minimum value of the number of extreme points in the fitting curve corresponding to the ith data point corresponding to the two laser detection angles in the bth combination; Δd i,b,g Δq represents the absolute value of the difference between the horizontal coordinates of the gth extreme point in the fitting curve corresponding to the i-th data point corresponding to the two laser detection angles in the b-th combination; i,b ' ,g It represents the absolute value of the difference in laser reflection intensity at the g-th extreme point in the fitting curve corresponding to the ith data point corresponding to the two laser detection angles in the b-th combination.

[0028] Further, according to the first degree and the second degree of the defect in the local space of each data point, the defect parameter index of each data point is obtained, and the specific steps include the following:

[0029] The product of the first degree of defect in the local space of the i-th data point and the second degree of defect in the local space of the i-th data point is recorded as the defect parameter index of the i-th data point.

[0030] Further, according to the defect parameter index of each data point in the three-dimensional point cloud data, the abnormality score of each data point in the three-dimensional point cloud data is obtained, and the specific steps included are as follows:

[0031] According to the defect parameter index of each data point in the three-dimensional point cloud data, the possibility that the defect parameter index of each data point at the first splitting belongs to the optimal splitting value is obtained;

[0032] In the three-dimensional point cloud data, the defect parameter index of the data point corresponding to the possibility that the largest defect parameter index belongs to the optimal splitting value is recorded as the optimal splitting value of the first splitting;

[0033] According to the optimal splitting value of the first splitting and the defect parameter index of each data point in the three-dimensional point cloud data, the optimal splitting value of each splitting of the three-dimensional point cloud data is obtained;

[0034] According to the optimal split value of each split of the three-dimensional point cloud data and the defect parameter index of each data point in the three-dimensional point cloud data, the abnormal score of each data point in the three-dimensional point cloud data is obtained through the isolation forest algorithm.

[0035] Furthermore, according to the defect parameter index of each data point in the three-dimensional point cloud data, the possibility that the defect parameter index of each data point at the first splitting belongs to the optimal splitting value is obtained, and the corresponding specific calculation formula is:

[0036]

[0037] Among them, F(1,y) represents the possibility that the defect parameter index of the yth data point at the first split belongs to the optimal split value; It represents the mean value of the defect parameter index of all data points in the three-dimensional point cloud data that are smaller than the defect parameter index of the y-th data point at the first split; represents the mean value of the defect parameter index of all data points greater than or equal to the defect parameter index of the yth data point in the three-dimensional point cloud data at the first split; J y Indicates the number of data points in the three-dimensional point cloud data that are smaller than the defect parameter index of the yth data point during the first split; s1 j,y Indicates the defect parameter index of the jth data point in the three-dimensional point cloud data that is smaller than the defect parameter index of the yth data point during the first split; J' y Indicates the number of data points of defect parameter index greater than or equal to the yth data point in the three-dimensional point cloud data at the first split; s2 j ' ,y It represents the defect parameter index of the j'th data point that is greater than or equal to the defect parameter index of the yth data point in the three-dimensional point cloud data at the first split, and || represents the absolute value function.

[0038] Further, according to the optimal splitting value of the first splitting and the defect parameter index of each data point in the three-dimensional point cloud data, the optimal splitting value of each splitting of the three-dimensional point cloud data is obtained, and the specific steps included are as follows:

[0039] According to the optimal split value of the first split and the defect parameter index of each data point in the three-dimensional point cloud data, the isolation forest algorithm is used to obtain the first split result of the three-dimensional point cloud data;

[0040] According to the method for obtaining the optimal splitting value of the first splitting, the optimal splitting value of the second splitting is obtained in the first splitting result of the three-dimensional point cloud data;

[0041] According to the first splitting result of the three-dimensional point cloud data and the optimal splitting value of the second splitting, the isolation forest algorithm is used to obtain the second splitting result of the three-dimensional point cloud data;

[0042] By analogy, the optimal splitting value for each splitting of the three-dimensional point cloud data is obtained.

[0043] The beneficial effects of the technical solution of the present invention are:

[0044] In the embodiment of the present invention, three-dimensional point cloud data of the anti-scaling and anti-corrosion casing is obtained, a local space corresponding to each data point is constructed in the three-dimensional point cloud data, a spherical space corresponding to each data point is constructed in the three-dimensional point cloud data, the density of each data point is obtained through the spherical space corresponding to each data point, and the first degree of defect of the local space of each data point is obtained according to the number of data points in the local space where each data point in the three-dimensional point cloud data is located, the distance between the data points and the density, the laser reflection intensity of each data point in the three-dimensional point cloud data at different laser detection angles is obtained, and the laser reflection intensity of each data point in the three-dimensional point cloud data at different laser detection angles is obtained. The second degree of defects in the local space of each data point is obtained by calculating the strength, the number of data points in the local space, and the distance between the data points. The defect parameter index of each data point is obtained according to the first degree and the second degree of defects in the local space of each data point. The defect parameter index of each data point is calculated to improve the accuracy of the optimal splitting value of each split of the isolated forest, thereby improving the accuracy of the detection of abnormal data by the isolated forest. According to the defect parameter index of each data point in the three-dimensional point cloud data, the abnormal score of each data point in the three-dimensional point cloud data is obtained. According to the abnormal score of each data point in the three-dimensional point cloud data, it is confirmed whether the anti-scaling and anti-corrosion casing has defects. The present invention improves the accuracy of the detection of abnormal data by the isolated forest by reducing the problems of unstable and inaccurate results caused by randomly selecting the segmentation value of the isolated forest, thereby improving the credibility of the final defect area result. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0046] Figure 1 It is a flow chart of the steps of the anti-scaling and anti-corrosion casing defect detection method based on three-dimensional point cloud data of the present invention. DETAILED DESCRIPTION

[0047] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the anti-scaling and anti-corrosion casing defect detection method based on three-dimensional point cloud data proposed by the present invention, its specific implementation method, structure, features and effects, in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0048] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0049] The specific scheme of the anti-scaling and anti-corrosion casing defect detection method based on three-dimensional point cloud data provided by the present invention is described in detail below with reference to the accompanying drawings.

[0050] See also Figure 1 , which shows a flowchart of a method for detecting anti-scaling and anti-corrosion casing defects based on three-dimensional point cloud data provided by an embodiment of the present invention, the method comprising the following steps:

[0051] Step S001: Acquire the three-dimensional point cloud data of the anti-scaling and anti-corrosion casing.

[0052] It should be noted that the three-dimensional point cloud data of the newly produced anti-scaling and anti-corrosion casing is collected from several laser detection angles through LiDAR laser detection and ranging to obtain three-dimensional coordinates and laser reflection intensity. This intensity information is related to the surface material, roughness, incident angle direction, and emission energy and laser wavelength of the anti-scaling and anti-corrosion casing. The three-dimensional point cloud data corresponding to the anti-scaling and anti-corrosion casing at each laser detection angle is thus obtained for analysis in subsequent processes. Among them, each data point in the three-dimensional point cloud data corresponds to a three-dimensional coordinate. When the laser detection angle is changed, the three-dimensional coordinate corresponding to the data point remains unchanged, but the laser reflection intensity corresponding to the data point will change with the change of the laser detection angle.

[0053] Step S002: construct a local space corresponding to each data point in the three-dimensional point cloud data; construct a spherical space corresponding to each data point in the three-dimensional point cloud data, and obtain the density of each data point according to the spherical space corresponding to each data point; obtain the first degree of defect of the local space of each data point according to the number of data points in the local space where each data point in the three-dimensional point cloud data is located, the distance between the data points and the density.

[0054] It should be noted that for all data points, their concave-convex defect characteristics in the local space can be analyzed first. Specifically, by analyzing the point cloud density and distribution uniformity in the local area, possible defects can be identified. Since the casing itself has an overall regular shape and symmetry, any defect will cause abnormal point cloud density and distribution uniformity in the local area. In addition, the laser reflection intensity at different angles will also produce more abnormal changes in the concave-convex defect area.

[0055] Specifically, taking the ith data point in the three-dimensional point cloud data as an example, the first spatial side length a=13 and the second spatial radius r=3 are preset in this embodiment. This is described as an example. Other implementations may be set to other values, which are not limited in this embodiment. In the three-dimensional point cloud data, with the ith data point as the center, a local space of the ith data point with a size of a×a×a is constructed. In the local space of the ith data point, the consistency of the local density of the points in the local space is analyzed.

[0056] Furthermore, in the three-dimensional point cloud data, a spherical space with a radius of the second space radius r is constructed with the i-th data point as the center. The ratio of the number of data points in the spherical space to the volume of the spherical space is recorded as the density of the i-th data point; according to the above method, the local space and density of each data point in the three-dimensional point cloud data are obtained.

[0057] Thus, the first degree of defect in the local space of each data point based on point cloud density analysis can be obtained.

[0058] The calculation formula is as follows:

[0059]

[0060] Among them, Q i represents the first degree of defect in the local space of the i-th data point; N represents the number of data points other than the i-th data point in the local space of the i-th data point; c i,n represents the density of the nth data point in the local space of the i-th data point except the i-th data point; Δd(i,n) represents the Euclidean distance between the i-th data point and the n-th data point in the local space of the i-th data point except the i-th point; Δc(i,n) represents the absolute value of the difference between the density of the i-th data point and the n-th data point in the local space of the i-th data point except the i-th point; Indicates that when n belongs to the interval [1,N], all c i,n The maximum value in ; Indicates that when n belongs to the interval [1,N], all c i,n The minimum value in; N' represents the number of data points in the local space of the i-th data point; c' i,n' Indicates the density of the n'th data point in the local space of the ith data point; Indicates that when n' is in the interval [1, N'], all c' i,n' The standard deviation of ; || represents the absolute value function; norm() is a linear normalization function, which means the normalization based on all data points of the three-dimensional point cloud data.

[0061] Specifically, It means that in the local space of the i-th data point, each data point is compared and analyzed with the i-th data point one by one. The closer the distance between the two and the smaller the density difference, the lower the probability that the two are concave-convex defect areas, and the smaller the first degree of the corresponding final defect; at the same time, It is corrected by two extreme values, because when the ratio of the maximum to the minimum density in the local space of the i-th data point is smaller (closer to 1), the corresponding density of the remaining points will naturally be closer, so the overall density in the local space is closer, and the probability of the existence of concave and convex defect areas is lower. It means that the uniformity of data point density distribution is reflected by the standard deviation of the density of all data points in the local space. If there are concave and convex areas in the local space, the density of different areas in the space is inconsistent. Therefore, the smaller the standard deviation after normalization, the better the uniformity of the density distribution of the current i-th data point in the local space in the overall three-dimensional point cloud data, and the smaller the probability of the existence of concave and convex defect areas.

[0062] Thus, the first degree of defect in the local space of each data point based on point cloud density analysis can be obtained.

[0063] Step S003: Obtain the laser reflection intensity of each data point in the three-dimensional point cloud data at different laser detection angles; obtain the second degree of defect in the local space of each data point based on the laser reflection intensity of each data point in the three-dimensional point cloud data at different laser detection angles, the number of data points in the local space, and the distance between the data points.

[0064] For each data point in the three-dimensional point cloud data of the anti-scaling and anti-corrosion casing, multiple laser detection angles are used to detect, and the laser reflection intensity of each data point in the three-dimensional point cloud data corresponding to each laser detection angle is obtained.

[0065] It should be noted that the defects of the local space determined only through the analysis of the density in the local space are relatively one-sided. Because the existence of the concave-convex defect area will not only affect the density of the local space, but also cause differences in the laser reflection intensity of each data point in the local space, especially the laser reflection intensity of each data point in the local space obtained at different angles. Due to the concave-convex defect area, the laser reflection intensity of each data point at different angles will show different changes. Therefore, the analysis of the laser reflection intensity in the local space can be combined with the subsequent analysis of the density to determine the defect performance that is more suitable for the local space.

[0066] Specifically, taking the local space of the ith data point corresponding to the jth laser detection angle as an example, in the local space of the ith data point corresponding to the jth laser detection angle, the laser reflection intensity of all data points is traversed from near to far starting from the distance from the center point to each of the remaining data points. In the local space of the ith data point corresponding to the jth laser detection angle, the distance from each data point to the ith data point is used as the horizontal axis, and the laser reflection intensity of each data point is used as the vertical axis. All data points are curve fitted using the least squares method to obtain the fitting curve corresponding to the ith data point corresponding to the jth laser detection angle. Calculate the second-order derivative of the fitting curve corresponding to the ith data point corresponding to the jth laser detection angle, obtain the extreme point in the fitting curve corresponding to the ith data point corresponding to the jth laser detection angle and mark it. The least squares method and the second-order derivative calculation method are both well-known methods, and the specific methods will not be introduced in detail here. Among all laser detection angles, any two non-repeating laser detection angles form a combination to obtain several combinations.

[0067] Thus, the second degree of defect in the local space of each data point based on the laser reflection intensity at different laser detection angles can be obtained.

[0068] The calculation formula is as follows:

[0069]

[0070] Among them, W i represents the second degree of defect in the local space of the i-th data point; L represents the number of laser detection angles; σ i,l represents the standard deviation of the laser reflection intensity of all data points in the local space of the i-th point in the 3D point cloud data corresponding to the l-th laser detection angle; N' represents the number of data points in the local space of the i-th data point; Δq i,l,n' represents the absolute value of the difference in laser reflection intensity between the n'th data point and its nearest data point in the local space of the ith point in the 3D point cloud data corresponding to the lth laser detection angle; B represents the number of combinations; G b,i represents the minimum value of the number of extreme points in the fitting curve corresponding to the ith data point corresponding to the two laser detection angles in the bth combination; Δd i,b,g Δq represents the absolute value of the difference between the horizontal coordinates of the gth extreme point in the fitting curve corresponding to the i-th data point corresponding to the two laser detection angles in the b-th combination; i,b ' ,g It represents the absolute value of the difference in laser reflection intensity at the g-th extreme point in the fitting curve corresponding to the ith data point corresponding to the two laser detection angles in the b-th combination.

[0071] It should be noted that the calculation of Δq i,l,n'When there are multiple nearest data points for the n'th data point, the average of the absolute values ​​of the laser reflection intensity differences between the n'th data point and the multiple nearest data points is taken.

[0072] Specifically, represents the performance of the laser reflection intensity in the local space of the i-th data point at different laser detection angles, where σ i,l It represents the standard deviation of laser reflection intensity at all laser detection angles. It is analyzed from the overall perspective. The smaller the value, the more stable the laser reflection intensity at each laser detection angle, and the smaller the possibility of defects in concave and convex areas. It represents the difference in laser reflection intensity between adjacent points at each laser detection angle. From a local analysis, the smaller the difference in laser reflection intensity between adjacent points, the smaller the possibility of the existence of concave-convex defect areas. The two are combined to jointly determine the comprehensive performance of the laser reflection intensity at different angles in the local space of the i-th data point from the overall and local aspects. It represents the abnormal performance analysis of the laser reflection intensity at different angles. The laser reflection intensities at all laser detection angles in the current local space are combined in pairs for comparative analysis. The distance between the extreme points with the same number refers to the smaller the difference in the horizontal coordinates of the extreme points in the curve and the smaller the difference in the laser reflection intensity of the extreme points with the same number, which means that the laser reflection intensity at the two laser detection angles corresponding to the combination is less affected by different acquisition angles, which further indicates that the abnormal laser reflection intensity of the local space of the current i-th data point at different acquisition angles changes little, and the possibility of concave-convex defects is small.

[0073] Thus, the second degree of defect in the local space of each data point based on the laser reflection intensity at different laser detection angles can be obtained.

[0074] Step S004: Obtain a defect parameter index for each data point according to the first degree and the second degree of defects in the local space of each data point.

[0075] It should be noted that due to the existence of concave-convex defect areas, the density of data points and the laser reflection intensity in the local space will be significantly different from those in the normal area. Therefore, the first degree and the second degree of defects in the local space of each data point can be combined. This can better determine the defect performance in the local space and improve the confidence in the defect performance in the local space.

[0076] The calculation formula is as follows:

[0077] E i =Q i ×W i

[0078] Among them, Ei represents the defect parameter index of the i-th data point; Q i represents the first degree of defect in the local space of the i-th data point; W i Represents the second degree of defect in the local space of the ith data point.

[0079] Thus, the defect parameter indicators of all data points are obtained and used as input data of the isolation forest for subsequent operations.

[0080] Step S005: Obtain an abnormality score for each data point in the three-dimensional point cloud data according to the defect parameter index of each data point in the three-dimensional point cloud data.

[0081] It should be noted that in the traditional isolation forest splitting, the splitting value is randomly selected, and an optimal value is compared through traversal analysis before each split. The existence of this optimal value minimizes the difference between the two categories after the split and maximizes the difference between the two categories.

[0082] According to the defect parameter index of each data point in the 3D point cloud data, the isolation forest algorithm is used to perform the first split on the 3D point cloud data.

[0083] The process of obtaining the optimal splitting value of the three-dimensional point cloud data of the anti-scaling and anti-corrosion casing for the first splitting is:

[0084] Taking the yth data point in the three-dimensional point cloud data as an example, the calculation formula is as follows:

[0085]

[0086] Among them, F(1,y) represents the possibility that the defect parameter index of the yth data point at the first split belongs to the optimal split value; It represents the mean value of the defect parameter index of all data points in the three-dimensional point cloud data that are smaller than the defect parameter index of the y-th data point at the first split; represents the mean value of the defect parameter index of all data points greater than or equal to the defect parameter index of the yth data point in the three-dimensional point cloud data at the first split; J y Indicates the number of data points in the three-dimensional point cloud data that are smaller than the defect parameter index of the yth data point during the first split; s1 j,y Indicates the defect parameter index of the jth data point in the three-dimensional point cloud data that is smaller than the defect parameter index of the yth data point during the first split; J' y Indicates the number of data points of defect parameter index greater than or equal to the yth data point in the three-dimensional point cloud data at the first split; s2 j ' ,yIt represents the defect parameter index of the j'th data point that is greater than or equal to the defect parameter index of the yth data point in the three-dimensional point cloud data at the first split, and || represents the absolute value function.

[0087] Specifically, through To express the difference between the two categories, it is reflected by the mean difference of the data in the two categories; It is used to represent the sum of the consistency of the two categories. The lower the consistency of the two categories, the smaller the difference within the category, and the smaller the sum of the consistency of the two categories.

[0088] According to the above method, the possibility that the defect parameter index of each data point at the first splitting belongs to the optimal splitting value is obtained; in the three-dimensional point cloud data, the defect parameter index of the data point corresponding to the possibility that the largest defect parameter index belongs to the optimal splitting value is recorded as the optimal splitting value of the first splitting.

[0089] It should be noted that when calculating the optimal split value for the first split, if there is a possibility that multiple largest defect parameter indicators belong to the optimal split value, the average of the defect parameter indicators of the data points corresponding to the possibility that multiple largest defect parameter indicators belong to the optimal split value is taken.

[0090] According to the optimal split value of the first split and the defect parameter index of each data point in the three-dimensional point cloud data, the isolation forest algorithm is used to obtain the first split result of the three-dimensional point cloud data.

[0091] According to the first splitting result of the three-dimensional point cloud data, the optimal splitting value of the second splitting is used, and then the isolation forest algorithm is used to obtain the second splitting result of the three-dimensional point cloud data;

[0092] According to the method for obtaining the optimal splitting value of the first splitting, the optimal splitting value of the second splitting is obtained in the first splitting result of the three-dimensional point cloud data;

[0093] By analogy, the optimal splitting value for each splitting of the three-dimensional point cloud data is obtained.

[0094] According to the optimal split value of each split of the three-dimensional point cloud data and the defect parameter index of each data point in the three-dimensional point cloud data, the isolation forest algorithm is used to obtain the abnormal score of each data point in the three-dimensional point cloud data.

[0095] Step S006: When the abnormal score of any data point in the three-dimensional point cloud data is greater than a preset abnormal threshold, it is determined that there is a defect in the anti-scaling and anti-corrosion casing.

[0096] It should be noted that the abnormal threshold value preset in this embodiment is 0.9, which is used as an example for description. Other implementations may be set to other values, which are not limited in this embodiment. When the abnormal score of any data point in the three-dimensional point cloud data is greater than the abnormal threshold value, it is determined that there is a defect in the anti-scaling and anti-corrosion casing.

[0097] So far, the present invention is completed.

[0098] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. Anti-scaling and anti-corrosion casing defect detection method based on three-dimensional point cloud data, It is characterized in that The method comprises the following steps: Obtain three-dimensional point cloud data of anti-scaling and anti-corrosion casing; Constructing a local space corresponding to each data point in the three-dimensional point cloud data; constructing a spherical space corresponding to each data point in the three-dimensional point cloud data, and obtaining the density of each data point according to the spherical space corresponding to each data point; obtaining the first degree of defect in the local space of each data point in the three-dimensional point cloud data according to the number of data points, the distance between the data points and the density; Obtain the laser reflection intensity of each data point in the three-dimensional point cloud data at different laser detection angles; obtain the second degree of defect in the local space of each data point according to the laser reflection intensity of each data point in the three-dimensional point cloud data at different laser detection angles, the number of data points in the local space, and the distance between the data points; Obtaining a defect parameter index for each data point according to a first degree and a second degree of a defect in a local space of each data point; According to the defect parameter index of each data point in the three-dimensional point cloud data, an abnormality score of each data point in the three-dimensional point cloud data is obtained; When the abnormal score of any data point in the three-dimensional point cloud data is greater than the preset abnormal threshold, it is determined that there is a defect in the anti-scaling and anti-corrosion casing; The method of obtaining the second degree of defect in the local space of each data point according to the laser reflection intensity of each data point in the three-dimensional point cloud data at different laser detection angles, the number of data points in the local space, and the distance between the data points comprises the following specific steps: In the The laser detection angle corresponds to the In the local space of data points, from each data point to the The distance of each data point is taken as the horizontal axis, and the laser reflection intensity of each data point is taken as the vertical axis. The least square method is used to fit the curve of all data points to obtain the The laser detection angle corresponds to the The fitting curve corresponding to the data points; Calculate the The laser detection angle corresponds to the The second-order derivative of the fitting curve corresponding to the data point is obtained The laser detection angle corresponds to the The extreme points in the fitting curve corresponding to the data points; Among all laser detection angles, any two non-repeating laser detection angles form a combination to obtain several combinations; The second degree of defect in the local space of each data point is obtained according to the laser reflection intensity of each data point at different laser detection angles, the number of data points in the local space, and the extreme point in the fitting curve corresponding to each data point corresponding to the two laser detection angles in the combination; According to the laser reflection intensity of each data point at different laser detection angles, the number of data points in the local space, and the extreme value points in the fitting curve corresponding to each data point corresponding to the two laser detection angles in the combination, the second degree of defect in the local space of each data point is obtained, and the corresponding specific calculation formula is: in, Indicates The second degree of defect in the local space of the data points; Indicates the number of laser detection angles; Indicated in The three-dimensional point cloud data corresponding to the laser detection angle The standard deviation of the laser reflection intensity of all data points in the local space of a point; Indicates The number of data points in the local space of data points; Indicated in The three-dimensional point cloud data corresponding to the laser detection angle In the local space of points, The absolute value of the difference in laser reflection intensity between a data point and its nearest data point; Indicates the number of combinations; Indicates The two laser detection angles in the combination correspond to the The minimum value of the number of extreme points in the fitting curve corresponding to the data points; Indicates The two laser detection angles in the combination correspond to the The first data point in the fitted curve The absolute value of the difference between the horizontal coordinates of the extreme points; Indicates The two laser detection angles in the combination correspond to the The first data point in the fitted curve The absolute value of the difference in laser reflection intensity at the extreme points.

2. According to the method for detecting anti-scaling and anti-corrosion casing defects based on three-dimensional point cloud data as described in claim 1, It is characterized in that The specific steps of constructing the local space corresponding to each data point in the three-dimensional point cloud data are as follows: In the 3D point cloud data, each data point is taken as the center and the size is constructed. The local space of It is the preset first space side length.

3. According to the method for detecting anti-scaling and anti-corrosion casing defects based on three-dimensional point cloud data as described in claim 1, It is characterized in that The spherical space corresponding to each data point is constructed in the three-dimensional point cloud data, and the density of each data point is obtained according to the spherical space corresponding to each data point, including the following specific steps: In the 3D point cloud data, each data point is taken as the center and the radius is constructed. spherical space, where is the preset second space radius; The ratio of the number of data points in the spherical space corresponding to each data point to the volume of the spherical space is recorded as the density of each data point.

4. According to the method for detecting anti-scaling and anti-corrosion casing defects based on three-dimensional point cloud data in claim 1, It is characterized in that According to the number of data points in the local space where each data point in the three-dimensional point cloud data is located, the distance between the data points and the density, the first degree of defect in the local space of each data point is obtained, and the corresponding specific calculation formula is: in, Indicates The first degree of defect in the local space of the data points; Indicates In the local space of data points, except for the The number of data points outside of data points; Indicates In the local space of data points, except for the data points outside The density of data points; Indicates Data points and In the local space of data points, except for the Points outside The Euclidean distance of the data points; Indicates Data points and In the local space of data points, except for the Points outside The absolute value of the density difference of the data points; express Belong to the interval All The maximum value in ; express Belong to the interval All The minimum value in ; Indicates The number of data points in the local space of data points; Indicates In the local space of the data point The density of data points; express Belong to the interval All internal time The standard deviation of Represents the normalization function.

5. According to the anti-scaling and anti-corrosion casing defect detection method based on three-dimensional point cloud data as claimed in claim 1, It is characterized in that The step of obtaining the defect parameter index of each data point according to the first degree and the second degree of the defect in the local space of each data point comprises the following specific steps: The first The first degree and the second degree of defects in the local space of the data points The product of the second degree of defects in the local space of data points is recorded as Defect parameter indicators for each data point.

6. According to the method for detecting anti-scaling and anti-corrosion casing defects based on three-dimensional point cloud data as claimed in claim 1, It is characterized in that The specific steps of obtaining the abnormality score of each data point in the three-dimensional point cloud data according to the defect parameter index of each data point in the three-dimensional point cloud data are as follows: According to the defect parameter index of each data point in the three-dimensional point cloud data, the possibility that the defect parameter index of each data point at the first splitting belongs to the optimal splitting value is obtained; In the three-dimensional point cloud data, the defect parameter index of the data point corresponding to the possibility that the largest defect parameter index belongs to the optimal splitting value is recorded as the optimal splitting value of the first splitting; According to the optimal splitting value of the first splitting and the defect parameter index of each data point in the three-dimensional point cloud data, the optimal splitting value of each splitting of the three-dimensional point cloud data is obtained; According to the optimal split value of each split of the three-dimensional point cloud data and the defect parameter index of each data point in the three-dimensional point cloud data, the abnormal score of each data point in the three-dimensional point cloud data is obtained through the isolation forest algorithm.

7. According to claim 6, the anti-scaling and anti-corrosion casing defect detection method based on three-dimensional point cloud data, It is characterized in that According to the defect parameter index of each data point in the three-dimensional point cloud data, the possibility that the defect parameter index of each data point at the first splitting belongs to the optimal splitting value is obtained, and the corresponding specific calculation formula is: in, Indicates the first split The probability that the defect parameter index of each data point belongs to the optimal splitting value; Indicates that the 3D point cloud data is smaller than the first split The mean value of the defect parameter index of all data points of the defect parameter index of data points; Indicates that the number of 3D point cloud data greater than or equal to the first split The mean value of the defect parameter index of all data points of the defect parameter index of data points; Indicates that the 3D point cloud data is smaller than the first split The number of data points for the defect parameter indicator with data points; Indicates that the 3D point cloud data is smaller than the first split The defect parameter index of the data point Defect parameter indicators for each data point; Indicates that the number of 3D point cloud data greater than or equal to the first split The number of data points for the defect parameter indicator with data points; Indicates that the number of 3D point cloud data greater than or equal to the first split The defect parameter index of the data point Defect parameter indicators for data points, represents the absolute value function.

8. According to claim 6, the anti-scaling and anti-corrosion casing defect detection method based on three-dimensional point cloud data, It is characterized in that The method of obtaining the optimal splitting value for each splitting of the three-dimensional point cloud data according to the optimal splitting value of the first splitting and the defect parameter index of each data point in the three-dimensional point cloud data includes the following specific steps: According to the optimal split value of the first split and the defect parameter index of each data point in the three-dimensional point cloud data, the isolation forest algorithm is used to obtain the first split result of the three-dimensional point cloud data; According to the method for obtaining the optimal splitting value of the first splitting, the optimal splitting value of the second splitting is obtained in the first splitting result of the three-dimensional point cloud data; According to the first splitting result of the three-dimensional point cloud data and the optimal splitting value of the second splitting, the isolation forest algorithm is used to obtain the second splitting result of the three-dimensional point cloud data; By analogy, the optimal splitting value for each splitting of the three-dimensional point cloud data is obtained.