CN116817771A - Aerospace part coating thickness measurement method based on cylindrical voxel characteristics - Google Patents
Aerospace part coating thickness measurement method based on cylindrical voxel characteristics Download PDFInfo
- Publication number
- CN116817771A CN116817771A CN202311084451.3A CN202311084451A CN116817771A CN 116817771 A CN116817771 A CN 116817771A CN 202311084451 A CN202311084451 A CN 202311084451A CN 116817771 A CN116817771 A CN 116817771A
- Authority
- CN
- China
- Prior art keywords
- point cloud
- point
- cylindrical
- coating
- descriptor
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Length Measuring Devices With Unspecified Measuring Means (AREA)
- Length Measuring Devices By Optical Means (AREA)
Abstract
The application discloses a method for measuring the thickness of a coating of a space part based on cylindrical voxel characteristics, which comprises the following steps: respectively scanning the space part and the space part coated with the coating to obtain point clouds、The method comprises the steps of carrying out a first treatment on the surface of the Acquiring denoised reference point cloudAnd target point cloudThe method comprises the steps of carrying out a first treatment on the surface of the Will be、Inputting a geometric feature extraction module to obtain a point cloud geometric descriptor; constructing a point cloud normal descriptor based on a normal vector; combining the point cloud geometric descriptor based on the cylinder with the point cloud normal descriptor to obtain a final point cloud feature descriptor，The method comprises the steps of carrying out a first treatment on the surface of the Measuring point cloud feature descriptors，The similarity is obtained by the transformed point cloud B ', and the transformed point cloud B' is matched with the reference point cloudPair Ji Peizhun; computing post-registration reference point cloudsAnd the distance between the transformed point clouds B' to obtain the thickness d of the coating of the aerospace part; the application solves the problems of low efficiency and low accuracy of the detection of the thickness of the electric vortex and improves the accuracy and the high efficiency of the detection of the thickness of the coating.
Description
Technical Field
The application relates to the technical field of product defect detection, in particular to a method for measuring the thickness of a coating of a space part based on cylindrical voxel characteristics.
Background
The surfaces of aerospace parts are often exposed to harsh operating environments such as high temperature, high pressure, oxidizing atmospheres, and the like, and are susceptible to corrosion. The corrosion resistance of the component can be improved and the service life of the component can be prolonged by coating the surface with the coating of the corrosion-resistant material; in the working environment of the space part, high-speed rotation, high temperature and high pressure, particulate matters and other factors tend to exist, the surface of the part is abraded, and the abrasion resistance of the part can be improved and the service life of the part can be prolonged by coating the surface with the abrasion-resistant material; the high-temperature parts of the aerospace parts are required to have high-temperature resistance, and the purpose of coating the thermal barrier coating is to protect the surfaces of the high-temperature parts and reduce the damage of high-temperature environments to the parts.
If the thickness of the coating does not meet the prescribed criteria, it may not provide adequate protection, resulting in premature wear, corrosion, and failure of the aerospace part components. Furthermore, if the coating is too thick, it may also lead to reduced performance of the aerospace part component. Therefore, it is very important to periodically detect the thickness of the coating of the aerospace part, so that the performance and the service life of the aerospace part can be ensured, and the safe operation can be ensured. However, there are still some problems in the existing thickness detection of the coating of aerospace parts, the thickness of the coating of aerospace parts is usually very small, usually between a few micrometers and hundreds of micrometers, which makes the accuracy of measuring the thickness of the coating very critical. Although the currently available detection techniques provide relatively high accuracy, there are still some limitations in terms of measurement range, signal-to-noise ratio, etc.; most coating thickness detection methods currently need to be performed in a laboratory, and the tested parts need to be disassembled and sent to the laboratory for detection, which consumes a lot of time and resources. Meanwhile, real-time detection cannot be performed on site, so that monitoring and maintenance of the coating of the aerospace part become more difficult; different detection techniques may require different coating and base materials, and may require different detection methods for some particular coatings and substrates, which may result in some coating thicknesses not being accurately measured.
Disclosure of Invention
In order to solve the problems, a method for measuring the thickness of a coating of an aerospace part based on cylindrical voxel characteristics is provided, and the method aims to solve the problems that the manual detection thickness omission ratio is high, the accuracy is low, and the accurate positioning and extraction of the abnormal position of the coating thickness cannot be performed in the prior art; the method can detect the thickness of the coating at each position of the aerospace part, accurately position the position which does not meet the standard coating thickness, and improve the accuracy and the high efficiency of the detection of the thickness of the coating of the aerospace part.
In order to achieve the above purpose, the present application provides the following technical solutions: the application provides a method for measuring the thickness of a coating of an aerospace part based on cylindrical voxel characteristics, which comprises the following steps:
s1, respectively scanning a space part and a space part coated with a coating by using a structured light scanner to obtain point clouds、/>；
S2, respectively point-to-point cloud、/>Preprocessing to obtain denoised reference point cloud ∈K>And target point cloud->；
S3, constructing a geometric feature extraction module, and performing、/>Inputting a geometric feature extraction module to obtain a point cloud geometric descriptor based on a cylinder;
s4, respectively calculating、/>Constructing a point cloud normal descriptor based on the normal vector at any point;
s5, combining the point cloud geometric descriptor based on the cylinder with the point cloud normal descriptor to obtain a final point cloud feature descriptor，/>；
S6, measuring point cloud feature descriptors，/>Similarity, establishing a reference point cloud->And target point cloud->The corresponding relation between the two point clouds B 'is obtained after transformation, and the point clouds B' and the reference point clouds are added>Pair Ji Peizhun;
s7, calculating the reference point cloud after registrationAnd the distance between the transformed point clouds B' to obtain the thickness d of the coating of the aerospace part.
Further, in step S2, the point clouds are respectively pointed、/>Preprocessing to obtain denoised reference point cloud ∈K>And target point cloud->The method specifically comprises the following steps:
s201, for space part point cloud、/>Any point p in the graph is searched for all neighborhood points u and neighborhood radiuses, wherein />Is Gaussian weight;
s202, calculating the distance between the point p and the neighborhood point u，/>Intensity difference->，/>，/>For the intensity of point p +.>The angle of the neighborhood point u;
s203, calculating the intensity value of the point pAnd calculating the sum of the intensity values of the neighborhood points u of the point p +.>；
S204, calculating the point p strength after bilateral filtering:；
s205, point cloud、/>Each point in the process is sequentially subjected to the four steps to obtain a denoised space part reference point cloud ++>Target point cloud->。
Further, the geometric feature extraction module in step S3 includes a point cloud converter and a feature extraction network, where the point cloud converter is a reference point cloud to be inputTarget point cloud->Mapping to the cylindrical space, thereby overcoming the rotation variance and not losing the key information of the local pattern; the feature extraction network is based on a 3D cylindrical voxel convolution, and the obtained cylindrical voxel tensor is input into the feature extraction network to obtain a compact and representative point cloud geometric descriptor for registration.
Further, the point cloud converter specifically includes the following steps:
for a reference point cloudTarget point cloud->Is->Obtaining the neighborhood radius +.>Neighborhood point set->Use +.>Estimating an axis pointing in the shooting direction>Point cloud->Make a transformation to->The axis is aligned with the z-axis, and the rotational variance of the z-axis is eliminated by the following specific transformation modes:
；
wherein ：is a transformed point cloud, which +.>The axis is aligned with the z-axis->Is a transformation matrix;
will beConsidered as spheres and uniformly divided into +.o along radial distance ρ, elevation angle φ and azimuth angle θ>Personal (S)Voxels, wherein J, H, K is the number of voxels divided by the radial distance, elevation angle and azimuth angle of the sphere, and the center point of each voxel is expressed as +.>Find the center point of each voxel +.>Neighborhood point set->Further eliminating the rotational variance on the XY plane;
for transformed spherical voxels, each spherical voxel has a set of neighborhood pointsEach spherical voxel is logically projected as a cylinder, denoted +.>。
Further, the feature extraction network based on 3D cylindrical voxel convolution comprises a full connection layer and a 3D cylindrical convolution layer, and specifically comprises the following steps:
inputting cylindrical voxels into full-connected layer to obtain full-connected characteristicsMLP weights are shared among all cylindrical voxels, < +.>And (3) for rotating the matrix, pooling the fully connected features to obtain pooled features: />A is a pooling function;
given voxels at locations (g, h, k) on the mth cylindrical feature map of the s-th layer, the 3D cylindrical convolution network is defined as follows:
；
wherein Is the size of the convolution kernel in the radial dimension, W and Z are the height and width of the convolution kernel, respectively,/->For learning parameters, M is a cylinderX, y, r are the current values of the height and width, radial dimensions, respectively, of the convolution kernel,/->A cylindrical-based point cloud geometry descriptor with rotational invariance is obtained.
Further, in step S4, reference point clouds are calculated respectivelyTarget point cloud->The normal vector at any point of the method is used for constructing a point cloud normal descriptor based on the normal vector, and the method specifically comprises the following steps of:
s401, for the denoised reference point cloudTarget point cloud->Is->Find its local neighborhood point set，/>,/>For neighborhood point->Is the neighborhood radius;
s402, calculating points by using covariance matrixNormal vector at->：/>；
S403, encoding the normal vector by using sine functions with different frequencies:
；
wherein ：is a normalized coefficient, ++>For->Index of->Is->、/>Angle between normals>For->Normal vector of neighborhood, T is transposed matrix, +.>Is a normal descriptor of the point cloud Q, +.>Is a reference point cloud->Is described in->Is the target point cloud->The dimension of which is the same as the dimension of the cylindrical-based point cloud geometry descriptor.
Further, the step S6 specifically includes the following steps:
s601, calculating feature descriptors，/>The closer the distance is, the more similar the distance measures similarity of feature descriptors;
s602, cloud from target pointFind and reference point cloud->The point with the maximum similarity in the two points is established with the corresponding relation of the point pairs;
s603, for target point cloudRepeating steps S601-S602 until the corresponding relation of all the point pairs is established;
s604, transforming the point pair into a matrix form, and transforming the point cloud B through a rotation matrix R and a translation vector t to obtain a transformed point cloud B';
；
；
；
wherein ,respectively and->A is a reference point cloud for corresponding point pairs>The point cloud after the corresponding point pair is established, B is the target point cloud +.>Establishing a point cloud after corresponding point pairs;
s605, SVD decomposition is carried out on the A and the B 'to obtain left singular vectors and right singular vectors of the A and the B', and then a rotation matrix R and a translation vector t are obtained according to the nature of the singular vectors;
s606, after the rotation matrix R and the translation vector t are obtained, the target point cloud is obtainedTransforming, namely transforming the transformed point cloud B' and the reference point cloud +.>Registration of alignment.
By the technical scheme, the application provides a method for measuring the thickness of a coating of an aerospace part based on cylindrical voxel characteristics. The method has at least the following beneficial effects:
the application combines the point cloud geometric descriptors obtained by the denoised reference point cloud and target point cloud input geometric feature extraction module with the point cloud normal descriptors based on normal vectors to finally obtain the point cloud feature descriptors, establishes the corresponding relationship between the reference point cloud and the target point cloud by measuring the similarity of the point cloud feature descriptors, and obtains the transformed point cloud and the reference point cloudAlignment registration, finally, the thickness of the coating of the aerospace part is calculated through the distance between the reference point cloud after registration and the transformed point cloud, and after the image of the detection article of the target point cloud is acquired, the fine defect in the image can be accurately identified and positioned, and the aerospace part can be subjected to aerospaceThe coating thickness of each position of the part is detected, the position which does not meet the standard coating thickness is accurately positioned, the problems of high missing detection rate and low accuracy of the image fine defect detection thickness in the prior art are solved, and the accuracy and the high efficiency of the aerospace part coating thickness detection are improved.
Drawings
The accompanying drawings are included to provide a further understanding of the application and are incorporated in and constitute a part of this specification, illustrate the application and together with the embodiments of the application, serve to explain the application. In the drawings:
FIG. 1 is a flow chart of a method for measuring the thickness of a coating of an aerospace part based on cylindrical voxel features of the present application;
FIG. 2 is a flow chart of a point cloud converter of the present application;
FIG. 3 is a flow chart of the present application for constructing a point cloud feature descriptor;
FIG. 4 is a registered aerospace part point cloud of an embodiment of the present application;
fig. 5 is a calculated coating thickness for an embodiment of the present application.
Detailed Description
In order that the above-recited objects, features and advantages of the present application will become more readily apparent, a more particular description of the application will be rendered by reference to the appended drawings and appended detailed description. Therefore, the realization process of how to apply the technical means to solve the technical problems and achieve the technical effects can be fully understood and implemented.
Those of ordinary skill in the art will appreciate that all or a portion of the steps in a method of implementing an embodiment described above may be implemented by a program to instruct related hardware, and thus, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
Referring to FIGS. 1-5, a specific implementation of the present embodiment is shownThe method comprises the steps of merging a point cloud geometric descriptor obtained by inputting a denoised reference point cloud and a target point cloud into a geometric feature extraction module with a point cloud normal descriptor based on a normal vector, finally obtaining a point cloud feature descriptor, establishing a corresponding relation between the reference point cloud and the target point cloud by measuring similarity of the point cloud feature descriptor, and obtaining the transformed point cloud and the reference point cloudThe method and the device can accurately identify and position fine defects in the image after acquiring the image of the detection product of the target point cloud, detect the coating thickness of each position of the aerospace part, accurately position the position which does not meet the standard coating thickness, solve the problems of high missing detection rate and low accuracy of the detection thickness of the fine defects of the image in the prior art, and improve the accuracy and high efficiency of the detection of the coating thickness of the aerospace part;
referring to fig. 1, a method for measuring thickness of a coating of an aerospace part based on a cylindrical point cloud feature descriptor includes the following steps:
s1, respectively scanning a space part and a space part coated with a coating by using a structured light scanner to obtain point clouds、/>。
S2, respectively point-to-point cloud、/>Preprocessing to obtain denoised reference point cloud ∈K>And target point cloud->；
Specifically, in step S2, the point clouds are respectively pointed、/>Preprocessing to obtain denoised reference point cloud ∈K>And target point cloud->The method comprises the following steps:
s201, for space part point cloud、/>Any point p in the graph is searched for all neighborhood points u and neighborhood radiuses, wherein />Is Gaussian weight;
s202, calculating the distance between the point p and the neighborhood point u，/>Intensity difference->，/>，/>Is taken as a pointIntensity of p->The angle of the neighborhood point u;
s203, calculating the intensity value of the point pAnd calculating the sum of the intensity values of the neighborhood points u of the point p +.>, wherein ,，/>the weight is as follows: />，/>Is a spatial weight;
s204, calculating the point p strength after bilateral filtering:；
s205, point cloud、/>Each point in the process is sequentially subjected to the four steps to obtain a denoised space part reference point cloud ++>Target point cloud->As shown in fig. 3.
S3, constructing a geometric feature extraction module, and performing、/>Inputting a geometric feature extraction module to obtain a point cloud geometric descriptor based on a cylinder;
specifically, the geometric feature extraction module in step S3 includes a point cloud converter and a feature extraction network, where the point cloud converter is a reference point cloud to be inputTarget point cloud->Mapping to the cylindrical space, thereby overcoming the rotation variance and not losing the key information of the local pattern; FIG. 2 is a flow chart of the point cloud converter of the present application; the feature extraction network is based on a 3D voxel convolution feature extraction network, and the obtained voxel tensor is input into the feature extraction network to obtain a compact and representative point cloud geometric descriptor for registration, as shown in fig. 3.
Specifically, the point cloud converter specifically includes the following steps:
for a reference point cloudTarget point cloud->Is->Obtaining the neighborhood radius +.>Neighborhood point set->Use +.>Estimating an axis pointing in the shooting direction>Point cloud->Make a transformation to->The axis is aligned with the z-axis, and the rotational variance of the z-axis is eliminated by the following specific transformation modes:
；
wherein ：is a transformed point cloud, which +.>The axis is aligned with the z-axis->Is a transformation matrix;
will beConsidered as spheres and uniformly divided into +.o along radial distance ρ, elevation angle φ and azimuth angle θ>Personal (S)Voxels, wherein J, H, K is the number of voxels divided by the radial distance, elevation angle and azimuth angle of the sphere, and the center point of each voxel is expressed as +.>Find the center point of each voxel +.>Neighborhood point set->Enter intoOne step eliminates the rotational variance on the XY plane;
for transformed spherical voxels, each spherical voxel has a set of neighborhood pointsEach spherical voxel is logically projected as a cylinder, denoted +.>. Fig. 2 is a flow chart of the point cloud converter according to the present application.
Specifically, the feature extraction network based on 3D cylindrical voxel convolution includes a full connection layer and a 3D cylindrical convolution layer, including:
inputting cylindrical voxels into full-connected layer to obtain full-connected characteristicsMLP weights are shared among all cylindrical voxels, < +.>And (3) for rotating the matrix, pooling the fully connected features to obtain pooled features: />A is a pooling function;
given voxels at locations (g, h, k) on the mth cylindrical feature map of the s-th layer, the 3D cylindrical convolution network is defined as follows:
；
wherein Is the size of the convolution kernel in the radial dimension, W and Z are the height and width of the convolution kernel, respectively,/->For the learned parameters, M is the number of cylinders, x, y and r are the current values of the height and width and radial dimension of the convolution kernel respectively, +.>A cylindrical-based point cloud geometry descriptor with rotational invariance is obtained.
S4, respectively calculating、/>Constructing a point cloud normal descriptor based on the normal vector at any point;
specifically, in step S4, reference point clouds are calculated respectivelyTarget point cloud->The normal vector at any point is used for constructing a point cloud normal descriptor based on the normal vector, and the method comprises the following steps of:
s401, for the denoised reference point cloudTarget point cloud->Is->Find its local neighborhood point set，/>,/>For neighborhood point->Is the neighborhood radius;
s402, calculating points by using covariance matrixNormal vector at->：/>；
S403, encoding the normal vector by using sine functions with different frequencies:
；
wherein ：is a normalized coefficient, ++>For->Index of->Is->、/>Angle between normals>For->Normal vector of neighborhood, T is transposed matrix, +.>Is a normal descriptor of the point cloud Q, +.>Is a reference point cloud->Is described in->Is the target point cloud->The dimension of which is the same as the dimension of the cylindrical-based point cloud geometry descriptor.
S5, combining the point cloud geometric descriptor based on the cylinder with the point cloud normal descriptor to obtain a final point cloud feature descriptor，/>。
S6, measuring point cloud feature descriptors，/>Similarity, establishing a reference point cloud->And target point cloud->The corresponding relation between the two point clouds B 'is obtained after transformation, and the point clouds B' and the reference point clouds are added>Pair Ji Peizhun;
specifically, the step S6 specifically includes the following steps:
s601, calculating feature descriptors，/>Distance of (2)Measuring similarity of feature descriptors, wherein the closer the distance is, the more similar the feature descriptors are;
s602, cloud from target pointFind and reference point cloud->The point with the maximum similarity in the two points is established with the corresponding relation of the point pairs;
s603, for target point cloudRepeating steps S601-S602 until the corresponding relation of all the point pairs is established;
s604, transforming the point pair into a matrix form, and transforming the point cloud B through a rotation matrix R and a translation vector t to obtain a transformed point cloud B';
；
；
；
wherein ,respectively and->A is a reference point cloud for corresponding point pairs>The point cloud after the corresponding point pair is established, B is the target point cloud +.>Establishing a point cloud after corresponding point pairs;
s605, SVD decomposition is carried out on the A and the B 'to obtain left singular vectors and right singular vectors of the A and the B', and then a rotation matrix R and a translation vector t are obtained according to the nature of the singular vectors;
s606, after the rotation matrix R and the translation vector t are obtained, the target point cloud is obtainedTransforming, namely transforming the transformed point cloud B' and the reference point cloud +.>Registration of alignment. Fig. 4 shows the space part point cloud obtained after registration according to an embodiment of the present application.
S7, calculating the reference point cloud after registrationAnd the distance between the transformed point clouds B' to obtain the thickness d of the coating of the aerospace part. As shown in fig. 5, the calculated coating thickness in the embodiment of the present application is shown, x, y, z in the upper left corner of fig. 5 are coordinates of the aerospace part for measuring thickness, nx, ny, nz are normal values of x, y, z, respectively, and c2c absolute distances represents the distance from the reference point cloud after registration to the point cloud after transformation, that is, the current point coating thickness.
The method can detect the thickness of the coating at each position of the aerospace part, accurately position the position which does not meet the standard thickness of the coating, is suitable for various coatings and base materials, is also suitable for special coatings and base materials, and improves the accuracy and the high efficiency of the thickness detection of the coating of the aerospace part.
In the present specification, each embodiment is described in a progressive manner, and each embodiment is mainly described in a different point from other embodiments, and identical and similar parts between the embodiments are all enough to refer to each other. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant points refer to the description of the method section.
The principles and embodiments of the present application have been described herein with reference to specific examples, the description of which is intended only to assist in understanding the methods of the present application and the core ideas thereof; also, it is within the scope of the present application to be modified by those of ordinary skill in the art in light of the present teachings. In view of the foregoing, this description should not be construed as limiting the application.
Claims (7)
1. The method for measuring the thickness of the coating of the aerospace part based on the cylindrical voxel characteristics is characterized by comprising the following steps of:
s1, respectively scanning a space part and a space part coated with a coating by using a structured light scanner to obtain point clouds、；
S2, respectively point-to-point cloud、/>Preprocessing to obtain denoised reference point cloud ∈K>And target point cloud->；
S3, constructing a geometric feature extraction module, and performing、/>Inputting a geometric feature extraction module to obtain a point cloud geometric descriptor based on a cylinder;
s4, respectively calculating、/>Constructing a point cloud normal descriptor based on the normal vector at any point;
s5, combining the point cloud geometric descriptor based on the cylinder with the point cloud normal descriptor to obtain a final point cloud feature descriptor，/>；
S6, measuring point cloud feature descriptors，/>Similarity, establishing a reference point cloud->And target point cloud->The corresponding relation between the two point clouds B 'is obtained after transformation, and the point clouds B' and the reference point clouds are added>Pair Ji Peizhun;
s7, calculating the reference point cloud after registrationAnd the distance between the transformed point clouds B' to obtain the thickness d of the coating of the aerospace part.
2. The method for measuring the thickness of a coating of an aerospace component based on cylindrical voxel characteristics as set forth in claim 1, wherein in step S2, the coating is formed byPoint-to-point cloud、/>Preprocessing to obtain denoised reference point cloud ∈K>And target point cloud->The method specifically comprises the following steps:
s201, for space part point cloud、/>Any point p in the graph is searched for all neighborhood points u and neighborhood radiuses, wherein />Is Gaussian weight;
s202, calculating the distance between the point p and the neighborhood point u，/>Intensity difference->，/>，/>For the intensity of point p +.>The angle of the neighborhood point u;
s203, calculating the intensity value of the point pAnd calculating the sum of the intensity values of the neighborhood points u of the point p +.>；
S204, calculating the point p strength after bilateral filtering:；
s205, point cloud、/>Each point in the process is sequentially subjected to the four steps to obtain a denoised space part reference point cloud ++>Target point cloud->。
3. The method for measuring thickness of coating of aerospace component based on cylindrical voxel feature of claim 1, wherein the geometrical feature extraction module in step S3 comprises a point cloud converter and a feature extraction network, the point cloud converter is a reference point cloud to be inputTarget point cloud->Mapping into a cylindrical space; the point cloud converter specifically comprises the following steps:
for a reference point cloudTarget point cloud->Is->Obtaining the neighborhood radius +.>Neighborhood point set->Use +.>Estimating an axis pointing in the shooting direction>Point cloud->Make a transformation to->The axis is aligned with the z-axis, and the rotational variance of the z-axis is eliminated by the following specific transformation modes:
；
wherein ：is a transformed point cloud, which +.>The axis is aligned with the z-axis->Is a transformation matrix;
will beConsidered as spheres and uniformly divided into +.o along radial distance ρ, elevation angle φ and azimuth angle θ>Personal (S)Voxels, wherein J, H, K is the number of voxels divided by the radial distance, elevation angle and azimuth angle of the sphere, and the center point of each voxel is expressed as +.>Find the center point of each voxel +.>Neighborhood point set->Further eliminating the rotational variance on the XY plane;
for transformed spherical voxels, each spherical voxel has a set of neighborhood pointsEach spherical voxel is logically projected as a cylinder, denoted +.>。
4. A method of measuring the thickness of a coating of a aerospace component based on voxel features as claimed in claim 3, wherein the feature extraction network is a 3D voxel convolution based feature extraction network, and the resulting voxel tensor is input into the feature extraction network to yield a compact and representative point cloud geometry descriptor for registration.
5. The method for measuring the thickness of the coating of the aerospace part based on the cylindrical voxel characteristics according to claim 4, wherein the characteristic extraction network based on the 3D cylindrical voxel convolution comprises a full-connection layer and a 3D cylindrical convolution layer, and specifically comprises the following steps:
inputting cylindrical voxels into full-connected layer to obtain full-connected characteristicsThe MLP weights are shared among all cylindrical voxels,and (3) for rotating the matrix, pooling the fully connected features to obtain pooled features: />A is a pooling function;
given voxels at locations (g, h, k) on the mth cylindrical feature map of the s-th layer, the 3D cylindrical convolution network is defined as follows:
；
wherein Is the size of the convolution kernel in the radial dimension, W and Z are the height and width of the convolution kernel, respectively,/->For the learned parameters, M is the number of cylinders, x, y and r are the current values of the height and width and radial dimension of the convolution kernel respectively, +.>A cylindrical-based point cloud geometry descriptor with rotational invariance is obtained.
6. The method for measuring the thickness of a coating of an aerospace component based on cylindrical voxel features as set forth in claim 1, wherein in step S4, reference point clouds are calculated respectivelyTarget point cloud->The normal vector at any point of the method is used for constructing a point cloud normal descriptor based on the normal vector, and the method specifically comprises the following steps of:
s401, for the denoised reference point cloudTarget point cloud->Is->Find its local neighborhood point set +.>，,/>For neighborhood point->Is the neighborhood radius;
s402, calculating points by using covariance matrixNormal vector at->：/>；
S403, encoding the normal vector by using sine functions with different frequencies:
；
wherein ：is a normalized coefficient, ++>For->Index of->Is->、/>Angle between normals>For->Normal vector of neighborhood, T is transposed matrix, +.>Is a normal descriptor of the point cloud Q, +.>Is a reference point cloud->Is described in the description of (a),is the target point cloud->The dimension of which is the same as the dimension of the cylindrical-based point cloud geometry descriptor.
7. The method for measuring the thickness of a coating of an aerospace component based on cylindrical voxel characteristics according to claim 1, wherein the step S6 specifically comprises the following steps:
s601, calculating feature descriptors，/>The closer the distance is, the more similar the distance measures similarity of feature descriptors;
s602, cloud from target pointFind and reference point cloud->The point with the maximum similarity in the two points is established with the corresponding relation of the point pairs;
s603, for target point cloudRepeating steps S601-S602 until the corresponding relation of all the point pairs is established;
s604, transforming the point pair into a matrix form, and transforming the point cloud B through a rotation matrix R and a translation vector t to obtain a transformed point cloud B';
；
；
；
wherein ,respectively and->A is a reference point cloud for corresponding point pairs>The point cloud after the corresponding point pair is established, B is the target point cloud +.>Establishing a point cloud after corresponding point pairs;
s605, SVD decomposition is carried out on the A and the B 'to obtain left singular vectors and right singular vectors of the A and the B', and then a rotation matrix R and a translation vector t are obtained according to the nature of the singular vectors;
s606, after the rotation matrix R and the translation vector t are obtained, the target point cloud is obtainedTransforming, namely transforming the transformed point cloud B' and the reference point cloud +.>Registration of alignment.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311084451.3A CN116817771B (en) | 2023-08-28 | 2023-08-28 | Aerospace part coating thickness measurement method based on cylindrical voxel characteristics |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311084451.3A CN116817771B (en) | 2023-08-28 | 2023-08-28 | Aerospace part coating thickness measurement method based on cylindrical voxel characteristics |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116817771A true CN116817771A (en) | 2023-09-29 |
| CN116817771B CN116817771B (en) | 2023-11-17 |
Family
ID=88141356
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311084451.3A Active CN116817771B (en) | 2023-08-28 | 2023-08-28 | Aerospace part coating thickness measurement method based on cylindrical voxel characteristics |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116817771B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117340554A (en) * | 2023-10-11 | 2024-01-05 | 湖南大学 | A method for manufacturing a rotary engine aluminum alloy cylinder block |
| CN118392054A (en) * | 2024-06-27 | 2024-07-26 | 南京航空航天大学 | Method for measuring thickness of aerospace wallboard based on point cloud data and Haosdorf measurement |
Citations (31)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104807409A (en) * | 2015-04-11 | 2015-07-29 | 成都陵川特种工业有限责任公司 | Wheel spoke thickness measuring method |
| US20150279121A1 (en) * | 2014-03-27 | 2015-10-01 | Knockout Concepts, Llc | Active Point Cloud Modeling |
| CN105956234A (en) * | 2016-04-21 | 2016-09-21 | 哈尔滨工业大学 | A virtual detection method for steam turbine components based on reverse engineering and a virtual repair method based on the virtual detection method |
| CN105976406A (en) * | 2016-04-26 | 2016-09-28 | 上海时元互联网科技有限公司 | Measurement system, measurement device, foot type measurement method, and foot type measurement system |
| CN106600622A (en) * | 2016-12-06 | 2017-04-26 | 西安电子科技大学 | Point cloud data partitioning method based on hyper voxels |
| CN109002418A (en) * | 2018-06-20 | 2018-12-14 | 厦门大学 | Tree breast-height diameter automatic calculating method based on Voxels growing and ground laser point cloud |
| US20190073744A1 (en) * | 2017-09-01 | 2019-03-07 | Tata Consultancy Services Limited | Systems and methods for shape constrained 3d point cloud registration |
| CN110108217A (en) * | 2019-03-22 | 2019-08-09 | 中交第二航务工程局有限公司 | A kind of tunnel just props up and two linings invade limit value and thickness analysis method |
| CN110246112A (en) * | 2019-01-21 | 2019-09-17 | 厦门大学 | Three-dimensional point cloud quality evaluating method in the room laser scanning SLAM based on deep learning |
| CN110375722A (en) * | 2019-06-25 | 2019-10-25 | 北京工业大学 | A method of duct pieces of shield tunnel joint open is extracted based on point cloud data |
| CN110942515A (en) * | 2019-11-26 | 2020-03-31 | 北京迈格威科技有限公司 | Point cloud-based target object three-dimensional computer modeling method and target identification method |
| CN111006601A (en) * | 2019-11-12 | 2020-04-14 | 黑龙江工程学院 | Key technology of three-dimensional laser scanning in deformation monitoring |
| JP2020076585A (en) * | 2018-11-05 | 2020-05-21 | 倉敷紡績株式会社 | Measuring method of three-dimensional shape of coating material, measuring system of three-dimensional shape of coating material, and application method of coating material |
| CN111539881A (en) * | 2020-04-16 | 2020-08-14 | 南京航空航天大学 | A deep learning-based point cloud denoising method for aerospace parts |
| US20200400442A1 (en) * | 2018-03-15 | 2020-12-24 | Kaarta, Inc. | Methods and systems for filtering point cloud map data for use with acquired scan frame data |
| CN112200843A (en) * | 2020-10-09 | 2021-01-08 | 福州大学 | CBCT and laser scanning point cloud data tooth registration method based on hyper-voxels |
| US20210065431A1 (en) * | 2019-09-04 | 2021-03-04 | Faro Technologies, Inc. | System and method for training a neural network to fill gaps between scan points in images and to de-noise point cloud images |
| AU2021105163A4 (en) * | 2021-08-09 | 2021-10-07 | Yunshigao Technology Company Limited | A learning based point cloud denoising algorithm |
| US20210347501A1 (en) * | 2020-05-09 | 2021-11-11 | Nanjing University Of Aeronautics And Astronautics | Fairing skin repair method based on measured wing data |
| CN114608461A (en) * | 2022-03-27 | 2022-06-10 | 河南科技大学 | Laser scanning measurement method for parts with non-uniform wall thickness |
| CN114663373A (en) * | 2022-03-11 | 2022-06-24 | 华中科技大学 | Point cloud registration method and device for detecting surface quality of part |
| CN114972459A (en) * | 2022-05-31 | 2022-08-30 | 哈尔滨理工大学 | Point cloud registration method based on low-dimensional point cloud local feature descriptor |
| CN115127478A (en) * | 2022-07-14 | 2022-09-30 | 苏州大学 | Detection method of quality deviation of shield machine segment assembly based on laser scanning |
| CN115409886A (en) * | 2022-11-02 | 2022-11-29 | 南京航空航天大学 | Part geometric feature measuring method, device and system based on point cloud |
| CN115619980A (en) * | 2022-11-23 | 2023-01-17 | 青岛小优智能科技有限公司 | Belted layer overlap flaw detection method based on point cloud |
| CN115619969A (en) * | 2022-10-17 | 2023-01-17 | 中国人民解放军军事科学院国防工程研究院 | Building damage trace measuring method based on laser point cloud difference |
| CN116204990A (en) * | 2023-04-28 | 2023-06-02 | 南京航空航天大学 | Three-dimensional measured data driven precise coordination repair method for large-scale framework of aircraft |
| US20230177707A1 (en) * | 2020-06-26 | 2023-06-08 | DJI Research LLC | Post-processing of mapping data for improved accuracy and noise-reduction |
| CN116258835A (en) * | 2023-05-04 | 2023-06-13 | 武汉大学 | Method and system for 3D reconstruction of point cloud data based on deep learning |
| US20230186437A1 (en) * | 2021-12-14 | 2023-06-15 | Faro Technologies, Inc. | Denoising point clouds |
| CN116363177A (en) * | 2023-02-23 | 2023-06-30 | 西北大学 | Three-dimensional point cloud registration method based on geometric transformation and Gaussian mixture model |
-
2023
- 2023-08-28 CN CN202311084451.3A patent/CN116817771B/en active Active
Patent Citations (31)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150279121A1 (en) * | 2014-03-27 | 2015-10-01 | Knockout Concepts, Llc | Active Point Cloud Modeling |
| CN104807409A (en) * | 2015-04-11 | 2015-07-29 | 成都陵川特种工业有限责任公司 | Wheel spoke thickness measuring method |
| CN105956234A (en) * | 2016-04-21 | 2016-09-21 | 哈尔滨工业大学 | A virtual detection method for steam turbine components based on reverse engineering and a virtual repair method based on the virtual detection method |
| CN105976406A (en) * | 2016-04-26 | 2016-09-28 | 上海时元互联网科技有限公司 | Measurement system, measurement device, foot type measurement method, and foot type measurement system |
| CN106600622A (en) * | 2016-12-06 | 2017-04-26 | 西安电子科技大学 | Point cloud data partitioning method based on hyper voxels |
| US20190073744A1 (en) * | 2017-09-01 | 2019-03-07 | Tata Consultancy Services Limited | Systems and methods for shape constrained 3d point cloud registration |
| US20200400442A1 (en) * | 2018-03-15 | 2020-12-24 | Kaarta, Inc. | Methods and systems for filtering point cloud map data for use with acquired scan frame data |
| CN109002418A (en) * | 2018-06-20 | 2018-12-14 | 厦门大学 | Tree breast-height diameter automatic calculating method based on Voxels growing and ground laser point cloud |
| JP2020076585A (en) * | 2018-11-05 | 2020-05-21 | 倉敷紡績株式会社 | Measuring method of three-dimensional shape of coating material, measuring system of three-dimensional shape of coating material, and application method of coating material |
| CN110246112A (en) * | 2019-01-21 | 2019-09-17 | 厦门大学 | Three-dimensional point cloud quality evaluating method in the room laser scanning SLAM based on deep learning |
| CN110108217A (en) * | 2019-03-22 | 2019-08-09 | 中交第二航务工程局有限公司 | A kind of tunnel just props up and two linings invade limit value and thickness analysis method |
| CN110375722A (en) * | 2019-06-25 | 2019-10-25 | 北京工业大学 | A method of duct pieces of shield tunnel joint open is extracted based on point cloud data |
| US20210065431A1 (en) * | 2019-09-04 | 2021-03-04 | Faro Technologies, Inc. | System and method for training a neural network to fill gaps between scan points in images and to de-noise point cloud images |
| CN111006601A (en) * | 2019-11-12 | 2020-04-14 | 黑龙江工程学院 | Key technology of three-dimensional laser scanning in deformation monitoring |
| CN110942515A (en) * | 2019-11-26 | 2020-03-31 | 北京迈格威科技有限公司 | Point cloud-based target object three-dimensional computer modeling method and target identification method |
| CN111539881A (en) * | 2020-04-16 | 2020-08-14 | 南京航空航天大学 | A deep learning-based point cloud denoising method for aerospace parts |
| US20210347501A1 (en) * | 2020-05-09 | 2021-11-11 | Nanjing University Of Aeronautics And Astronautics | Fairing skin repair method based on measured wing data |
| US20230177707A1 (en) * | 2020-06-26 | 2023-06-08 | DJI Research LLC | Post-processing of mapping data for improved accuracy and noise-reduction |
| CN112200843A (en) * | 2020-10-09 | 2021-01-08 | 福州大学 | CBCT and laser scanning point cloud data tooth registration method based on hyper-voxels |
| AU2021105163A4 (en) * | 2021-08-09 | 2021-10-07 | Yunshigao Technology Company Limited | A learning based point cloud denoising algorithm |
| US20230186437A1 (en) * | 2021-12-14 | 2023-06-15 | Faro Technologies, Inc. | Denoising point clouds |
| CN114663373A (en) * | 2022-03-11 | 2022-06-24 | 华中科技大学 | Point cloud registration method and device for detecting surface quality of part |
| CN114608461A (en) * | 2022-03-27 | 2022-06-10 | 河南科技大学 | Laser scanning measurement method for parts with non-uniform wall thickness |
| CN114972459A (en) * | 2022-05-31 | 2022-08-30 | 哈尔滨理工大学 | Point cloud registration method based on low-dimensional point cloud local feature descriptor |
| CN115127478A (en) * | 2022-07-14 | 2022-09-30 | 苏州大学 | Detection method of quality deviation of shield machine segment assembly based on laser scanning |
| CN115619969A (en) * | 2022-10-17 | 2023-01-17 | 中国人民解放军军事科学院国防工程研究院 | Building damage trace measuring method based on laser point cloud difference |
| CN115409886A (en) * | 2022-11-02 | 2022-11-29 | 南京航空航天大学 | Part geometric feature measuring method, device and system based on point cloud |
| CN115619980A (en) * | 2022-11-23 | 2023-01-17 | 青岛小优智能科技有限公司 | Belted layer overlap flaw detection method based on point cloud |
| CN116363177A (en) * | 2023-02-23 | 2023-06-30 | 西北大学 | Three-dimensional point cloud registration method based on geometric transformation and Gaussian mixture model |
| CN116204990A (en) * | 2023-04-28 | 2023-06-02 | 南京航空航天大学 | Three-dimensional measured data driven precise coordination repair method for large-scale framework of aircraft |
| CN116258835A (en) * | 2023-05-04 | 2023-06-13 | 武汉大学 | Method and system for 3D reconstruction of point cloud data based on deep learning |
Non-Patent Citations (5)
| Title |
|---|
| BAI, CG 等: "Faster-LIO: Lightweight Tightly Coupled Lidar-Inertial odometry Using Parallel Sparse Incremental Voxels", 《IEEE ROBOTICS AND AUTOMATION LETTERS》, vol. 7, no. 2, pages 4861 - 4868 * |
| 余飞祥 等: "基于扫描线点云的飞机蒙皮边界特征提取", 计算机集成制造系统, vol. 23, no. 04, pages 701 - 707 * |
| 李艺达 等: "特征自适应的飞机大型蒙皮曲面测量系统与测量规划", 《南京航空航天大学学报》, vol. 54, no. 3, pages 428 - 438 * |
| 李雪梅 等: "基于超体素双向最近邻距离比的点云配准方法", 《吉林大学学报（工学版）》, vol. 52, no. 8, pages 1918 - 1925 * |
| 段晓峰 等: "铁路线路点云数据可视化重构方法研究", 铁道标准设计, vol. 61, no. 04, pages 14 - 18 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117340554A (en) * | 2023-10-11 | 2024-01-05 | 湖南大学 | A method for manufacturing a rotary engine aluminum alloy cylinder block |
| CN118392054A (en) * | 2024-06-27 | 2024-07-26 | 南京航空航天大学 | Method for measuring thickness of aerospace wallboard based on point cloud data and Haosdorf measurement |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116817771B (en) | 2023-11-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN116817771A (en) | Aerospace part coating thickness measurement method based on cylindrical voxel characteristics | |
| Xia et al. | Detection method of manufacturing defects on aircraft surface based on fringe projection | |
| Cao et al. | Large-complex-surface defect detection by hybrid gradient threshold segmentation and image registration | |
| Wang et al. | Optical measurement method for blade profiles based on blade self-features | |
| Zhou et al. | A measurement system based on internal cooperation of cameras in binocular vision | |
| Dong et al. | The calibration method of the circle-structured light measurement system for inner surfaces considering systematic errors | |
| CN111369609B (en) | Building local deformation analysis method based on point cloud curved surface feature constraint | |
| CN111637851A (en) | Aruco code-based visual measurement method and device for plane rotation angle | |
| Hu et al. | Measurement method of rotation angle and clearance in intelligent spherical hinge | |
| Yu et al. | A novel method of multitarget augmented reality assembly result inspection for large complex scenes | |
| Al Khawli et al. | Introducing data analytics to the robotic drilling process | |
| Gadelmawla et al. | Measurement and Inspection of Roundness Using Computer Vision. | |
| An et al. | Tube inner wall defect detection method based on rotary scanning micro laser distance sensor | |
| Sun et al. | Precision work-piece detection and measurement combining top-down and bottom-up saliency | |
| Huang et al. | Vision measurement for flat parts based on local line-angle contour segmentation | |
| Duan et al. | The icosahedron marker for Robots 6-dof pose estimation | |
| Chen et al. | Aircraft skin countersink primitive extraction from 3-d measurement point clouds via deep clustering and fitting | |
| CN116124047A (en) | Roughness anisotropic parameter measurement method based on normal vector statistical characteristics | |
| Zhao et al. | 3D measurement of precise part in complex circumstance using line structured light and improved U-Net | |
| Liao et al. | A Coupled 3d Morphological Reconstruction Approach for Surface Microcrack in Si3n4 Ceramic Bearing Roller Based on Adaptive Nano Feature Extraction & Multiscale Depth Fusion | |
| Tang et al. | A robust algorithm for rapid pre-alignment of multiple types and sizes of wafers | |
| Xie et al. | Real-time Reconstruction of unstructured scenes based on binocular vision depth | |
| Zhang et al. | Intelligent detection method for welding seam defects of automobile wheel hub based on yolo | |
| Yang et al. | Research on the application of machine vision technology in industrial automation assembly line | |
| Liu et al. | Measurement method for radial runout of sprocket based on machine vision |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |