CN116778260B - Aviation rivet flushness detection method, device and system based on AdaBoost ensemble learning - Google Patents
Aviation rivet flushness detection method, device and system based on AdaBoost ensemble learning Download PDFInfo
- Publication number
- CN116778260B CN116778260B CN202311038170.4A CN202311038170A CN116778260B CN 116778260 B CN116778260 B CN 116778260B CN 202311038170 A CN202311038170 A CN 202311038170A CN 116778260 B CN116778260 B CN 116778260B
- Authority
- CN
- China
- Prior art keywords
- rivet
- point
- region
- training set
- point cloud
- 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.)
- Active
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 35
- 238000004364 calculation method Methods 0.000 claims abstract description 35
- 238000000034 method Methods 0.000 claims abstract description 31
- 230000011218 segmentation Effects 0.000 claims abstract description 20
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 19
- 238000000605 extraction Methods 0.000 claims description 15
- 238000011176 pooling Methods 0.000 claims description 9
- 102100029469 WD repeat and HMG-box DNA-binding protein 1 Human genes 0.000 claims 1
- 101710097421 WD repeat and HMG-box DNA-binding protein 1 Proteins 0.000 claims 1
- 238000005303 weighing Methods 0.000 claims 1
- 239000013598 vector Substances 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 230000001419 dependent effect Effects 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 239000003550 marker Substances 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 238000003672 processing method Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000000691 measurement method Methods 0.000 description 1
- 238000012821 model calculation Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
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
- Image Analysis (AREA)
Abstract
The application discloses an aviation rivet flushness detection method based on AdaBoost integrated learning, which comprises the following steps: acquiring a point cloud model training set of a marked rivet regionCalculating single-point local characteristics; adaBoost-based integrated learning algorithm, training set is usedGenerating a strong classifier for segmenting rivet regions; point cloud model for acquiring rivet-containing area to be testedPoint cloud modelIs a single point of each ofPerforming a near point query, calculating local characteristics and recording the local characteristics as a data setThe method comprises the steps of carrying out a first treatment on the surface of the Will be assembledInputting the rivet region into a strong classifier for dividing the rivet region to obtain a divided rivet regionNon-rivet areasThe method comprises the steps of carrying out a first treatment on the surface of the To rivet regionNon-rivet areasAnd performing the calculation of the flushness of the target surface. According to the application, the strong classifier formed by the weak classifiers iterated for many times is used for carrying out semantic segmentation on the point cloud data on the surface of the skin, so that high-precision rivet flushness calculation is realized, and the riveting quality of the aircraft is better controlled.
Description
Technical Field
The application relates to the technical field of digital detection of aircraft rivets, in particular to an aircraft rivet flushness detection method, device and system based on AdaBoost integrated learning.
Background
Riveting is a critical process used to securely join together various components or structural elements during aircraft manufacturing. However, due to manufacturing and assembly complexity, there may be rivet flushness issues in aircraft staking. The flushness of a rivet refers to the difference in flatness or perpendicularity between the riveting surface and the rivet head. This has a significant impact on the aerodynamic profile and performance of the aircraft. At present, the number of rivets on the aircraft skin is numerous, the detection efficiency and the reliability are low by using the traditional manual method, quantitative detection cannot be realized, and the method is too dependent on experience. Although the rivet can be identified by the image processing method, the image lacks three-dimensional information, and the flatness cannot be detected.
The three-dimensional laser scanning technology can efficiently acquire three-dimensional information of the aircraft skin surface, and has the advantages of high precision and accurate reflection of the real shape. However, since the difference between the rivet point and the non-rivet point of the skin surface is small, the conventional point cloud segmentation algorithm has difficulty in effectively distinguishing the rivet region, so that calculation of flatness is performed. Thus, the present study is directed to the problem of quantitative detection of rivet flatness of aircraft skin surfaces; therefore, the application provides a digital twinning-based real-time measurement method for assembly gaps to solve the problems.
Disclosure of Invention
In order to solve the problems, an aviation rivet flushness detection method, device and system based on AdaBoost ensemble learning are provided, and the method, device and system aim to solve the problem that small differences between rivet points and non-rivet points on the surface of a skin are difficult to effectively distinguish rivet areas by a traditional point cloud segmentation algorithm, so that calculation of flushness is performed; according to the application, based on an AdaBoost integrated learning algorithm, the strong classifier formed by the weak classifier iterated for many times is used for carrying out semantic segmentation on the point cloud data on the skin surface, so that high-precision rivet flushness calculation is realized, the riveting quality of the aircraft is better controlled, and the performance of the aircraft is improved.
In order to achieve the above purpose, the present application provides the following technical solutions: the application provides an aviation rivet flushness detection method based on AdaBoost integrated learning, which comprises the following steps:
s1, acquiring a point cloud model training set of a marked rivet regionComputing training set +.>A medium single point local feature;
s2, based on AdaBoost integrated learning algorithm, training set is usedThe medium single-point local features generate a strong classifier for segmenting rivet regions;
s3, obtaining the to-be-detectedPoint cloud model with rivet areaPoint cloud model->Is>Performing a nearby point query and calculating a point cloud model +.>Is recorded as a data set +.>；
S4, gathering the dataInputting the rivet region segmentation into a strong classifier to obtain segmented rivet regions and non-rivet regions;
s5, aligning rivet areasAnd non-rivet area->And carrying out the calculation of the flushness of the target surface to obtain a detection result.
Further, in step S1, a training set of the point cloud model of the marked rivet region is obtainedComputing training set +.>The single-point local feature of the medium specifically comprises the following steps:
s11, training setI < th > point->Using kd-tree to inquire the adjacent points;
s12, calculating a target point based on the adjacent pointsComprises a Gaussian curvature>PFH descriptor->And Density->；
S13, single point of recordIs characterized by->Memory training set->Is as followsWherein->Is the number of samples in the training set, +.>，For the sample feature of point i, +.>For the marker set of point i, +.>。
Further, the training set is used in step S2The strong classifier for dividing the rivet region is generated by the single-point local features, and specifically comprises the following steps:
s21, initializing sample weights, and weighting each sample in the training setInitialized to->；
S22, starting iterative training of a weak classifier model,a weak classifier model for the t-th iteration, wherein +.>Is a set of sample features, +.>Let T iterations be needed ∈ ->；
S23, calculating the classification error rate of the weak classifier modelThe calculation formula is as follows:
；
wherein,is a sample feature->Is (are) true category->Is a weak classifier model->For sample characteristicsClassification result of->Weighting the current sample;
s24, calculating the weight of the weak classifier modelThe calculation formula is as follows:
；
s25, updating the sample weight according to the classification accuracy of the current weak classifier modelThe calculation formula is as follows:
；
s26, repeating S22-S25 for T-round iteration to obtainWeak classifier model and corresponding weights +.>And then combining them into a strong classifier, the calculation formula of which is as follows:
；
wherein,the representation will->For its sign, i.e. when->Is +1 when->And is-1.
Further, the weak classifier model in step S22 includes two convolution layersAnd->Each convolution layer is followed by a maximum pooling layer->、/>In->And a full connection layer FC is arranged.
Further, the convolution layerThe convolution kernel size of (2) is 3x3, the convolution kernel depth is 6, the step size is 1, and the maximum pooling layer is +.>The convolution kernel size of (2 x 2), step size of 2, said convolution layer +.>The convolution kernel size of (2) is 3x3, the convolution kernel depth is 6, the step size is 1, and the maximum pooling layer is +.>The convolution kernel size of (2) is 2x2, the step size is 2.
Further, in step S3, the point cloud model is obtainedComprises calculating the point Gaussian curvature +.>And PFH descriptor->And local Density->Record as data set->。
Further, to the rivet regionAnd non-rivet area->The method comprises the following steps of:
s51, using a region growing algorithm to perform rivet regionDividing ∈>Divided into->Individual rivet areas，/>；
S52, using RANSAC to the non-rivet areaPerforming plane fitting and recording the fitted planeIs->；
S53, calculatingEach point of->For plane->Is>，/>，/>Is->The number of midpoints;
s54, calculatingEach point in the region is +.>Is>Mean value of>Obtaining the ith rivet->Is a high level of (2).
The technical scheme also provides a device for realizing the aviation rivet flushness detection method based on AdaBoost integrated learning, which comprises the following steps:
training set local feature extraction moduleThe training set local feature extraction module is used for obtaining a point cloud model training set marked with rivet areasComputing training set +.>A medium single point local feature;
the strong classifier generating module is used for using a training set based on an AdaBoost integrated learning algorithmThe medium single-point local features generate a strong classifier for segmenting rivet regions;
the local characteristic extraction module of the point cloud model is used for obtaining the point cloud model of the rivet-containing region to be detectedPoint cloud model->Is>Performing the inquiry of the nearby points and calculating the point cloud modelIs recorded as a data set +.>；
The rivet region segmentation module is used for inputting the data set into the strong classifier to segment rivet regions, and obtaining segmented rivet regionsAnd non-rivet area->；
The detection result acquisition module is used for acquiring a rivet regionNon-rivet areasAnd carrying out the calculation of the flushness of the target surface to obtain a detection result.
The technical scheme also provides a system for realizing the aviation rivet flushness detection method based on AdaBoost integrated learning, which comprises the following steps:
a processor;
a memory;
and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the processor, the programs for a computer to perform the methods described above.
By the technical scheme, the application provides an aviation rivet flushness detection method, device and system based on AdaBoost integrated learning. The method has at least the following beneficial effects:
the rivet flatness on the aircraft skin has important influence on the aerodynamic shape and performance of the aircraft, and the traditional manual method is low in detection efficiency and reliability, cannot realize quantitative detection and is too dependent on experience. Rivets can be identified based on an image processing method, but images lack three-dimensional information and cannot be detected in flatness; the three-dimensional laser scanning technology can efficiently acquire three-dimensional information of the aircraft skin surface, and has the advantages of high precision and accurate reflection of the real shape; however, since the difference between the rivet point and the non-rivet point of the skin surface is small, the rivet region is difficult to be effectively distinguished by the traditional point cloud segmentation algorithm, so that the calculation of the flatness is performed; therefore, the rivet flushness detection is carried out by utilizing the three-dimensional laser scanning technology, the strong classifier formed by the weak classifier iterated for many times is used for carrying out semantic segmentation on the point cloud data on the surface of the skin based on the AdaBoost integrated learning algorithm, the accurate segmentation on the rivet region can be realized, the detection precision and efficiency are improved, and the flushness calculation is carried out by the method, so that the high-precision rivet flushness calculation is further realized, the riveting quality of the aircraft is better controlled, and the aircraft performance is 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 rivet level detection method based on AdaBoost ensemble learning;
FIG. 2 is a diagram of a weak classifier network according to the present application;
FIG. 3 is a graph of rivet extraction results provided by the practice of the present application;
FIG. 4 is a graph of the calculation result of rivet flatness provided by the embodiment of the application;
fig. 5 is a schematic block diagram of a device of the rivet flushness detection method based on AdaBoost ensemble learning.
In the figure: 100, a training set local feature extraction module; 200 strong classifier generating module; a 300-point cloud model local feature extraction module; 400 rivet zone segmentation modules; 500 to obtain a detection result module.
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 fig. 1-5, a specific implementation manner of the present embodiment is shown, in the method of the present embodiment, based on an AdaBoost integrated learning algorithm, a strong classifier formed by a weak classifier iterated multiple times is used to perform semantic segmentation on point cloud data on a skin surface, so as to further implement high-precision calculation of rivet flushness, thereby better controlling riveting quality of an aircraft, improving performance of the aircraft, and solving the problem that a traditional point cloud segmentation algorithm is difficult to effectively distinguish a rivet region from a small difference between a rivet point and a non-rivet point on the skin surface, so as to perform calculation of flushness.
Referring to fig. 1, a rivet level detection method based on AdaBoost ensemble learning includes the following steps:
s1, acquiring a point cloud model training set of a marked rivet regionComputing training set +.>A medium single point local feature;
specifically, step S1 includes the steps of:
s11, training setI < th > point->Using kd-tree to inquire the adjacent points; inquired +.>The nearest neighbor point is marked as point set +.>Wherein->Is the target point->Is the nearest neighbor of (a);
s12, calculating a target point based on the adjacent pointsComprises a Gaussian curvature>PFH descriptor->And Density->；
Wherein the target point is calculatedGaussian curvature +.>PFH descriptor->And Density->The method specifically comprises the following steps:
s121 is toIs>Calculate its relative to the target point->Coordinate offset vector +.>：
；
S122, calculating covariance matrix of neighborhood points：
；
S123, opposite covariance matrixDecomposing the characteristic value to obtain characteristic value +.>And->And their corresponding feature vectors +.>And->Calculating the available Gaussian curvature +.>：
；
S124, at the target pointThe coordinate system is defined as follows:
；
wherein the method comprises the steps ofFor the target point->Normal vector of->Is->Three axes of the coordinate system defined above;
；
wherein the method comprises the steps ofFor->Normal vector of (2);
s125, utilizeConstructing PFH operator->；
S126, for the coordinate offset vector calculated in step S121Taking->，Calculate->Point cloud Density at->The calculation formula is expressed as follows:
；
s13, notePoint(s)Is characterized by->Memory training set->Is as followsWherein->Is the number of samples in the training set, +.>，For the sample feature of point i, +.>For the marker set of point i, +.>。
S2, based on AdaBoost integrated learning algorithm, training set is usedThe medium single-point local features generate a strong classifier for segmenting rivet regions;
specifically, step S2 includes the steps of:
s21, initializing sample weights, and weighting each sample in the training setInitialized to->Wherein->Is the number of samples in the training set;
s22, starting iterative training of a weak classifier model, and normalizing the current sample weightUse the current sample weight +.>Training a weak classifier model, +.>A weak classifier model for the t-th iteration, wherein +.>Is a set of sample features, +.>Let T iterations be needed ∈ ->；
In particular, the weak classifier model includes two convolutional layersAnd->As shown in FIG. 2, each convolution layer is followed by a maximum pooling layer +.>、/>In->A full connection layer FC is configured at the back;
in particular, a convolution layerThe convolution kernel size of (2) is 3x3, convolutionCore depth 6, step size 1, maximum pooling layer->The convolution kernel size of (2 x 2), step size of 2, convolution layer +.>The convolution kernel size of (2) is 3x3, the convolution kernel depth is 6, the step size is 1, and the maximum pooling layer is +.>The convolution kernel size of (2) is 2x2, the step length is 2;
s23, calculating the classification error rate of the weak classifier modelThe calculation formula is as follows:
；
wherein,is a sample feature->Is (are) true category->Is a weak classifier model->For sample characteristicsClassification result of->Weighting the current sample;
s24, calculating the weight of the weak classifier modelCalculation thereofThe formula is as follows:
；
s25, updating the sample weight according to the classification accuracy of the current weak classifier modelThe calculation formula is as follows:
；
s26, repeating S22-S25 for T-round iteration to obtainWeak classifier model and corresponding weights +.>And then combining them into a strong classifier, the calculation formula of which is as follows:
；
wherein,the representation will->For its sign, i.e. when->Is +1 when->And is-1.
S3, acquiring a point cloud model of the rivet-containing region to be testedPoint cloud model->Is>Performing a nearby point query and calculating a point cloud model +.>Is recorded as a data set +.>；
Specifically, in step S3, a point cloud model is calculatedComprises calculating the point Gaussian curvature +.>And PFH descriptor->And local Density->Record as data set->；
First, an aircraft skin point cloud model of a rivet-containing region is acquired using a laser scannerSecondly, point cloud modelIs fixed by a fixed amount using the kd-Tree algorithm for each single point of (2)>Is based on the proximity point to calculate the gaussian curvature of the point +.>PFH descriptor->And local Density->The specific calculation process is the same as S121-S126, and the calculation result is marked as a set +.>。
S4, gathering the dataInputting the rivet region segmentation into a trained strong classifier to obtain a classification result of each single point sample, namely a rivet segmentation result, as shown in fig. 3, wherein fig. 3 is a rivet extraction result diagram provided by the implementation of the application, and outputting segmented point cloud groups->Point cloud group->Including rivet region->And non-rivet area->Finally, the segmented rivet region is obtained>And non-rivet area->As shown in FIG. 2, rivet region is shown as Rivet region +.>Non-rivet region is the Non-rivet region +.>。
S5, aligning rivet areasAnd non-rivet area->Performing the flushness calculation of the target surface to obtain a detection result;
in particular, to the rivet regionAnd non-rivet area->The method for calculating the flushness of the target surface to obtain the detection result comprises the following steps:
s51, using a region growing algorithm to perform rivet regionDividing ∈>Divided into->Individual rivet areas，/>；
S52, using RANSAC to the non-rivet areaPerforming plane fitting, wherein the fitted plane is marked as +.>Record->The equation of (2) is: />；
S53, independent rivet areaEach point of->Calculate its distance plane +.>Is>，/>，/>Is->The number of midpoints, its vertical distance +.>The calculation formula of (2) is as follows:
；
s54, calculatingEach point in the region is +.>Is>Mean value of>I.e. the ith rivetAs shown in FIG. 4, FIG. 4 is a graph showing the calculation result of rivet flatness provided by the implementation of the present application, wherein the average value +.>The calculation formula of (2) is as follows:
；
the technical scheme also provides a device for realizing the method for detecting the rivet flushness based on AdaBoost integrated learning, as shown in FIG. 5, FIG. 5 is a device schematic block diagram of the method for detecting the rivet flushness based on AdaBoost integrated learning, which comprises the following steps:
the training set local feature extraction module 100, wherein the training set local feature extraction module 100 is used for obtaining a point cloud model training set of a marked rivet regionComputing training set +.>A medium single point local feature;
the strong classifier generating module 200 is configured to use training set based on AdaBoost ensemble learning algorithm by the strong classifier generating module 200The medium single-point local features generate a strong classifier for segmenting rivet regions;
the local feature extraction module 300 of the point cloud model, wherein the local feature extraction module 300 of the point cloud model is used for obtaining the point cloud model of the rivet-containing region to be detectedPoint cloud model->Is>Performing a nearby point query and calculating a point cloud model +.>Is recorded as a data set +.>；
The rivet region segmentation module 400 is configured to input the data set into a strong classifier to segment the rivet region, thereby obtaining a segmented rivet regionAnd non-rivet area->；
The detection result acquisition module 500 is used for acquiring a rivet region of the rivetAnd non-rivet area->And carrying out the calculation of the flushness of the target surface to obtain a detection result.
The technical scheme also provides a system for realizing the aviation rivet flushness detection method based on AdaBoost integrated learning, which comprises the following steps:
a processor;
a memory;
and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the processor, the programs for a computer to perform the methods described above.
According to the application, based on an AdaBoost integrated learning algorithm, the strong classifier formed by the weak classifier iterated for many times is used for carrying out semantic segmentation on the point cloud data on the skin surface, so that high-precision rivet flushness calculation is realized, the riveting quality of the aircraft is better controlled, and the performance of the aircraft is improved.
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. An aircraft rivet flushness detection method based on AdaBoost integrated learning is characterized by comprising the following steps:
s1, acquiring a point cloud model training set D of a marked rivet region, and calculating single-point local features in the training set D;
s2, generating a strong classifier for dividing the rivet region by using single-point local features in the training set D based on an AdaBoost integrated learning algorithm;
wherein, the step S2 specifically includes the following steps:
s21, initializing sample weights, and weighing the weight w of each sample in the training set i Initializing to
S22, starting iterative training of a weak classifier model, g t (x) A weak classifier model for the t-th iteration, where x is the set of sample features, x= { x 1 ,x 2 ,…,x m Setting the iteration time to be T times, and then T epsilon T;
s23, calculating classification error rate co of the weak classifier model t The calculation formula is as follows:
wherein y is i Is the sample characteristic x i True category, g t (x i ) Is a weak classifier model g t (x) For sample feature x i Is used for classifying the result of the classification of (a),weighting the current sample;
s24, calculating the weight alpha of the weak classifier model t The calculation formula is as follows:
s25, updating the sample weight w according to the classification accuracy of the current weak classifier model i The calculation formula is as follows:
s26, repeating S22-S25 for T-round iteration to obtain T weak classifier models and corresponding weights alpha t They are combined into a strong classifier whose calculation formula is as follows:
wherein sign (z) represents that z is the sign thereof, namely +1 when z is more than or equal to 0, and-1 when z is less than 0;
s3, acquiring a point cloud model M of the rivet-containing region to be tested, wherein each single point M of the point cloud model M i Performing near point inquiry, calculating single-point local characteristics of the point cloud model M, and recording the single-point local characteristics as a data set N;
s4, inputting the data set N into a strong classifier to perform rivet region segmentation to obtainDivided rivet region Q 1 And non-rivet region Q 2 ；
S5, to rivet region Q 1 And non-rivet region Q 2 Performing the flushness calculation of the target surface to obtain a detection result;
wherein, the step S5 specifically includes the following steps:
s51, using a region growing algorithm to rivet the region Q 1 Dividing Q 1 Divided into k independent rivet areas Q 1i ，i∈(0,k)；
S52, using RANSAC to the non-rivet region Q 2 Performing plane fitting, and marking the fitted plane as alpha;
s53, calculate Q 1i Each point q of (3) j (x 1 ,y 1 ,z 1 ) Perpendicular distance h to plane alpha j J epsilon (0, n), n is Q 1i The number of midpoints;
s54, calculate Q 1i Perpendicular distance h of each point in the region to plane alpha j The average value H of (i) is the ith rivet Q 1i Is a high level of (2).
2. The method for detecting the level of the aviation rivet based on AdaBoost ensemble learning according to claim 1, wherein in the step S1, a training set D of a point cloud model of a marked rivet area is obtained, and single-point local features in the training set D are calculated, specifically comprising the following steps:
s11, for the ith point P of the training set D i Using kd-tree to inquire the adjacent points;
s12, calculating a target point P based on the adjacent points i Including gaussian curvature K, PFH descriptor F and density d;
s13, record single point P i Is characterized by x i = { K, F, D }, the dataset of training set D is remembered as D = { (x) 1 ,y 1 ),(x 2 ,y 2 ),…,(x m ,y m ) M is the number of samples in the training set, i.e. (0, m), x i For sample feature at point i, y i Set of markers for the ith point, y i ∈{-1,+1}。
3. The method for detecting the aircraft rivet level based on AdaBoost ensemble learning according to claim 1, wherein the weak classifier model in the step S22 comprises two convolution layers Conv_1 and Conv_2, wherein each convolution layer is followed by a maximum pooling layer pool_1 and pool_2, and a full connection layer FC is configured after the pool_2.
4. The method for detecting the level of the aviation rivet based on AdaBoost ensemble learning according to claim 3, wherein the convolution kernel size of the convolution layer Conv_1 is 3x3, the convolution kernel depth is 6, the step length is 1, the convolution kernel size of the maximum pooling layer pool_1 is 2x2, the step length is 2, the convolution kernel size of the convolution layer Conv_2 is 3x3, the convolution kernel depth is 6, the step length is 1, the convolution kernel size of the maximum pooling layer pool_2 is 2x2, and the step length is 2.
5. The method for detecting the level of the blind rivet based on AdaBoost ensemble learning according to claim 1, wherein the single-point local feature of the point cloud model M in the step S3 comprises calculating the point Gaussian curvature K, the PFH descriptor F and the local density d, and recording the calculated point Gaussian curvature K, the PFH descriptor F and the local density d as the data set N.
6. An apparatus for implementing the AdaBoost ensemble learning-based method for detecting the flatness of an aircraft rivet according to any one of the preceding claims 1 to 5, characterized by comprising:
the training set local feature extraction module (100), wherein the training set local feature extraction module (100) is used for obtaining a point cloud model training set D of a marked rivet region and calculating single-point local features in the training set D;
the strong classifier generation module (200) is used for generating a strong classifier for dividing the rivet region by using single-point local features in the training set D based on an AdaBoost ensemble learning algorithm;
the device comprises a point cloud model local feature extraction module (300), wherein the point cloud model local feature extraction module (300) is used for acquiring a rivet-containing region to be detectedPoint cloud model M, each single point M of point cloud model M i Performing near point inquiry, calculating single-point local characteristics of the point cloud model M, and recording the single-point local characteristics as a data set N;
a rivet region segmentation module (400), wherein the rivet region segmentation module (400) is used for inputting the data set N into a strong classifier to carry out rivet region segmentation to obtain a segmented rivet region Q 1 And non-rivet region Q 2 ；
A detection result acquisition module (500), wherein the detection result acquisition module (500) is used for the rivet region Q 1 And non-rivet region Q 2 And carrying out the calculation of the flushness of the target surface to obtain a detection result.
7. A system for implementing an AdaBoost ensemble learning-based aerial rivet flushness detection method according to any one of the preceding claims 1-5, characterized in that it comprises:
a processor;
a memory;
and one or more programs, wherein the one or more programs are stored in memory and configured to be executed by the processor, the program for a computer to perform the method of any of claims 1-5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311038170.4A CN116778260B (en) | 2023-08-17 | 2023-08-17 | Aviation rivet flushness detection method, device and system based on AdaBoost ensemble learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311038170.4A CN116778260B (en) | 2023-08-17 | 2023-08-17 | Aviation rivet flushness detection method, device and system based on AdaBoost ensemble learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116778260A CN116778260A (en) | 2023-09-19 |
| CN116778260B true CN116778260B (en) | 2023-11-17 |
Family
ID=87994793
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311038170.4A Active CN116778260B (en) | 2023-08-17 | 2023-08-17 | Aviation rivet flushness detection method, device and system based on AdaBoost ensemble learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116778260B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN204152918U (en) * | 2014-10-08 | 2015-02-11 | 江苏神州碳制品有限公司 | A kind of wind-force carbon brush joint rivet |
| CN111028220A (en) * | 2019-12-11 | 2020-04-17 | 南京航空航天大学 | Automatic detection method for levelness of point cloud rivet |
| CN112053426A (en) * | 2020-10-15 | 2020-12-08 | 南京航空航天大学 | Deep learning-based large-scale three-dimensional rivet point cloud extraction method |
| CN114626470A (en) * | 2022-03-18 | 2022-06-14 | 南京航空航天大学深圳研究院 | Aircraft skin key feature detection method based on multi-type geometric feature operator |
| 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 |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2009282685A (en) * | 2008-05-21 | 2009-12-03 | Sony Corp | Information processor, information processing method, and program |
-
2023
- 2023-08-17 CN CN202311038170.4A patent/CN116778260B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN204152918U (en) * | 2014-10-08 | 2015-02-11 | 江苏神州碳制品有限公司 | A kind of wind-force carbon brush joint rivet |
| CN111028220A (en) * | 2019-12-11 | 2020-04-17 | 南京航空航天大学 | Automatic detection method for levelness of point cloud rivet |
| CN112053426A (en) * | 2020-10-15 | 2020-12-08 | 南京航空航天大学 | Deep learning-based large-scale three-dimensional rivet point cloud extraction method |
| CN114626470A (en) * | 2022-03-18 | 2022-06-14 | 南京航空航天大学深圳研究院 | Aircraft skin key feature detection method based on multi-type geometric feature operator |
| 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 |
Non-Patent Citations (3)
| Title |
|---|
| Quality verification of a riveting process by statistical analysis and neural network;Wang, H. L;《Experimental Techniques 》;34-42 * |
| 基于三维点云的机械加工精度自动检测方法;杨坤等;《工具技术》;第53卷(第4期);112-116 * |
| 基于图像-点云映射的铆钉平齐度高效检测;国荣辉等;《激光与光电子学进展》;第58卷(第20期);320-328 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116778260A (en) | 2023-09-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110070074B (en) | Method for constructing pedestrian detection model | |
| CN107657279B (en) | Remote sensing target detection method based on small amount of samples | |
| CN108052942B (en) | Visual image recognition method for aircraft flight attitude | |
| Mohlin et al. | Probabilistic orientation estimation with matrix fisher distributions | |
| CN110633604B (en) | Information processing method and information processing apparatus | |
| US20100322474A1 (en) | Detecting multiple moving objects in crowded environments with coherent motion regions | |
| CN110610505A (en) | Image segmentation method fusing depth and color information | |
| CN109815814B (en) | Face detection method based on convolutional neural network | |
| CN109284779A (en) | Object detection method based on deep full convolution network | |
| CN109708643B (en) | Evaluation and selection method for asteroid surface optical navigation road sign | |
| CN110942077B (en) | Feature line extraction method based on weight local change degree and L1 median optimization | |
| CN110852317A (en) | Small-scale target detection method based on weak edge | |
| CN111553280A (en) | Target part identification method based on deep learning | |
| CN109919215B (en) | Target detection method for improving characteristic pyramid network based on clustering algorithm | |
| CN116778260B (en) | Aviation rivet flushness detection method, device and system based on AdaBoost ensemble learning | |
| Sahloul et al. | An accurate and efficient voting scheme for a maximally all-inlier 3D correspondence set | |
| CN111275748B (en) | Point cloud registration method based on laser radar in dynamic environment | |
| CN111563904A (en) | Wing fork ear point cloud segmentation method based on three-dimensional feature descriptor | |
| CN117197159A (en) | Efficient and high-quality high-temperature object point cloud segmentation method | |
| Scully et al. | Simulating solar radio bursts using generative adversarial networks | |
| CN110415281A (en) | A kind of point set rigid registration method based on Loam curvature weighting | |
| Yang et al. | Research on Edge Detection of LiDAR Images Based on Artificial Intelligence Technology | |
| CN115546205A (en) | Planar point cloud contour line generation method based on density field perception | |
| CN112509018B (en) | Quaternion space optimized three-dimensional image registration method | |
| CN115205462A (en) | Three-dimensional reconstruction method aiming at light section type and based on semi-ordered point cloud |
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 |