CN113658137A - Aircraft surface defect detection method and system based on capsule network - Google Patents
Aircraft surface defect detection method and system based on capsule network Download PDFInfo
- Publication number
- CN113658137A CN113658137A CN202110943780.3A CN202110943780A CN113658137A CN 113658137 A CN113658137 A CN 113658137A CN 202110943780 A CN202110943780 A CN 202110943780A CN 113658137 A CN113658137 A CN 113658137A
- Authority
- CN
- China
- Prior art keywords
- capsule network
- sample
- defect
- surface defect
- image
- 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.)
- Pending
Links
- 230000007547 defect Effects 0.000 title claims abstract description 73
- 239000002775 capsule Substances 0.000 title claims abstract description 58
- 238000001514 detection method Methods 0.000 title claims abstract description 21
- 238000000034 method Methods 0.000 claims abstract description 47
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 14
- 238000004364 calculation method Methods 0.000 claims abstract description 8
- 239000013598 vector Substances 0.000 claims description 12
- 238000004590 computer program Methods 0.000 claims description 8
- 238000013507 mapping Methods 0.000 claims description 5
- 230000008569 process Effects 0.000 claims description 4
- 238000012549 training Methods 0.000 claims description 4
- 238000011176 pooling Methods 0.000 claims description 3
- 238000013527 convolutional neural network Methods 0.000 abstract description 2
- 230000006870 function Effects 0.000 description 9
- 238000007689 inspection Methods 0.000 description 8
- 238000010586 diagram Methods 0.000 description 7
- 238000007781 pre-processing Methods 0.000 description 7
- 238000005516 engineering process Methods 0.000 description 6
- 238000005070 sampling Methods 0.000 description 6
- 238000013528 artificial neural network Methods 0.000 description 4
- 230000008859 change Effects 0.000 description 4
- 238000004804 winding Methods 0.000 description 4
- 238000000605 extraction Methods 0.000 description 3
- 238000003062 neural network model Methods 0.000 description 3
- 230000003287 optical effect Effects 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 230000008439 repair process Effects 0.000 description 3
- 230000000007 visual effect Effects 0.000 description 3
- 230000003416 augmentation Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000001914 filtration Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 230000000644 propagated effect Effects 0.000 description 2
- 206010063385 Intellectualisation Diseases 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 230000008034 disappearance Effects 0.000 description 1
- 239000000835 fiber Substances 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 230000005764 inhibitory process Effects 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 230000001502 supplementing effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G06T5/77—
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
- G01N2021/8854—Grading and classifying of flaws
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
- G01N2021/8887—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges based on image processing techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
Abstract
The invention discloses an aircraft surface defect detection method and system based on a capsule network, wherein the method comprises the following steps: obtaining an airplane surface defect sample, and performing downsampling treatment on the airplane surface defect sample; establishing a capsule network model, and inputting the processed airplane surface defect sample into the capsule network model after the model is trained; calculating the defect confidence of the sample according to the loss function in the capsule network model; and taking the calculation result with the maximum confidence coefficient as a final defect prediction result. According to the method and the system, an airplane surface defect detection model based on the capsule network is adopted on the basis of the airplane surface skin defect, and the efficient convolutional neural network algorithm is established, so that the general target detection procedure and steps are simplified, the image defect classification is optimized, and the target detection and classification of the capsule network can be efficiently carried out.
Description
Technical Field
The invention relates to an aircraft surface detection technology, in particular to an aircraft surface defect detection method and system based on a capsule network.
Background
With the continuous development of civil aviation industry, the occurrence frequency of airplane safety accidents is also continuously improved, so that the health and safety inspection of airplanes must be paid attention. The inspection of the surface defects of the airframe at home and abroad at present adopts a mode of combining visual winding inspection with periodic inspection, wherein the visual winding inspection is a method for inspecting the state of each device of the aircraft by a pilot and crew in a visual mode according to a specified route of the winding inspection. With the development of artificial intelligence technology, in particular to a deep neural network, the computer vision technology based on the deep neural network is possible to replace the winding inspection before the flight. The concept of Deep Neural Networks (DNNs) stems from the study of artificial neural networks. DNN presents attribute classes or features by combining low-level features to form more abstract high-level features to discover distributed features of data.
At present, the deep neural network model is used for image recognition at home and abroad, is mainly applied to airport security check, unknown aircraft recognition and the like in the large-scale application in the civil aviation field, and is not applied to civil aviation engineering. The deep neural network model is introduced into the maintenance of the air route to replace manual work to realize the intellectualization of the pre-flight inspection, and compared with the traditional technology based on feature extraction, the deep neural network model is easily influenced by the external environment, and has high recognition rate and strong anti-interference performance.
Disclosure of Invention
One of the purposes of the invention is to provide an aircraft surface defect detection method and system based on a capsule network, which adopt an aircraft surface defect detection model based on the capsule network on the basis of aircraft surface skin defects, simplify the procedures and steps of general target detection and optimize image defect classification by establishing an efficient convolutional neural network algorithm, so that the target detection and classification of the capsule network can be efficiently carried out.
One of the purposes of the invention is to provide an aircraft surface defect detection method and system based on a capsule network, which greatly improve the identification rate of surface defects through a capsule network technology and have strong external interference resistance compared with the traditional technology.
One of the purposes of the invention is to provide an aircraft surface defect detection method and system based on a capsule network, the method and system change the combination mode of Primary Caps layer capsules to reduce the calculation overhead of a dynamic routing algorithm, improve the training speed of the network, and improve the accuracy of model identification by increasing the capsule compression function of the information carrying capacity of the capsules.
In order to achieve at least one of the above-mentioned objects, the present invention further provides an aircraft surface defect detection method based on a capsule network, the method comprising the steps of:
obtaining an airplane surface defect sample, and performing downsampling treatment on the airplane surface defect sample;
establishing a capsule network model, and inputting the processed airplane surface defect sample into the capsule network model after the model is trained;
calculating the defect confidence of the sample according to the loss function in the capsule network model;
and taking the calculation result with the maximum confidence coefficient as a final defect prediction result.
According to one preferred embodiment of the invention, after the surface defect sample of the airplane is obtained, the defect sample is repaired, and the repairing method of the defect sample comprises the following steps:
identifying and determining the edge of the area to be repaired, and selecting a pixel point P with the highest priority on the edge;
establishing a template block by using the pixel point P, and matching a complete image through the template block;
obtaining a minimum sample restoration module by adopting an image matching algorithm;
and updating the confidence of the repaired pixel point and the edge of the repaired area.
According to another preferred embodiment of the present invention, the method for calculating the priority of the pixel point P includes:
P(p)=C(p)D(p);
wherein, C (p) is a confidence term, and the meaning of the representation is the number of known pixel points in the sample block; d (p) is a data item representing the amount of structural information.
According to another preferred embodiment of the present invention, the method performs data enhancement on the defect sample data after acquiring the defect sample data, and the method of data enhancement includes: rotation, flipping and random cropping.
According to another preferred embodiment of the present invention, the method of down-sampling comprises uniformly down-sampling the sample image by cubic convolution interpolation:
wherein row represents an image row, col represents an image column, a is a constant, and x is a distance from 16 adjacent pixels in the target image to P; v, u represent row number offset and column number offset, i, j represent the mapping of the original image on the interpolated image, F represents the mapping function, and F (i + v, j + u) represents the corresponding coordinate point on the interpolated image; s (x) is a cubic convolution interpolation formula.
According to another preferred embodiment of the invention, the convolutional layers of the capsule network are 4 layers, and the number of channels per convolutional layer is 128, and a convolution filter with a dimension of 3 x 3 is used.
According to another preferred embodiment of the present invention, in the process of capsule network training and identification, the pixel values are normalized: first, pixel point values x are calculatediMean and variance of (c):
the normalized value of the pixel is further calculated:
wherein ε represents the convolutional layer parameter, μBRepresenting pixel point values xiAverage value of (a) ("sigma2 BRepresenting pixel point values xiThe variance of (c).
According to another preferred embodiment of the present invention, a dynamic path algorithm is used instead of pooling operations in the capsule network, the dynamic path algorithm comprising:
whereinInput vector, v, being a fully connected structurejAs an output vector, bijIs dynamicRouting parameters, cijFor the tuning parameters, k is the number of dynamic routing parameters, sjThe adjusted intermediate vector is dynamically routed.
In order to achieve at least one of the above objects, the present invention further provides an aircraft surface defect detecting system based on a capsule network, which executes the aircraft surface defect detecting method based on the capsule network,
the present invention further provides a computer-readable storage medium having stored thereon a computer program executable by a processor to perform a method of capsule network-based aircraft surface defect detection as described above.
Drawings
Fig. 1 is a schematic flow chart showing an aircraft surface defect detection method based on a capsule network according to the present invention. Fig. 2 is a schematic diagram showing the architecture of an aircraft surface defect detection and classification system based on a capsule network according to the present invention.
FIG. 3 is a schematic diagram showing a flow of image preprocessing according to the present invention.
Fig. 4 is a schematic diagram showing the comparison of the down-sampling effect of the different methods of the present invention.
Detailed Description
The following description is presented to disclose the invention so as to enable any person skilled in the art to practice the invention. The preferred embodiments in the following description are given by way of example only, and other obvious variations will occur to those skilled in the art. The basic principles of the invention, as defined in the following description, may be applied to other embodiments, variations, modifications, equivalents, and other technical solutions without departing from the spirit and scope of the invention.
It is understood that the terms "a" and "an" should be interpreted as meaning that a number of one element or element is one in one embodiment, while a number of other elements is one in another embodiment, and the terms "a" and "an" should not be interpreted as limiting the number.
Referring to fig. 1-4, the present invention discloses a method and a system for detecting surface defects of an aircraft based on a capsule network, wherein the method mainly comprises the following steps:
the method comprises the steps of firstly obtaining a sample of the surface defect of the airplane, wherein the surface defect of the airplane can be obtained from in-service airplanes and airplane overhaul factories at home and abroad, obtaining an image sample of the surface defect of the airplane in a research mode, summarizing and classifying various samples, further judging whether the obtained image of the surface defect of the airplane needs to be subjected to image restoration because the image is incomplete, and matching and supplementing the missing image because the image of the surface of the airplane possibly has the defects of shielding objects, a photographing mode, a photographing angle and the like, thereby avoiding the problem of poor accuracy of a final output result caused by the influence of the image sample.
That is to say, after obtaining the airplane defect image sample, firstly, it needs to do the preprocessing of the airplane defect image, and the preprocessing process includes repairing the airplane defect sample image, where the repairing method includes:
determining the edge omega of the area to be repaired, wherein the determination mode of the area to be repaired comprises but is not limited to manual identification and machine identification, and if the area to be repaired cannot be determined, the area to be repaired does not exist, and then the repair is finished;
after the area to be repaired is determined, acquiring a pixel point P with the highest priority on the edge of the area to be repaired;
taking the pixel point P with the highest priority as a central structure modeling block, preferentially, the size of the modeling block is 9 x 9, and matching the image in the template block by searching the image of the intact region so that the image in the intact region can be filled into the corresponding modeling block;
matching the images in the smallest sample repairing template block by using an SSD (image matching algorithm);
the confidence of the repaired image point and the edge of the repaired area are further updated.
It is worth mentioning that the calculation formula of the pixel point priority includes:
P(p)=C(p)D(p);
where c (p) is a confidence term, meaning how many known pixel points in the sample block. The larger the c (p) is, the larger the proportion of the known information contained in the pixel block constructed with the pixel point p as the center is, that is, the higher the confidence is, the higher the repair should be prioritized. D (p) is a data item representing the amount of structural information. The larger D (p) indicates the more complex the surface linear structure, and the repair should be prioritized.
Further the preprocessing step further comprises image augmentation of the sample, the image augmentation method including but not limited to rotation, flipping and random cropping, wherein:
rotating: and then rotating the sample image for a certain degree to change the relative position of the sample seed characteristics. The rotation angle is randomly generated by the script between 90-270 degrees.
Turning: and taking a certain point or a certain line as a central axis, and carrying out symmetrical mirror image on the sample.
Random cutting: and randomly cutting defects in the sample, and taking the cut part as a new sample.
The preprocessing step comprises the steps of solving the problem of the purity of the plane surface defect image by adopting a Gaussian blur algorithm, solving the problem of illumination of the plane surface defect image by adopting a white balance method, and solving the problem of sampling blur by adopting a Laplace sharpening method.
The preprocessing step further comprises down-sampling the sample image, wherein the down-sampling method comprises: uniformly downsampling the sample image to 400 × 3 by adopting a cubic convolution interpolation method, wherein a concrete downsampling formula is as follows:
the cubic convolution interpolation method relates to a target pixel point and the calculation of 16 pixel points which surround the target pixel point, namely 4 rows and 4 columns, and the formula comprises the following steps: row represents a row pixel point, col represents a column pixel point, a is a constant and can be set according to requirements, x is the distance from 16 adjacent pixels in a target image to P, v and u represent row number deviation and column number deviation, i and j represent the mapping of an original image on an interpolated image, F (i + v, j + u) represents a corresponding coordinate point on the interpolated image, and S (x) is a cubic convolution interpolation formula, wherein the change of the interpolation formula can change the interpolation effect.
After the preprocessing of the plane surface defect image is finished, a capsule network model is further established, the capsule network is trained, the preprocessed plane surface defect image is input into a convolution layer to be subjected to feature extraction, then the extracted features are input into the capsule network, a dynamic routing algorithm is adopted to replace pooling operation in traditional image calculation, the confidence coefficient of the defect is calculated according to a loss function, the capsule network can predict the plane surface defect image for multiple times, the confidence value of each prediction is calculated, and the maximum confidence value is taken as the final defect prediction result by adopting a non-maximum inhibition algorithm.
Because the capsule network is an airplane surface defect sample based on classification, the trained capsule network can solve the technical problem that the accuracy of image classification is improved by avoiding gradient disappearance in the iterative process of the capsule network.
The convolution layers of the capsule network are set to be 4 layers, the filtering capability and the feature extraction capability of the network are improved in a mode of increasing the depth of the convolution layers, the number of channels of each convolution layer is 128, and a convolution filter with the size of 3 x 3 is adopted for filtering, so that the parameter quantity of the convolution layers is small, and the capsule network has more nonlinearity.
The invention needs to carry out normalization processing on training data, and the normalization processing formula comprises:
then the data is normalized according to the following formula
Wherein ε represents the convolutional layer parameter, μBRepresenting pixel point values xiAverage value of (a) ("sigma2 BRepresenting pixel point values xiM is the number of sampled samples.
Further, the dynamic route adjustment formula is as follows:
whereinInput vector, v, being a fully connected structurejAs an output vector, bijFor dynamic routing parameters, cijFor the tuning parameters, k is the number of dynamic routing parameters, sjThe adjusted intermediate vector is dynamically routed.
Wherein the output vector vjThe calculation function of (a) is:
the invention further carries out network reconstruction on the capsule network, and the reconstruction method comprises the following steps:
inputting: the 28 × 28 feature maps are subjected to convolution operations with a convolution kernel of (3,3) and a step size of 1, to obtain 128 feature maps of 26 × 26: parameters 1, 28, 128, 100352;
the 128 26 × 26 feature maps are convolved (convolution kernel 3 × 3) to 128 24 × 24 feature maps: parameters 128 x 3 x 128 — 147456;
the 128 24 × 24 feature maps are convolved (convolution kernel 3 × 3) to 128 22 × 22 feature maps: parameters 128 x 3 x 128 — 147456;
entering a capsule network layer: from 128 convolution of 22 by 22 signatures (convolution kernel 3 by 3, step size 2) to 128 convolution of 10 by 10 signatures: parameters 128 x 3 x 128 — 147456;
the 128 10 × 10 feature maps are changed into 3 × 16 vectors by the dynamic routing algorithm: no parameters;
therefore, the total number of parameters of the new reconstructed network is: 100352+147456+147456+147456 is 542720. And taking the maximum value of the confidence degrees of the output vectors in the capsule network as a final recognition result.
In particular, according to the embodiments of the present disclosure, the processes described above with reference to the flowcharts may be implemented as computer software programs. For example, embodiments of the present disclosure include a computer program product comprising a computer program embodied on a computer readable medium, the computer program comprising program code for performing the method illustrated in the flow chart. In such an embodiment, the computer program may be downloaded and installed from a network via the communication section, and/or installed from a removable medium. The computer program, when executed by a Central Processing Unit (CPU), performs the above-described functions defined in the method of the present application. It should be noted that the computer readable medium mentioned above in the present application may be a computer readable signal medium or a computer readable storage medium or any combination of the two. The computer readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples of the computer readable storage medium may include, but are not limited to: an electrical connection having one or more wire segments, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the present application, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. In this application, however, a computer readable signal medium may include a propagated data signal with computer readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated data signal may take many forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A computer readable signal medium may also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to: wireless section, wire section, fiber optic cable, RF, etc., or any suitable combination of the foregoing.
The flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.
It will be understood by those skilled in the art that the embodiments of the present invention described above and illustrated in the drawings are given by way of example only and not by way of limitation, the objects of the invention having been fully and effectively achieved, the functional and structural principles of the present invention having been shown and described in the embodiments, and that various changes or modifications may be made in the embodiments of the present invention without departing from such principles.
Claims (10)
1. An aircraft surface defect detection method based on a capsule network is characterized by comprising the following steps:
obtaining an airplane surface defect sample, and performing downsampling treatment on the airplane surface defect sample;
establishing a capsule network model, and inputting the processed airplane surface defect sample into the capsule network model after the model is trained;
calculating the defect confidence of the sample according to the loss function in the capsule network model;
and taking the calculation result with the maximum confidence coefficient as a final defect prediction result.
2. The method for detecting the surface defects of the airplane based on the capsule network as claimed in claim 1, wherein after the surface defect samples of the airplane are obtained, the defect samples are repaired, and the method for repairing the defect samples comprises the following steps:
identifying and determining the edge of the area to be repaired, and selecting a pixel point P with the highest priority on the edge;
establishing a template block by using the pixel point P, and matching a complete image through the template block;
obtaining a minimum sample restoration module by adopting an image matching algorithm;
and updating the confidence of the repaired pixel point and the edge of the repaired area.
3. The method for detecting the surface defect of the airplane based on the capsule network as claimed in claim 2, wherein the method for calculating the priority of the pixel point P comprises: p (c), (p) d (p);
wherein, C (p) is a confidence term, and the meaning of the representation is the number of known pixel points in the sample block; d (p) is a data item representing the amount of structural information.
4. The method for detecting the surface defect of the aircraft based on the capsule network according to claim 1, wherein the method performs data enhancement on the defect sample data after acquiring the defect sample data, and the data enhancement method includes: rotation, flipping and random cropping.
5. The method for detecting the surface defects of the airplane based on the capsule network is characterized in that the method for downsampling comprises the following steps of uniformly downsampling a sample image through a cubic convolution interpolation method:
wherein row represents an image row, col represents an image column, a is a constant, and x is a distance from 16 adjacent pixels in the target image to P; v, u represent row number offset and column number offset, i, j represent the mapping of the original image on the interpolated image, F represents the mapping function, and F (i + v, j + u) represents the corresponding coordinate point on the interpolated image; s (x) is a cubic convolution interpolation formula.
6. The method of claim 1, wherein the number of convolutional layers of the capsule network is 4, the number of channels in each convolutional layer is 128, and a convolutional filter with a dimension of 3 x 3 is used.
7. The method for detecting the surface defects of the airplane based on the capsule network as claimed in claim 1, wherein in the process of the capsule network training and identification, the pixel values are normalized: first, pixel point values x are calculatediMean and variance of (c):
the normalized value of the pixel is further calculated:
wherein ε represents the convolutional layer parameter, μBRepresenting pixel point values xiAverage value of (a) ("sigma2 BRepresenting pixel point values xiThe variance of (c).
8. The method of claim 1, wherein a dynamic path algorithm is used to replace pooling operations in the capsule network, the dynamic path algorithm comprising:
9. An aircraft surface defect detection and classification system based on a capsule network is characterized in that the system executes the aircraft surface defect detection method based on the capsule network according to any one of claims 1 to 8.
10. A computer-readable storage medium, characterized in that the computer-readable storage medium stores a computer program, which can be executed by a processor to perform a method for detecting surface defects of an aircraft based on a capsule network according to any one of claims 1 to 8.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110943780.3A CN113658137A (en) | 2021-08-17 | 2021-08-17 | Aircraft surface defect detection method and system based on capsule network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110943780.3A CN113658137A (en) | 2021-08-17 | 2021-08-17 | Aircraft surface defect detection method and system based on capsule network |
Publications (1)
Publication Number | Publication Date |
---|---|
CN113658137A true CN113658137A (en) | 2021-11-16 |
Family
ID=78480019
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110943780.3A Pending CN113658137A (en) | 2021-08-17 | 2021-08-17 | Aircraft surface defect detection method and system based on capsule network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113658137A (en) |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106971152A (en) * | 2017-03-16 | 2017-07-21 | 天津大学 | A kind of method of Bird's Nest in detection transmission line of electricity based on Aerial Images |
CN108805198A (en) * | 2018-06-08 | 2018-11-13 | Oppo广东移动通信有限公司 | Image processing method, device, computer readable storage medium and electronic equipment |
CN111160256A (en) * | 2019-12-30 | 2020-05-15 | 武汉科技大学 | Automatic identification method and system for transformer substation pointer instrument |
CN111948213A (en) * | 2020-07-09 | 2020-11-17 | 黄鹏 | Lipstick defect detection device based on attention capsule network and detection method thereof |
CN112270659A (en) * | 2020-08-31 | 2021-01-26 | 中国科学院合肥物质科学研究院 | Rapid detection method and system for surface defects of pole piece of power battery |
CN112927163A (en) * | 2021-03-22 | 2021-06-08 | 深圳市唯特视科技有限公司 | Image data enhancement method and device, electronic equipment and storage medium |
-
2021
- 2021-08-17 CN CN202110943780.3A patent/CN113658137A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106971152A (en) * | 2017-03-16 | 2017-07-21 | 天津大学 | A kind of method of Bird's Nest in detection transmission line of electricity based on Aerial Images |
CN108805198A (en) * | 2018-06-08 | 2018-11-13 | Oppo广东移动通信有限公司 | Image processing method, device, computer readable storage medium and electronic equipment |
CN111160256A (en) * | 2019-12-30 | 2020-05-15 | 武汉科技大学 | Automatic identification method and system for transformer substation pointer instrument |
CN111948213A (en) * | 2020-07-09 | 2020-11-17 | 黄鹏 | Lipstick defect detection device based on attention capsule network and detection method thereof |
CN112270659A (en) * | 2020-08-31 | 2021-01-26 | 中国科学院合肥物质科学研究院 | Rapid detection method and system for surface defects of pole piece of power battery |
CN112927163A (en) * | 2021-03-22 | 2021-06-08 | 深圳市唯特视科技有限公司 | Image data enhancement method and device, electronic equipment and storage medium |
Non-Patent Citations (1)
Title |
---|
李小彤: ""基于图像处理和胶囊网络的金属带材表面缺陷检测研究"", 《中国优秀硕士学位论文全文数据库 (基础科学辑)工程科技I辑》, no. 5, pages 21 - 42 * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108764202B (en) | Airport foreign matter identification method and device, computer equipment and storage medium | |
CN108921799B (en) | Remote sensing image thin cloud removing method based on multi-scale collaborative learning convolutional neural network | |
CN109858372B (en) | Lane-level precision automatic driving structured data analysis method | |
CN106897681B (en) | Remote sensing image contrast analysis method and system | |
US10157314B2 (en) | Aerial image processing | |
Abdollahi et al. | Extraction of road features from UAV images using a novel level set segmentation approach | |
CN110827297A (en) | Insulator segmentation method for generating countermeasure network based on improved conditions | |
KR102270834B1 (en) | Method and system for recognizing marine object using hyperspectral data | |
CN115797350B (en) | Bridge disease detection method, device, computer equipment and storage medium | |
CN110490155B (en) | Method for detecting unmanned aerial vehicle in no-fly airspace | |
Doi et al. | The effect of focal loss in semantic segmentation of high resolution aerial image | |
CN113139543A (en) | Training method of target object detection model, target object detection method and device | |
CN113822314A (en) | Image data processing method, apparatus, device and medium | |
CN112001317A (en) | Lead defect identification method and system based on semantic information and terminal equipment | |
CN113792742A (en) | Semantic segmentation method of remote sensing image and training method of semantic segmentation model | |
CN114066755A (en) | Remote sensing image thin cloud removing method and system based on full-band feature fusion | |
CN114812398A (en) | High-precision real-time crack detection platform based on unmanned aerial vehicle | |
CN114841923A (en) | High-precision real-time crack detection method based on unmanned aerial vehicle | |
CN116485779B (en) | Adaptive wafer defect detection method and device, electronic equipment and storage medium | |
CN112200879B (en) | Map lightweight compression transmission method | |
CN113569911A (en) | Vehicle identification method and device, electronic equipment and storage medium | |
CN113537017A (en) | Optical remote sensing image airplane detection method and device based on cascade regression correction | |
CN113658137A (en) | Aircraft surface defect detection method and system based on capsule network | |
Bolten et al. | Evaluation of Deep Learning based 3D-Point-Cloud Processing Techniques for Semantic Segmentation of Neuromorphic Vision Sensor Event-streams. | |
Babanin et al. | The conceptual level of analysis and its place in the infological system of information and analytical support in the tasks of operational assessment of threats posed by unmanned aerial vehicles |
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 |