CN110222603B - Exponential entropy multiplicative fuzzy defect characteristic analysis reconstruction method based on infrared thermal imaging - Google Patents

Exponential entropy multiplicative fuzzy defect characteristic analysis reconstruction method based on infrared thermal imaging Download PDF

Info

Publication number
CN110222603B
CN110222603B CN201910433604.8A CN201910433604A CN110222603B CN 110222603 B CN110222603 B CN 110222603B CN 201910433604 A CN201910433604 A CN 201910433604A CN 110222603 B CN110222603 B CN 110222603B
Authority
CN
China
Prior art keywords
defect
matrix
iteration
representing
characteristic
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
Application number
CN201910433604.8A
Other languages
Chinese (zh)
Other versions
CN110222603A (en
Inventor
张博
程玉华
殷春
黄雪刚
张昊楠
薛婷
李毅
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
University of Electronic Science and Technology of China
Original Assignee
University of Electronic Science and Technology of China
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by University of Electronic Science and Technology of China filed Critical University of Electronic Science and Technology of China
Priority to CN201910433604.8A priority Critical patent/CN110222603B/en
Publication of CN110222603A publication Critical patent/CN110222603A/en
Application granted granted Critical
Publication of CN110222603B publication Critical patent/CN110222603B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00Pattern recognition
    • G06F18/20Analysing
    • G06F18/23Clustering techniques
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00Image analysis
    • G06T7/0002Inspection of images, e.g. flaw detection
    • G06T7/0004Industrial image inspection
    • G06T7/0008Industrial image inspection checking presence/absence
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00Scenes; Scene-specific elements
    • G06V20/40Scenes; Scene-specific elements in video content
    • G06V20/46Extracting features or characteristics from the video content, e.g. video fingerprints, representative shots or key frames

Abstract

The invention discloses an exponential entropy multiplicative fuzzy defect feature analysis reconstruction method based on infrared thermal imaging, which extracts and analyzes features of defects with different degrees in different spaces through a reconstruction model and an optimized fuzzy algorithm, so that the defect features with different degrees in different spaces can be accurately divided; meanwhile, in the optimized fuzzy algorithm, a new objective function is constructed, one part of the objective function comprises the product of the praise degree and the hesitation degree, the feature information of elements is enriched, the other part of the objective function comprises the exponential fuzzy entropy, the uncertainty of the feature information is described, the correlation of the gain function of the damage region is enhanced, the features are emphasized, and effective help is provided for distinguishing defects.

Description

Exponential entropy multiplicative fuzzy defect characteristic analysis reconstruction method based on infrared thermal imaging
Technical Field
The invention belongs to the technical field of defect detection, and particularly relates to an exponential entropy multiplicative fuzzy defect feature analysis reconstruction method based on infrared thermal imaging.
Background
In recent years, infrared thermal imaging detection technology has been rapidly developed. The nondestructive testing method has the advantages of no damage to the body, rapidness, high efficiency and the like, can effectively solve the problems of high labor intensity, long period, low efficiency, poor safety and the like of the traditional nondestructive testing method, realizes large-area rapid testing, and saves a large amount of manpower and material resources.
If the surface of the test piece to be tested has defects, the heat distribution of the test piece to be tested is influenced. Heating a test piece to be detected so as to generate a high-temperature area and a low-temperature area, wherein due to temperature difference, heat in the high-temperature area is transferred to the low-temperature area through heat conduction to cause the temperature of different areas of the tested piece to change, acquiring the change process of the temperature of the test piece through a thermal infrared imager, and then sending the acquired thermal image video to a computer for analysis and processing so as to acquire related information of the tested piece and realize qualitative and quantitative detection of defects.
The traditional defect analysis method mainly aims at defect characteristics of different regions in the same space, the defect distribution is clear, the defects can be clearly divided through corresponding technologies, but for the defects on the aviation material, the defects are distributed in the spaces of different layers, and factors such as impact, corrosion and the like cause mutual interference of defect distribution in different degrees, so that the external environment can also influence the defects. The damage information of different spaces is ignored, and thus the defect type of the material is erroneously determined. Currently, due to the influence of the external environment and the properties of the material itself, various known defect types, such as scattered pit defects, surface crack defects, internal delamination defects, internal peeling defects, and the like, are formed, and these known defect types can be used as a modeling basis for defect evaluation.
Since the actual defect information target is intermediate in form and category and has no definite boundary for distinguishing, the FCM fuzzy algorithm is used for classifying and processing the defect information aiming at the defects of different areas. However, the traditional FCM algorithm cannot completely express the characteristics of each element, and defect information of different degrees in different spaces loses a lot of useful pixel characteristics in the processing process. On the other hand, the uncertainty of the variable is also described by fuzzy entropy in the new objective function, so that the relevance of the defect is more clear. Therefore, the method can accurately perform interference elimination clustering on the defects in different spaces.
In order to judge the defect condition of the material more accurately, the invention provides an effective detection method which can judge the damage of the surface space of the material to be detected, remove noise interference and more importantly can obtain the damage condition of the inner layer space more accurately.
Disclosure of Invention
The invention aims to overcome the defects of the prior art, provides an exponential entropy multiplicative fuzzy defect feature analysis reconstruction method based on infrared thermal imaging, and realizes defect detection and feature extraction of a plurality of regions in different spaces through an improved fuzzy algorithm.
In order to achieve the above object, the present invention provides an exponential entropy multiplicative fuzzy defect feature analysis reconstruction method based on infrared thermal imaging, which is characterized by comprising the following steps:
(1) preprocessing of video stream to be detected
(1.1) representing the video stream to be detected as:
Figure BDA0002069822670000021
wherein N isI×NJRepresenting spatial information, NTRepresenting time information;
(1.2) converting the matrix block into a two-dimensional matrix Y through a vector operator Vec ();
Y=[Vec(Y(1)),Vec(Y(2)),…,Vec(Y(NT))]
wherein the content of the first and second substances,
Figure BDA0002069822670000022
NIJ=NI×NJ
(1.3) according to the prior I defect types, rewriting the video stream to be detected into:
Figure BDA0002069822670000023
wherein, Xθ(tθ) Indicating the t-th in the video stream to be detectedθThe frame is divided into a theta defect feature row vector, where theta is 1,2, …, l,
Figure BDA0002069822670000024
αθ(tθ) Is Xθ(tθ) The corresponding characteristic coefficient matrix column vector,
Figure BDA0002069822670000025
Figure BDA0002069822670000026
NCrepresenting the total frame number representing characteristic defects in the video stream to be detected;
(2) defect reconstruction
(2.1) decomposing the rewritten video stream Y into the following parts by adopting a singular value decomposition algorithm: y isT=U∑VTWherein, in the step (A),
Figure BDA0002069822670000027
in the form of a left-hand singular matrix,
Figure BDA0002069822670000028
is a matrix of singular values and is,
Figure BDA0002069822670000029
is a right singular matrix;
(2.2) calculating a whitening matrix W and a mixing coefficient matrix lambda;
Figure BDA0002069822670000031
(2.3) representing the reconstruction defect as X ═ W × Y in a matrix;
(2.4) eliminating the correlation between the reconstructed defect characteristics by a Newton iteration method:
w*=E{xg(wTx)}-E{xg(wTx)}w
wherein X is the row vector of X, W is the row vector of W, G (-) is the derivative of the contrast function G, E {. cndot. } represents the expectation;
(2.5) processing all the rows of the X according to the method in the step (2.4) to obtain a reconstructed defect characteristic information matrix: x*=W*×Y,
Figure BDA0002069822670000032
(2.6) whitening matrix W*Performing time domain analysis to obtain actual distribution conditions of different defect characteristics, and comparing W*Middle row vector w*Trend of the curve of (1), will w*Judging the defects with similar curve trends as the same defect type, otherwise, judging the defects as different defect types;
merging defect frame numbers of the same defect type to form a new defect frame number
Figure BDA0002069822670000033
Wherein m is 1,2, …% and x is less than or equal to l,
Figure BDA0002069822670000034
thus from X*To obtain
Figure BDA0002069822670000035
A reconstructed image of the m-th defect type, denoted
Figure BDA0002069822670000036
Figure BDA0002069822670000037
(3) Feature extraction of reconstructed image
(3.1) reconstruction of an image with a Butterworth filter
Figure BDA0002069822670000038
Performing texture and color difference segmentation;
the characteristic color difference is expressed as:
Figure BDA0002069822670000039
wherein the content of the first and second substances,
Figure BDA00020698226700000310
tito represent
Figure BDA00020698226700000311
Middle pixel point (t)i,tj) Line coordinate of (d), tjTo represent
Figure BDA00020698226700000312
Middle pixel point (t)i,tj) Column coordinates of (D)0Is a threshold value;
the feature texture is represented as:
Figure BDA00020698226700000313
(3.2) optimizing the characteristic chromatic aberration;
(3.2.1) setting an iteration termination condition epsilon, iteration times c and maximum iteration times cmNumber of clusters τ, operation coefficient δ, weight coefficient b, popularity μj(hi)cDegree xi of hesitationj(hi)cMean value of degree of hesitation
Figure BDA00020698226700000314
Cluster center vj cObjective function Jc
(3.2.2) calculating the popularity mu of each single pixel point in each class in the c iterationj(hi)c
Figure BDA0002069822670000041
Wherein h isiRepresenting the pixel value of the ith pixel point in the characteristic chromatic aberration H, wherein j is not equal to s; | l | · | | represents the norm calculation;
(3.2.3) calculating the clustering center v of each class in the c iterationj c
Figure BDA0002069822670000042
Wherein g represents the number of pixel points in the characteristic chromatic aberration H;
(3.2.4) calculating the hesitation xi of each single pixel point in each class during the c iterationj(hi)c
Figure BDA0002069822670000043
(3.2.5) calculating the average value of the hesitations of single pixel points in each class during the c-th iteration
Figure BDA0002069822670000044
Figure BDA0002069822670000045
(3.2.6) calculating an objective function J at the c-th iterationcA value;
Figure BDA0002069822670000046
(3.2.7) and after the c iteration is finished, judging whether the relation formula is met: | Jc-Jc-1If | | ≦ ε, entering step (3.2.8), otherwise, stopping iteration, and jumping to step (3.3);
(3.2.8), judging whether the current iteration number c reaches the set maximum iteration number cmIf c < cmIf so, increasing the current iteration number c by 1, and then returning to the step (3.2.2); otherwise, iteration stops and jumps intoStep (3.3);
(3.3) maximizing the criterion according to the popularity: mj=arg max(μj(hi)c) Find out the maximum praise degree Mj,MjA set of pixels representing a j-th class; then according to MjClassifying the characteristic color difference H, and after classification is finished, v isj cIs assigned to the corresponding MjObtaining the optimized characteristic color difference H*
(3.4) optimized reconstruction characteristics of
Figure BDA0002069822670000047
The invention aims to realize the following steps:
according to the index entropy multiplicative fuzzy defect feature analysis reconstruction method based on infrared thermal imaging, defects of different degrees in different spaces are extracted and subjected to feature analysis through a reconstruction model and an optimized fuzzy algorithm, so that the defect features of different degrees in different spaces can be accurately divided; meanwhile, in the optimized fuzzy algorithm, a new objective function is constructed, one part of the objective function comprises the product of the praise degree and the hesitation degree, the feature information of elements is enriched, the other part of the objective function comprises the exponential fuzzy entropy, the uncertainty of the feature information is described, the correlation of the gain function of the damage region is enhanced, the features are emphasized, and effective help is provided for distinguishing defects.
Meanwhile, the index entropy multiplicative fuzzy defect characteristic analysis and reconstruction method based on infrared thermal imaging also has the following beneficial effects:
(1) by carrying out feature model reconstruction and evaluation analysis on the defects of different spaces and different degrees, accurate defect extraction is realized, and a basic theory is provided for further establishing a defect model impact behavior database;
(2) in the optimized fuzzy objective function, the product of the praise and the hesitation enriches the color difference information of the characteristic defect, the description of the outline is more accurate, and the application of the exponential fuzzy entropy enhances the description of the uncertainty of the characteristic elements, accurately clusters the characteristic elements and ensures that the description of the defect is more complete and accurate;
(3) and the characteristic information of the defect is divided into a characteristic outline and a characteristic chromatic aberration for processing by using a Butterworth filter, so that the whole algorithm process is more efficient.
Drawings
FIG. 1 is a flow chart of an index entropy multiplicative fuzzy defect characteristic analysis and reconstruction method based on infrared thermal imaging;
FIG. 2 is a flow chart of a feature color difference optimization process;
FIG. 3 is a schematic view of a defect signature of a test piece;
FIG. 4 is a schematic view of a feature defect texture and a feature defect color difference;
FIG. 5 is a graph of feature defect color difference after optimization;
fig. 6 is a defect feature map of the extracted reconstructed image.
Detailed Description
The following description of the embodiments of the present invention is provided in order to better understand the present invention for those skilled in the art with reference to the accompanying drawings. It is to be expressly noted that in the following description, a detailed description of known functions and designs will be omitted when it may obscure the subject matter of the present invention.
Examples
FIG. 1 is a flow chart of an exponential entropy multiplicative fuzzy defect characteristic analysis and reconstruction method based on infrared thermal imaging.
In this embodiment, as shown in fig. 1, the method for reconstructing exponential entropy multiplicative fuzzy defect characteristic analysis based on infrared thermal imaging mainly includes three steps: s1, preprocessing a video stream to be detected; s2, reconstructing defects; s3, extracting the characteristics of the reconstructed image;
in the following we will describe in detail the three steps described above.
S1 preprocessing video stream to be detected
S1.1, representing a video stream to be detected by using a matrix block as follows:
Figure BDA0002069822670000061
wherein N isI×NJRepresenting spatial information, NTRepresenting time information;
s1.2, converting the matrix block into a two-dimensional matrix Y through a vector operator Vec ();
Y=[Vec(Y(1)),Vec(Y(2)),…,Vec(Y(NT))]
wherein the content of the first and second substances,
Figure BDA0002069822670000062
NIJ=NI×NJ
s1.3, in order to reconstruct different defect information, rewriting a video stream to be detected into the following types according to prior one defect types:
Figure BDA0002069822670000063
wherein, Xθ(tθ) Indicating the t-th in the video stream to be detectedθThe frame is divided into a theta defect feature row vector, where theta is 1,2, …, l,
Figure BDA0002069822670000064
αθ(tθ) Is Xθ(tθ) The corresponding characteristic coefficient matrix column vector,
Figure BDA0002069822670000065
Figure BDA0002069822670000066
NCrepresenting the total frame number representing characteristic defects in the video stream to be detected;
s2 defect reconstruction
S2.1, decomposing the rewritten video stream Y by using a singular value decomposition algorithm to reconstruct the surface defect characteristicsComprises the following steps: y isT=U∑VTWherein, in the step (A),
Figure BDA0002069822670000067
in the form of a left-hand singular matrix,
Figure BDA0002069822670000068
is a matrix of singular values and is,
Figure BDA0002069822670000069
is a right singular matrix;
the specific process of decomposing the rewritten video stream Y by the singular value decomposition algorithm is as follows:
1) constructing a covariance matrix A;
Figure BDA0002069822670000071
2) calculating a singular value;
by calculation of
Figure BDA0002069822670000072
Get the right singular value
Figure BDA0002069822670000073
Singular value
Figure BDA0002069822670000074
And left singular value
Figure BDA0002069822670000075
Wherein the content of the first and second substances,
Figure BDA0002069822670000076
representing a characteristic value;
3) according to singular value
Figure BDA0002069822670000077
Constructing a singular value matrix sigma';
4) setting an empirical threshold aF、aB、aNAnd are all positive integers;
in the matrix of singular values ∑, the singular values are combined
Figure BDA0002069822670000078
Satisfy the requirement of
Figure BDA0002069822670000079
In (1)
Figure BDA00020698226700000710
An
Figure BDA00020698226700000711
Representing surface defect features by satisfying singular values
Figure BDA00020698226700000712
Of the hour
Figure BDA00020698226700000713
An
Figure BDA00020698226700000714
Representing internal spallation defect characteristics, satisfying singular values
Figure BDA00020698226700000715
Of the hour
Figure BDA00020698226700000716
An
Figure BDA00020698226700000717
Representing non-defect region features;
according to
Figure BDA00020698226700000718
The surface defect characteristics,
Figure BDA00020698226700000719
Internal delamination defect characterization and
Figure BDA00020698226700000720
updating the singular value matrix sigma' according to the characteristics of the non-defective area to obtain an updated singular value matrix sigma, and then obtaining a finally rewritten video stream Y which is expressed as YT=U∑VT
S2.2, calculating a whitening matrix W and a mixing coefficient matrix lambda;
Figure BDA00020698226700000721
s2.3, representing the reconstruction defect as X in a matrix of W multiplied by Y;
s2.4, eliminating the correlation between the reconstruction defect characteristics through a Newton iteration method:
w*=E{xg(wTx)}-E{xg(wTx)}w
wherein X is the row vector of X, W is the row vector of W, G (-) is the derivative of the contrast function G, E {. cndot. } represents the expectation;
s2.5, processing all the rows of the X according to the method in the step S2.4 to obtain a reconstructed defect characteristic information matrix: x*=W*×Y,
Figure BDA00020698226700000722
S2.6 defect feature matrix representing different spaces
Figure BDA00020698226700000723
In the method, a line vector representing corresponding defect characteristics is characterized by an image, and each line represents damage characteristics in different spaces.
By making a whiten matrix W*Performing time domain analysis to obtain actual distribution conditions of different defect characteristics, and comparing W*Middle row vector w*Trend of the curve of (1), will w*Judging the defects with similar curve trends as the same defect type, otherwise, judging the defects as different defect types;
merging defect frame numbers of the same defect type to form a new defect frame number
Figure BDA0002069822670000081
Wherein m is 1,2, …% and x is less than or equal to l,
Figure BDA0002069822670000082
thus from X*To obtain
Figure BDA0002069822670000083
A reconstructed image of the m-th defect type, denoted
Figure BDA0002069822670000084
Figure BDA0002069822670000085
S3, feature extraction of reconstructed image
S3.1, reconstructing the image by using a Butterworth filter
Figure BDA0002069822670000086
Performing texture and color difference segmentation;
the characteristic color difference is expressed as:
Figure BDA0002069822670000087
wherein the content of the first and second substances,
Figure BDA0002069822670000088
tito represent
Figure BDA0002069822670000089
Middle pixel point (t)i,tj) Line coordinate of (d), tjTo represent
Figure BDA00020698226700000810
Middle pixel point (t)i,tj) Column coordinates of (D)0Is a threshold value;
the feature texture is represented as:
Figure BDA00020698226700000811
s3.2, as shown in the figure 2, optimizing the characteristic chromatic aberration, wherein the specific process is as follows;
s3.2.1, setting an iteration termination condition epsilon, an iteration number c and a maximum iteration number cmNumber of clusters τ, operation coefficient δ, weight coefficient b, popularity μj(hi)cDegree xi of hesitationj(hi)cMean value of degree of hesitation
Figure BDA00020698226700000812
Cluster center vj cObjective function Jc
S3.2.2, calculating the popularity mu of each single pixel point in each class during the c-th iterationj(hi)c
Figure BDA00020698226700000813
Wherein h isiRepresenting the pixel value of the ith pixel point in the characteristic chromatic aberration H, wherein j is not equal to s; | l | · | | represents the norm calculation;
s3.2.3, calculating the clustering center v of each class in the c iterationj c
Figure BDA00020698226700000814
Wherein g represents the number of pixel points in the characteristic chromatic aberration H;
s3.2.4, calculating the hesitation degree xi of each single pixel point in each class during the c iterationj(hi)c
Figure BDA0002069822670000091
S3.2.5, calculating average value of hesitation degree of single pixel point in each class during the c iteration
Figure BDA0002069822670000092
Figure BDA0002069822670000093
S3.2.6, calculating the objective function J in the c-th iterationcA value;
Figure BDA0002069822670000094
in the present embodiment, it is conventional
Figure BDA0002069822670000095
Cannot express the characteristics of each element completely, and in order to solve the problem of information loss, a new objective function is constructed, the first part of the objective function is muj(hi)(2-ξj(hi) In which the approval degree muj(hi) And hesitation degree xij(hi) The characteristic information of each element is enriched, and the mutual influence between the same area and different damages is eliminated. In another aspect, entropy can measure uncertainty in a system, fuzzy entropy is applied in an objective function
Figure BDA0002069822670000096
The higher the uncertainty of the variable, the higher the entropy. Meanwhile, the exponential entropy enhances the correlation of the gain function of the damage area and emphasizes the characteristics.
S3.2.7, after the c-th iteration is finished, judging whether the relation is satisfied: | Jc-Jc-1If | | is less than or equal to epsilon, the step S3.2.8 is entered, otherwise the iteration is stopped, and the step S3.3 is entered;
s3.2.8, judging whether the current iteration number c reaches the set maximum iteration number cmIf c < cmIf yes, increase the current iteration number c by 1, and then return to step S3.2.2; otherwise, stopping iteration and jumping to the step S3.3;
S3.3according to the approval maximization criterion: mj=arg max(μj(hi)c) Find out the maximum praise degree Mj,MjA set of pixels representing a j-th class; then according to MjClassifying the characteristic color difference H, and after classification is finished, v isj cIs assigned to the corresponding MjObtaining the optimized characteristic color difference H*
S3.4, the optimized reconstruction characteristics are
Figure BDA0002069822670000097
Simulation of experiment
In order to better enhance the information of the test piece defect, the embodiment performs defect detection on the Whipple guard plate rear plate.
According to the algorithm process, firstly, the original data sequence of the test piece is collected, wherein N isT=543,NI=512,NJ640, the characteristic is NC3. G (-) adopts a Gaussian function, a in singular value decompositionF=27,aB=5,aH3, optimizing membership degree and clustering center algorithm, wherein iteration termination condition epsilon is 10-5Maximum number of iterations cmThe number of clusters τ is 25, the operation coefficient δ is 0.5, and the weighting coefficient b is 2, which are 100, resulting in the defect characteristics shown in fig. 3.
The background area of the test piece reflects the influence of external thermal excitation, the background area of the edge of the test piece is reflected, due to the influence of skin effect of the outline of the edge of the defect, surface impurities, external environment, light source and other factors, the reconstruction coefficient of some non-defect areas is too high, the research on the actual defect is influenced, the reconstruction coefficient becomes a large factor for analyzing the interference defect degree, and the redundant interference information belongs to redundant interference information.
Aiming at the surface defect part, a space impact experiment is simulated, an aerospace material is impacted at a super-high speed, impact pits with different sizes are distributed on the surface of a test piece in an uneven mode, the damage characterization of the impact can be irregularly diffused along with the impact position to the periphery, the distribution of defects formed on the surface of the test piece after the impact is carried out can be clearly observed in the drawing, the defects are distributed in a gathering type scattered mode, and the defects are more dense when being closer to the center and more scattered when being closer to the edge.
Aiming at the internal spalling part of the test piece, the test piece subjected to ultra-high-speed impact can obviously observe the impact pits in the gathering type scattered distribution on the surface, only the pit-shaped distribution on the surface can be observed by means of human vision, but through algorithm model analysis, the internal part of the impact position of the experimental test piece can be seen, the area damaged by the energy wave is formed, because the impact position is relatively concentrated, the energy wave is not scattered in the internal part of the test piece as the surface is, but is concentrated in the internal part of the impact position to form the gathering type defect
In this embodiment, the texture and the chromatic aberration of the feature defect after passing through the butterworth filter are shown in fig. 4, and the chromatic aberration of the feature defect after optimization is shown in fig. 5, so that the internal defect information can be observed more clearly by extracting the internal spallation profile, the defect distribution is relatively gathered, and the energy wave is relatively concentrated on the diffusion opposite surface in the material.
After the above algorithm processing is combined. The internal defect can be represented as shown in fig. 6, and from the above fig. 6, it can be clearly observed that the internal delamination information in the surface defect.
Due to the particularity of the internal spalling defect, the internal spalling defect is often ignored by people, and through the analysis of the internal defect and the research of combining the surface defect and the internal spalling, the performance of the material and the related defect mechanism can be discussed completely, the damage caused by unpredictable impact can be reasonably and effectively prevented, and the normal working performance of the equipment can be ensured.
Although illustrative embodiments of the present invention have been described above to facilitate the understanding of the present invention by those skilled in the art, it should be understood that the present invention is not limited to the scope of the embodiments, and various changes may be made apparent to those skilled in the art as long as they are within the spirit and scope of the present invention as defined and defined by the appended claims, and all matters of the invention which utilize the inventive concepts are protected.

Claims (2)

1. An exponential entropy multiplicative fuzzy defect feature analysis reconstruction method based on infrared thermal imaging is characterized by comprising the following steps:
(1) preprocessing of video stream to be detected
(1.1) representing the video stream to be detected as:
Figure FDA0003379671260000011
wherein N isI×NJRepresenting spatial information, NTRepresenting time information;
(1.2) converting the matrix block into a two-dimensional matrix through a vector operator Vec () to obtain a rewritten video stream Y;
Y=[Vec(Y(1)),Vec(Y(2)),…,Vec(Y(NT))]
wherein the content of the first and second substances,
Figure FDA0003379671260000012
NIJ=NI×NJ
(1.3) according to the prior I defect types, rewriting the video stream to be detected into:
Figure FDA0003379671260000013
wherein, Xθ(tθ) Indicating the t-th in the video stream to be detectedθThe frame is divided into a theta defect feature row vector, where theta is 1,2, …, l,
Figure FDA0003379671260000014
αθ(tθ) Is Xθ(tθ) The corresponding characteristic coefficient matrix column vector,
Figure FDA0003379671260000015
Figure FDA0003379671260000016
NCrepresenting the total frame number representing characteristic defects in the video stream to be detected;
(2) defect reconstruction
(2.1) decomposing the rewritten video stream Y into the following parts by adopting a singular value decomposition algorithm: y isT=U∑VTWherein, in the step (A),
Figure FDA0003379671260000017
in the form of a left-hand singular matrix,
Figure FDA0003379671260000018
is a matrix of singular values and is,
Figure FDA0003379671260000019
is a right singular matrix;
(2.2) calculating a whitening matrix W and a mixing coefficient matrix lambda;
Figure FDA00033796712600000110
(2.3) representing the reconstruction defect as X ═ W × Y in a matrix;
(2.4) eliminating the correlation between the reconstructed defect characteristics by a Newton iteration method:
w*=E{xg(wTx)}-E{xg(wTx)}w
wherein X is the row vector of X, W is the row vector of W, G (-) is the derivative of the contrast function G, E {. cndot. } represents the expectation;
(2.5) processing all the rows of the X according to the method in the step (2.4) to obtain a reconstructed defect characteristic information matrix: x is W is multiplied by Y,
Figure FDA0003379671260000021
(2.6) whitening matrix W*Performing time domain analysis to obtain actual distribution conditions of different defect characteristics, and comparing W*Of mid-row vectors wThe curve trends, namely regarding the curves with similar w-x curve trends as the same defect type, and regarding the curves with similar w-x curve trends as different defect types if not;
merging defect frame numbers of the same defect type to form a new defect frame number
Figure FDA0003379671260000022
Wherein m is 1,2, …% and x is less than or equal to l,
Figure FDA0003379671260000023
thus from X*To obtain
Figure FDA0003379671260000024
A reconstructed image of the m-th defect type, denoted
Figure FDA0003379671260000025
(3) Feature extraction of reconstructed image
(3.1) reconstruction of an image with a Butterworth filter
Figure FDA0003379671260000026
Performing texture and color difference segmentation;
the characteristic color difference is expressed as:
Figure FDA0003379671260000027
wherein the content of the first and second substances,
Figure FDA0003379671260000028
tito represent
Figure FDA0003379671260000029
Middle pixel point (t)i,tj) Line coordinate of (d), tjTo represent
Figure FDA00033796712600000210
Middle pixel point (t)i,tj) Column coordinates of (D)0Is a threshold value;
the feature texture is represented as:
Figure FDA00033796712600000211
(3.2) optimizing the characteristic chromatic aberration;
(3.2.1) setting an iteration termination condition epsilon, iteration times c and maximum iteration times cmNumber of clusters τ, operation coefficient δ, weight coefficient b, popularity μj(hi)cDegree xi of hesitationj(hi)cMean value of degree of hesitation
Figure FDA00033796712600000214
Cluster center vj cObjective function Jc
(3.2.2) calculating the popularity mu of each single pixel point in each class in the c iterationj(hi)c
Figure FDA00033796712600000212
Wherein h isiRepresenting the pixel value of the ith pixel point in the characteristic chromatic aberration H, wherein j is not equal to s; | l | · | | represents the norm calculation;
(3.2.3) calculating the clustering center v of each class in the c iterationj c
Figure FDA00033796712600000213
Wherein g represents the number of pixel points in the characteristic chromatic aberration H;
(3.2.4) calculating the hesitation xi of each single pixel point in each class during the c iterationj(hi)c
Figure FDA0003379671260000031
(3.2.5) calculating the average value of the hesitations of single pixel points in each class during the c-th iteration
Figure FDA0003379671260000039
Figure FDA0003379671260000032
(3.2.6) calculating an objective function J at the c-th iterationcA value;
Figure FDA0003379671260000033
(3.2.7) and after the c iteration is finished, judging whether the relation formula is met: | Jc-Jc-1If | | ≦ ε, entering step (3.2.8), otherwise, stopping iteration, and jumping to step (3.3);
(3.2.8), judging whether the current iteration number c reaches the set maximum iteration number cmIf c < cmIf so, increasing the current iteration number c by 1, and then returning to the step (3.2.2); otherwise, stopping iteration and jumping to the step (3.3);
(3.3) maximizing the criterion according to the popularity: mj=argmax(μj(hi)c) Find out the maximum praise degree Mj,MjA set of pixels representing a j-th class; then according to MjClassifying the characteristic color difference H, and after classification is finished, v isj cIs assigned to the corresponding MjObtaining the optimized characteristic color difference H*
(3.4) optimized reconstruction characteristics
Figure FDA00033796712600000310
2. The method for reconstructing exponential entropy multiplicative fuzzy defect characteristic analysis based on infrared thermal imaging as claimed in claim 1, wherein the singular value decomposition algorithm decomposes the rewritten video stream Y by:
1) constructing a covariance matrix A;
Figure FDA0003379671260000034
2) calculating a singular value;
by calculation of
Figure FDA0003379671260000035
Get the right singular value
Figure FDA00033796712600000311
Singular value
Figure FDA0003379671260000036
And left singular value
Figure FDA0003379671260000037
Wherein the content of the first and second substances,
Figure FDA0003379671260000038
representing a characteristic value;
3) according to singular value
Figure FDA00033796712600000312
Constructing a singular value matrix sigma';
4) setting an empirical threshold aF、aB、aNAnd are all positive integers;
in the matrix of singular values ∑, the singular values are combined
Figure FDA0003379671260000041
Satisfy the requirement of
Figure FDA0003379671260000042
In (1)
Figure FDA0003379671260000043
An
Figure FDA0003379671260000044
Representing surface defect features by satisfying singular values
Figure FDA0003379671260000045
Of the hour
Figure FDA0003379671260000046
An
Figure FDA0003379671260000047
Representing internal spallation defect characteristics, satisfying singular values
Figure FDA0003379671260000048
Of the hour
Figure FDA0003379671260000049
An
Figure FDA00033796712600000410
Representing non-defect region features;
according to
Figure FDA00033796712600000411
The surface defect characteristics,
Figure FDA00033796712600000412
Internal delamination defect characterization and
Figure FDA00033796712600000413
updating the singular value matrix sigma' according to the characteristics of the non-defect area to obtain the updated singular value matrix sigma, and then obtaining the updated singular value matrix sigmaThe final rewritten video stream Y is denoted YT=U∑VT
CN201910433604.8A 2019-05-23 2019-05-23 Exponential entropy multiplicative fuzzy defect characteristic analysis reconstruction method based on infrared thermal imaging Active CN110222603B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910433604.8A CN110222603B (en) 2019-05-23 2019-05-23 Exponential entropy multiplicative fuzzy defect characteristic analysis reconstruction method based on infrared thermal imaging

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910433604.8A CN110222603B (en) 2019-05-23 2019-05-23 Exponential entropy multiplicative fuzzy defect characteristic analysis reconstruction method based on infrared thermal imaging

Publications (2)

Publication Number Publication Date
CN110222603A CN110222603A (en) 2019-09-10
CN110222603B true CN110222603B (en) 2022-03-25

Family

ID=67818255

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910433604.8A Active CN110222603B (en) 2019-05-23 2019-05-23 Exponential entropy multiplicative fuzzy defect characteristic analysis reconstruction method based on infrared thermal imaging

Country Status (1)

Country Link
CN (1) CN110222603B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115781408B (en) * 2023-01-06 2023-04-14 吉德特智能科技(苏州)有限公司 Regulation and control system for inhibiting temperature rise of electric spindle of machine tool

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1620603A (en) * 2002-01-23 2005-05-25 马雷纳系统有限公司 Employing infrared thermography for defect detection and analysis
CN108288250A (en) * 2017-12-29 2018-07-17 郑州升达经贸管理学院 Infrared remote sensing image super-resolution rebuilding method based on TGV regularizations
CN109598711A (en) * 2018-11-30 2019-04-09 电子科技大学 A kind of thermal image defect extracting method based on feature mining and neural network

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7551769B2 (en) * 2003-02-18 2009-06-23 Marena Systems Corporation Data structures and algorithms for precise defect location by analyzing artifacts

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1620603A (en) * 2002-01-23 2005-05-25 马雷纳系统有限公司 Employing infrared thermography for defect detection and analysis
CN108288250A (en) * 2017-12-29 2018-07-17 郑州升达经贸管理学院 Infrared remote sensing image super-resolution rebuilding method based on TGV regularizations
CN109598711A (en) * 2018-11-30 2019-04-09 电子科技大学 A kind of thermal image defect extracting method based on feature mining and neural network

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
Sparse Reconstruction-Based Thermal Imaging for Defect Detection;Deboshree Roy 等;《IEEE Transactions on Instrumentation and Measurement》;20190114;第4550-4558页 *
基于多模态特征图融合的红外热图像目标区域提取算法;朱莉 等;《红外与毫米波学报》;20190228;第125-131页 *

Also Published As

Publication number Publication date
CN110222603A (en) 2019-09-10

Similar Documents

Publication Publication Date Title
CN107657279B (en) Remote sensing target detection method based on small amount of samples
CN110211103B (en) Information entropy additive fuzzy defect characteristic analysis reconstruction method based on infrared thermal imaging
Al-Hiary et al. Fast and accurate detection and classification of plant diseases
CN112435221A (en) Image anomaly detection method based on generative confrontation network model
CN106778687B (en) Fixation point detection method based on local evaluation and global optimization
CN111046964B (en) Convolutional neural network-based human and vehicle infrared thermal image identification method
CN112101271A (en) Hyperspectral remote sensing image classification method and device
CN110619352A (en) Typical infrared target classification method based on deep convolutional neural network
CN105760877A (en) Wool and cashmere identification algorithm based on gray level co-occurrence matrix model
CN113837974A (en) NSST (non-subsampled contourlet transform) domain power equipment infrared image enhancement method based on improved BEEPS (Bayesian particle swarm optimization) filtering algorithm
CN110222603B (en) Exponential entropy multiplicative fuzzy defect characteristic analysis reconstruction method based on infrared thermal imaging
CN113537236B (en) Quantitative identification method for defect of thermal diffusion effect for infrared detection of spacecraft damage
Takruri et al. Multi-classifier decision fusion for enhancing melanoma recognition accuracy
Clarke et al. Does free-sorting provide a good estimate of visual similarity
CN110222740B (en) Infrared technology defect reconstruction and feature extraction method based on additive fuzzy
CN110390663B (en) Information entropy multiplicative fuzzy defect characteristic analysis reconstruction method based on infrared thermal imaging
CN110389154B (en) Index entropy additive fuzzy defect characteristic analysis reconstruction method based on infrared thermal imaging
Rehman et al. Fruit leaf diseases classification: A hierarchical deep learning framework
Zhang et al. Design of tire damage image recognition system based on deep learning
Miao et al. Research on Soybean Disease Identification Method Based on Deep Learning
Xie et al. Improved locally linear embedding and its application on multi-pose ear recognition
CN115272761A (en) Power grid transmission line defect detection method based on collaborative representation bilateral self-encoder
Feng et al. Research of image recognition of plant diseases and pests based on deep learning
Indukuri et al. Paddy Disease Classifier using Deep learning Techniques
CN110223274B (en) Infrared technology defect reconstruction and feature extraction method based on multiplicative fuzzy

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