CN110175976A - II plate-type ballastless track Crack Detection classification method of CRTS based on machine vision - Google Patents
II plate-type ballastless track Crack Detection classification method of CRTS based on machine vision Download PDFInfo
- Publication number
- CN110175976A CN110175976A CN201910065861.0A CN201910065861A CN110175976A CN 110175976 A CN110175976 A CN 110175976A CN 201910065861 A CN201910065861 A CN 201910065861A CN 110175976 A CN110175976 A CN 110175976A
- Authority
- CN
- China
- Prior art keywords
- crack
- image
- crts
- ballastless track
- plate
- 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
- 238000000034 method Methods 0.000 title claims abstract description 85
- 238000001514 detection method Methods 0.000 title claims abstract description 47
- 238000010801 machine learning Methods 0.000 claims abstract description 13
- 239000013598 vector Substances 0.000 claims description 21
- 230000006870 function Effects 0.000 claims description 16
- 238000012549 training Methods 0.000 claims description 12
- 238000003708 edge detection Methods 0.000 claims description 11
- 238000012706 support-vector machine Methods 0.000 claims description 10
- 230000006378 damage Effects 0.000 claims description 7
- 230000009466 transformation Effects 0.000 claims description 7
- 238000000605 extraction Methods 0.000 claims description 6
- 238000005286 illumination Methods 0.000 claims description 6
- 208000027418 Wounds and injury Diseases 0.000 claims description 5
- 230000003044 adaptive effect Effects 0.000 claims description 5
- 208000014674 injury Diseases 0.000 claims description 5
- 230000008569 process Effects 0.000 claims description 5
- 230000002146 bilateral effect Effects 0.000 claims description 4
- 238000009795 derivation Methods 0.000 claims description 4
- 238000011156 evaluation Methods 0.000 claims description 4
- 238000001914 filtration Methods 0.000 claims description 4
- 230000003902 lesion Effects 0.000 claims description 4
- 238000005457 optimization Methods 0.000 claims description 4
- 238000005070 sampling Methods 0.000 claims description 4
- 239000000284 extract Substances 0.000 claims description 2
- 201000010099 disease Diseases 0.000 abstract description 4
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 abstract description 4
- 230000011218 segmentation Effects 0.000 description 12
- 238000003709 image segmentation Methods 0.000 description 6
- 238000007689 inspection Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 4
- 239000006247 magnetic powder Substances 0.000 description 4
- 238000000638 solvent extraction Methods 0.000 description 4
- 230000008859 change Effects 0.000 description 3
- 238000012886 linear function Methods 0.000 description 3
- 238000012423 maintenance Methods 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 238000005336 cracking Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000001788 irregular Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 231100000241 scar Toxicity 0.000 description 2
- 241000276425 Xiphophorus maculatus Species 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 239000003245 coal Substances 0.000 description 1
- 239000000571 coke Substances 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000007598 dipping method Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 239000012634 fragment Substances 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 239000011229 interlayer Substances 0.000 description 1
- 230000005415 magnetization Effects 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000001737 promoting effect Effects 0.000 description 1
- 239000007921 spray Substances 0.000 description 1
- 239000004575 stone Substances 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000012800 visualization Methods 0.000 description 1
Classifications
-
- 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
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- 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
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/187—Segmentation; Edge detection involving region growing; involving region merging; involving connected component labelling
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Data Mining & Analysis (AREA)
- Quality & Reliability (AREA)
- Pathology (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- General Health & Medical Sciences (AREA)
- Immunology (AREA)
- Health & Medical Sciences (AREA)
- Signal Processing (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The CRTS II plate-type ballastless track Crack Detection classification method based on machine vision that the present invention provides a kind of.This method will improve the accuracy of crack identification, effectively reduce false detection rate.Since the shape in crack determines its degree of disease, cause the repairing method in crack different.Therefore this method solves the problems, such as that fracture shape is classified by machine learning algorithm.
Description
Technical field
The CRTS II plate-type ballastless track Crack Detection classification method based on machine vision that the present invention relates to a kind of.
Background technique
In recent years, as China railways develop rapidly, ballastless track structure is the operation for guaranteeing the steady safety of bullet train
Important leverage is provided, CRTS II platy ballastless track structure is a kind of ballastless track structure form being widely adopted.Without the tiny fragments of stone, coal, etc.
Each structure sheaf of track is usually made of concrete or armored concrete, since concrete tensile strength is low, in addition for a long time by train
The influence of load action and harsh natural environment, cracking are inevitable.Although being provided with a fixed spacing in track plates top surface
Precracking has been also found that crack at non-precracking to prevent its cracking, but in factual survey.Crack or structure interlayer off-seam meeting
Make arrangement works reduced performance, depression of bearing force even jeopardizes traffic safety.
Currently, CRTS II plate-type ballastless track maintenance work relies primarily on manual inspection, it is complete by range estimation and hand dipping
Based on.This routine inspection mode inefficiency, and be illuminated by the light, subjective factor influence it is big.And the detection means such as ultrasonic wave, current vortex,
It is not suitable for surface inspection, but is suitable for track interior and detects a flaw.Magnetic powder inspection is because needing to spray magnetic powder in raceway surface, then
According to the magnetic resistance change rate at scar after track magnetization, magnetic powder shows scar, and detection efficiency is low in this way for institute, needs a large amount of magnetic
Powder is not suitable for needing the track flaw detection of quickly detection, long range for detecting.
Summary of the invention
The CRTS II plate-type ballastless track Crack Detection classification based on machine vision that the purpose of the present invention is to provide a kind of
Method.
To solve the above problems, the present invention provides a kind of II plate-type ballastless track crack CRTS inspection based on machine vision
Survey classification method, comprising:
Step1: the original image in the crack of acquisition CRTS II plate-type ballastless track;
Step2: pretreatment operation is carried out to the original image in the crack, obtains gray level image;
Step3: binarization operation is carried out to the gray level image and obtains bianry image;
Step4: being split the bianry image, to be partitioned into crack image;
Step5: edge detection is carried out to the crack image, obtains the contour images in crack;
Step6: feature extraction is carried out to the contour images, and calculates the width in crack;
Step7: classified by shape of the support vector machines machine learning algorithm to the crack;
Step8: security level evaluation is carried out to the crack classified.
Further, in the above-mentioned methods, in the Step1, when acquiring image, the fixed position and coke of camera are kept
Away from identical light source when night acquires.
Further, in the above-mentioned methods, it in the Step2, is drawn high using contrast to carry out image enhancement, is made first
The original image brightness for obtaining the entirely crack keeps uniform as far as possible.Then bilateral filtering is used, filters out in image and significantly makes an uproar
Image is finally carried out greyscale transformation, obtains gray level image by point.
Further, in the above-mentioned methods, in the Step3, using the binarization operation based on illumination adaptive threshold value.
Further, in the above-mentioned methods, in the Step4, the bianry image is divided using region-growing method
It cuts, to be partitioned into crack image.
Further, in the above-mentioned methods, in the Step5, fracture image carries out Canny edge detection, is split
The contour images of seam.
Further, in the above-mentioned methods, in the Step6, using minimum circumscribed rectangle as feature to the crack
Contour images carry out feature extraction, and calculate the width in crack.
Further, in the above-mentioned methods, in the Step7, characterized by rectangle long axis and horizontal angle, pass through branch
It holds vector machine machine learning algorithm to learn it, obtains sample database.Different types of crack is compared with sample database,
If long axis and horizontal angle then judge this crack for horizontal shape crack, if long axis between (- 45 °, 45 °) and (135 °, 225 °)
With horizontal angle between (45 °, 135 °) and (225 °, 315 °), then judge this crack for vertical-shaped crack.
Further, in the above-mentioned methods, in the Step7, the process of support vector machines study and classification is as follows:
(1) fracture profile figure is divided into the subimage block of W M × N, each subgraph constitutes a feature vector xi=p,
Wherein i=1,2,3 ..., W;P is crack most boundary rectangle long axis and horizontal angle;
(2) by manual method in xiMiddle selection a part can represent target area and the feature vector of nontarget area is come
As trained feature vector, it is expressed as (xj, yj), wherein j ∈ { 1,2 ..., W }, yjIt is class formative;
(3) A, B are set and respectively represents horizontal shape crack target area and vertical-shaped crack nontarget area.Then yjIt can indicate are as follows:
All correctly classify to all training characteristics samples, linear discriminant function must satisfy:
yj(wTxj+b)-1≥0 ②
Wherein, w is weight vectors, and b is constant, the gap size in the classification gap (M) of two class samples are as follows:
M=2/ | | w | |2 ③
At this point, optimal classification surface problem is converted into following optimization problem, i.e., under condition constraint 3., the derivation of equation is 4. most
Small value:
The minimum value of formula 4. is acquired by the constraint condition of formula 2., obtains globally optimal solution w*、b*, then linear optimal
Classification decision function are as follows:
F (x)=sgn (w*x+b*) ⑤
Wherein, sgn is sign function, and x is sampling feature vectors.It is non-with one for the training sample of Nonlinear separability
The training sample space of points is mapped to higher-dimension sample space by linear function, and linear classification is carried out on the sample space of higher-dimension;
(4) by set of eigenvectors x to be sortedi(i=1,2,3 ..., W) is brought into 5., if f (xi) value be 1,
So corresponding xiBelong to A class, otherwise xiBelong to B class, it can thus be appreciated that xiRepresentative profile belongs to horizontal shape crack or vertical-shaped splits
Seam.
Further, in the above-mentioned methods, in the Step8, fracture length is less than 5mm, is assessed as slight crack,
With the wire tag of green, with text prompt " this crack minor injury, please keep continue tracking ";Fracture length is greater than 5mm, is less than
10mm is assessed as early warning crack, is demarcated with the line of yellow, with text prompt, " this crack moderate lesion please send someone to be repaired
It is multiple ";Fracture length is greater than 10mm, is assessed as high-risk crack, with red wire tag, with text prompt " this fracture height
Damage please be replaced immediately ".
Compared with prior art, the CRTS II plate-type ballastless track Crack Detection based on machine vision that set forth herein a kind of
Classification method, proposes the image processing techniques for track Crack Detection, and using the classification method based on machine learning into
The classification of row track plank split is finally gone back fracture security level and is evaluated, to improve the track plates in railway traffic transport
Detection makes significant contribution with maintenance work.The present invention provides a kind of CRTS II plate-type ballastless track based on machine vision and splits
Seam detection classification method.This method will improve the accuracy of crack identification, effectively reduce false detection rate.Since the shape in crack is determined
Its fixed degree of disease causes the repairing method in crack different.Therefore this method solves fracture shape point by machine learning algorithm
The problem of class.
Detailed description of the invention
Fig. 1 is the CRTS II plate-type ballastless track Crack Detection classification side based on machine vision of one embodiment of the invention
The flow chart of method.
Specific embodiment
In order to make the foregoing objectives, features and advantages of the present invention clearer and more comprehensible, with reference to the accompanying drawing and specific real
Applying mode, the present invention is described in further detail.
As shown in Figure 1, the present invention provides a kind of CRTS II plate-type ballastless track Crack Detection classification based on machine vision
Method, comprising:
Step1: acquisition CRTS II plate-type ballastless track crack original image.
Step2: pretreatment operation, including image enhancement, image gray-scale transformation are carried out to original image.
Step3: binarization operation is carried out to grayscale image.
Step4: being split bianry image using region-growing method, divisible crack image out.
Step5: Canny edge detection is carried out to the crack image after segmentation, obtains the contour images in crack.
Step6: feature extraction is carried out to contour images using minimum circumscribed rectangle as feature, and calculates the width in crack.
Step7: classified by support vector machines machine learning algorithm fracture shape.
Step8: security level evaluation is carried out to the crack classified.
Here, track plates are very important always component part in the railway traffic transport process of construction in China.
The one kind of crack as track plates disease affects its safe operation, therefore most important to the effectively detection of track plank split.
Although the demand for detecting rail safety especially to track plates Crack Detection has been increased to a new height, due to
Traditional artificial detection method job costs are high, and detection efficiency is low, and testing result accuracy and confidence level are lower.With existing
The continuous improvement of rapid development and computer hardware performance for computer technology makes the most rapid machine vision technique of development
It is applicable in track plates Crack Detection and is also possibly realized, so, the method for proposing new visualization Crack Detection and classification
It is imperative to the health evaluating of track plates to complete, and there is very big significance.
The CRTS II plate-type ballastless track Crack Detection classification method based on machine vision that set forth herein a kind of, proposes
Dividing for track plank split is carried out for the image processing techniques of track Crack Detection, and using the classification method based on machine learning
Class is finally gone back fracture security level and is evaluated, and does to improve the track plates detection in railway traffic transport with maintenance work
Significant contribution out.The present invention provides a kind of CRTS II plate-type ballastless track Crack Detection classification method based on machine vision.It should
Method will improve the accuracy of crack identification, effectively reduce false detection rate.Since the shape in crack determines its degree of disease, cause
The repairing method in crack is different.Therefore this method solves the problems, such as that fracture shape is classified by machine learning algorithm.
In one embodiment of CRTS II plate-type ballastless track Crack Detection classification method based on machine vision of the invention,
Step1 requires the fixed position and focal length that keep camera as far as possible when acquiring image.Night must also consider when acquiring with identical
Light source.
In one embodiment of CRTS II plate-type ballastless track Crack Detection classification method based on machine vision of the invention,
The period of Step2 acquisition image is distributed in different moments, and the exposure of different time sections illumination and camera can produce image
Raw different influence, so first having to carry out pretreatment operation to image, its object is to remove noise spot or promote image
Contrast, lay the groundwork for next image segmentation.
It is drawn high using contrast to carry out image enhancement first, so that the brightness in whole image keeps uniformly, connecing as far as possible
Apparent noise in image is filtered out using bilateral filtering, image is finally subjected to greyscale transformation, obtains gray level image.
In one embodiment of CRTS II plate-type ballastless track Crack Detection classification method based on machine vision of the invention,
Since the image irradiation of different moments is different, when image binaryzation, needs according to different brightness to each picture hand Step3
The setting of the completion threshold value of work, so that the robustness of algorithm is deteriorated.Therefore it is based on illumination adaptive threshold value that the present invention, which uses,
Binarization operation so that the robustness of Binarization methods greatly improves.
In one embodiment of CRTS II plate-type ballastless track Crack Detection classification method based on machine vision of the invention,
The purpose of Step4 image segmentation is in order to which extracted region interested in image is come out, for example interested here is crack
Region is it is necessary to extract crack area by certain partitioning algorithm, and partitioning algorithm selection here is region-growing method.
Region-growing method can usually come out the connection region segmentation with same characteristic features, and can provide good boundary information and segmentation
As a result.When no priori knowledge can use, the optimum efficiency of optimal performance and complicated image segmentation can be obtained.And
The image in CRTS II plate-type ballastless track crack has complicated and irregular characteristic, so selective area growth method is come to figure
Ideal segmentation effect can be obtained as carrying out Threshold segmentation.
In one embodiment of CRTS II plate-type ballastless track Crack Detection classification method based on machine vision of the invention,
The purpose of Step5 edge detection is to find the violent pixel collection of brightness change in image, shows often profile.Such as
Edge accurately can be measured and be positioned in fruit image, then, it is meant that actual object can be positioned and be measured, including
Shape of the area of object, the diameter of object, object etc. can be measured.Method used herein is that Canny edge detection is calculated
Method, the method have the characteristics that low False Rate and can inhibit false edge.
In one embodiment of CRTS II plate-type ballastless track Crack Detection classification method based on machine vision of the invention,
Step6 chooses feature of the minimum circumscribed rectangle of profile as crack, and its purpose is to facilitate the direction and the width that calculate crack
Degree.Boundary rectangle long axis and horizontal angle can represent the direction in crack, and the length of boundary rectangle short axle then represents crack
Width.
In one embodiment of CRTS II plate-type ballastless track Crack Detection classification method based on machine vision of the invention,
Step7 learns it by support vector machines machine learning algorithm, is obtained characterized by rectangle long axis and horizontal angle
Sample database.Different types of crack is compared with sample database, if long axis and horizontal angle at (- 45 °, 45 °) and
Between (135 °, 225 °), then judge this crack for horizontal shape crack.If long axis and horizontal angle (45 °, 135 °) and (225 °,
315 °) between, then judge this crack for vertical-shaped crack.
The process of support vector machines study and classification is as follows:
(1) fracture profile figure is divided into the subimage block of W M × N, each subgraph constitutes a feature vector xi=p,
Wherein i=1,2,3 ..., W;P is crack most boundary rectangle long axis and horizontal angle.
(2) by manual method in xiMiddle selection a part can represent target area and the feature vector of nontarget area is come
As trained feature vector, it is expressed as (xj, yj), wherein j ∈ { 1,2 ..., W }, yjIt is class formative.
(3) A, B are set and respectively represents horizontal shape crack target area and vertical-shaped crack nontarget area.Then yjIt can indicate are as follows:
All correctly classify to all training characteristics samples, linear discriminant function must satisfy:
yj(wTxj+b)-1≥0 ②
Wherein, w is weight vectors, and b is constant, its effect is to cross origin in order to avoid linear classification face is certain, makes this
Method is more flexible.The gap size in the classification gap (M) of two class samples are as follows:
M=2/ | | w | |2 ③
At this point, optimal classification surface problem is converted into following optimization problem, i.e., under condition constraint 3., the derivation of equation is 4. most
Small value:
The minimum value of formula 4. is acquired by the constraint condition of formula 2., obtains globally optimal solution w*、b*, then linear optimal
Classification decision function are as follows:
F (x)=sgn (w*x+b*) ⑤
Wherein, Sgn is sign function, and x is sampling feature vectors.It is non-with one for the training sample of Nonlinear separability
The training sample space of points is mapped to higher-dimension sample space by linear function, and linear classification is carried out on the sample space of higher-dimension.
(4) by set of eigenvectors x to be sortedi(i=1,2,3 ..., W) is brought into 5., if f (xi) value be 1,
So corresponding xiBelong to A class, otherwise xiBelong to B class.It can thus be appreciated that xiRepresentative profile belongs to horizontal shape crack or vertical-shaped splits
Seam.
In one embodiment of CRTS II plate-type ballastless track Crack Detection classification method based on machine vision of the invention,
Step8 fracture length is less than 5mm, is assessed as slight crack, and with the wire tag of green, with text prompt, " this crack is slight
Damage please keep continuing tracking ".Fracture length is greater than 5mm, is less than 10mm, is demarcated with the line of yellow, with text prompt, " this splits
Moderate lesion is stitched, please send someone to be repaired ".Fracture length is greater than 10mm, with red wire tag, with text prompt " this crack
High injury please immediately replace ".
Specifically, during Image Acquisition of the present invention, it is desirable that the fixation position of holding camera and focal length as far as possible.Night adopts
Collection image must use identical light source.Then collected original image is pre-processed, greyscale transformation and adaptive threshold
The operation of binaryzation, reuse region growing method carry out Threshold segmentation, to after segmentation image carry out Canny edge detection and
Feature extraction is classified finally by machine learning algorithm fracture type.
A kind of CRTS II plate-type ballastless track Crack Detection classification method program circuit packet based on machine vision of the present invention
Following 6 steps are included, flow chart is as shown in Figure 1.Specific step is as follows.
Step1: setting up multiple cameras on track plates side, every the image of acquisition in 6 hours.During setting up camera
The fixed position and focal length of camera are kept as far as possible.Night must also consider when acquiring with identical light source.
Step2: pretreatment operation, including image enhancement, image gray-scale transformation are carried out to image.The period of Image Acquisition
It is distributed in different moments, the exposure of different time sections illumination and camera can generate image different influences, so first
Pretreatment operation is carried out to image, it is therefore intended that removal noise spot or the contrast for promoting image are next image
Segmentation is laid the groundwork.
It is drawn high using contrast to carry out image enhancement first, so that the brightness in whole image keeps uniformly, connecing as far as possible
Use bilateral filtering, filter out apparent noise in image, image be finally subjected to greyscale transformation, obtains gray level image.
Step3: grayscale image is subjected to binarization operation.Since the image irradiation of different moments is different, when image binaryzation
The setting according to different brightness to the completion threshold value of each picture craft is needed, so that the robustness of algorithm is deteriorated.Therefore
Of the present invention is the binarization operation based on illumination adaptive threshold value, so that the robustness of Binarization methods mentions significantly
It is high.
Step4: being split bianry image using region-growing method, divisible crack image out.The mesh of image segmentation
Be in order to which extracted region interested in image is come out.For example interested here is crack area it is necessary to pass through certain
Partitioning algorithm extracts crack area, and partitioning algorithm selection here is region-growing method.The usual energy of region-growing method
Connection region segmentation with same characteristic features is come out, and good boundary information and segmentation result can be provided.In no priori
When knowledge can use, the optimum efficiency of optimal performance and complicated image segmentation can be obtained.And CRTS II plate-type ballastless rail
The image in road crack has complicated and irregular characteristic, so selective area growth method to can use image progress Threshold segmentation
Obtain ideal segmentation effect.
Step5: Canny edge detection is carried out to the crack image after segmentation, obtains the contour images in crack.Edge detection
Purpose be to find the violent pixel collection of brightness change in image, show often profile.If edge in image
Accurately it can measure and position, then, it is meant that actual object can be positioned and be measured, area including object,
Diameter, shape of object of object etc. can be measured.Method used herein is Canny edge detection algorithm, the method tool
The features such as having low False Rate and false edge can be inhibited.
Step6: feature extraction is carried out to contour images using minimum circumscribed rectangle as feature, and calculates the width in crack.
Choose feature of the minimum circumscribed rectangle of profile as crack, in order to facilitate the direction for calculating crack and width.It is external
Rectangle long axis and horizontal angle can represent the direction in crack, and the length of boundary rectangle short axle then represents the width in crack.
Step7: characterized by rectangle long axis and horizontal angle, it is carried out by support vector machines machine learning algorithm
Study, obtains sample database.Different types of crack is compared with sample database, if long axis and horizontal angle (- 45 °,
45 °) and (135 °, 225 °) between, then judge this crack for horizontal shape crack.If long axis and horizontal angle at (45 °, 135 °) and
Between (225 °, 315 °), then judge this crack for vertical-shaped crack.
The process of support vector machines study and classification is as follows:
(1) fracture profile figure is divided into the subimage block of W M × N, each subgraph constitutes a feature vector xi=p,
Wherein i=1,2,3 ..., W;P is crack most boundary rectangle long axis and horizontal angle.
(2) by manual method in xiMiddle selection a part can represent target area and the feature vector of nontarget area is come
As trained feature vector, it is expressed as (xj, yj), wherein j ∈ { 1,2 ..., W }, yjIt is class formative.
(3) A, B are set and respectively represents horizontal shape crack target area and vertical-shaped crack nontarget area.Then yjIt can indicate are as follows:
All correctly classify to all training characteristics samples, linear discriminant function must satisfy:
yj(wTxj+b)-1≥0 ②
Wherein, w is weight vectors, and b is constant, its effect is to cross origin in order to avoid linear classification face is certain, makes this
Method is more flexible.The gap size in the classification gap (M) of two class samples are as follows:
M=2/ | | w | |2 ③
At this point, optimal classification surface problem is converted into following optimization problem, i.e., under condition constraint 3., the derivation of equation is 4. most
Small value:
The minimum value of formula 4. is acquired by the constraint condition of formula 2., obtains globally optimal solution w*, b*, then linear optimal
Classification decision function are as follows:
F (x)=sgn (w*x+b*) ⑤
Wherein, sgn is sign function, and x is sampling feature vectors.It is non-with one for the training sample of Nonlinear separability
The training sample space of points is mapped to higher-dimension sample space by linear function, and linear classification is carried out on the sample space of higher-dimension.
(4) by set of eigenvectors x to be sortedi(i=1,2,3 ..., W) is brought into 5., if f (xi) value be 1,
So corresponding xiBelong to A class, otherwise xiBelong to B class.It can thus be appreciated that xiRepresentative profile belongs to horizontal shape crack or vertical-shaped splits
Seam.
Step8: security level evaluation is carried out to the crack classified.Fracture length is less than 5mm, is assessed as slightly splitting
Seam, with the wire tag of green, with text prompt " this crack minor injury, please keep continue tracking ".Fracture length is greater than 5mm,
Less than 10mm, demarcated with the line of yellow, be assessed as early warning crack, with text prompt " this crack moderate lesion, please send someone into
Row is repaired ".Fracture length is greater than 10mm, with red wire tag, high-risk crack is assessed as, with text prompt " this crack
High injury please immediately replace ".
Each embodiment in this specification is described in a progressive manner, the highlights of each of the examples are with other
The difference of embodiment, the same or similar parts in each embodiment may refer to each other.
Professional further appreciates that, unit described in conjunction with the examples disclosed in the embodiments of the present disclosure
And algorithm steps, can be realized with electronic hardware, computer software, or a combination of the two, in order to clearly demonstrate hardware and
The interchangeability of software generally describes each exemplary composition and step according to function in the above description.These
Function is implemented in hardware or software actually, the specific application and design constraint depending on technical solution.Profession
Technical staff can use different methods to achieve the described function each specific application, but this realization is not answered
Think beyond the scope of this invention.
Obviously, those skilled in the art can carry out various modification and variations without departing from spirit of the invention to invention
And range.If in this way, these modifications and changes of the present invention belong to the claims in the present invention and its equivalent technologies range it
Interior, then the invention is also intended to include including these modification and variations.
Claims (10)
1. a kind of CRTS II plate-type ballastless track Crack Detection classification method based on machine vision characterized by comprising
Step1: the original image in the crack of acquisition CRTS II plate-type ballastless track;
Step2: pretreatment operation is carried out to the original image in the crack, obtains gray level image;
Step3: binarization operation is carried out to the gray level image and obtains bianry image;
Step4: being split the bianry image, to be partitioned into crack image;
Step5: edge detection is carried out to the crack image, obtains the contour images in crack;
Step6: feature extraction is carried out to the contour images, and calculates the width in crack;
Step7: classified by shape of the support vector machines machine learning algorithm to the crack;
Step8: security level evaluation is carried out to the crack classified.
2. a kind of CRTS II plate-type ballastless track Crack Detection classification side based on machine vision according to claim 1
Method, which is characterized in that in the Step1, when acquiring image, keep the fixed position and focal length of camera, phase when night acquires
Same light source.
3. a kind of CRTS II plate-type ballastless track Crack Detection classification side based on machine vision according to claim 1
Method, which is characterized in that in the Step2, drawn high using contrast to carry out image enhancement first, so that the entire crack
Original image brightness is kept uniformly as far as possible, is then used bilateral filtering, is filtered out apparent noise in image, finally carry out image
Greyscale transformation obtains gray level image.
4. a kind of CRTS II plate-type ballastless track Crack Detection classification side based on machine vision according to claim 1
Method, which is characterized in that in the Step3, using the binarization operation based on illumination adaptive threshold value.
5. a kind of CRTS II plate-type ballastless track Crack Detection classification side based on machine vision according to claim 1
Method, which is characterized in that in the Step4, the bianry image is split using region-growing method, to be partitioned into crack pattern
Picture.
6. a kind of CRTS II plate-type ballastless track Crack Detection classification side based on machine vision according to claim 1
Method, which is characterized in that in the Step5, fracture image carries out Canny edge detection, obtains the contour images in crack.
7. a kind of CRTS II plate-type ballastless track Crack Detection classification side based on machine vision according to claim 1
Method, which is characterized in that in the Step6, feature is carried out to the contour images in the crack using minimum circumscribed rectangle as feature
It extracts, and calculates the width in crack.
8. a kind of CRTS II plate-type ballastless track Crack Detection classification side based on machine vision according to claim 1
Method, which is characterized in that in the Step7, characterized by rectangle long axis and horizontal angle, pass through support vector machines machine learning
Algorithm learns it, obtains sample database.Different types of crack is compared with sample database, if long axis and horizontal folder
Angle then judges this crack for horizontal shape crack, if long axis and horizontal angle exist between (- 45 °, 45 °) and (135 °, 225 °)
Between (45 °, 135 °) and (225 °, 315 °), then judge this crack for vertical-shaped crack.
9. a kind of CRTS II plate-type ballastless track Crack Detection classification side based on machine vision according to claim 1
Method, which is characterized in that in the Step7, the process of support vector machines study and classification is as follows:
(1) fracture profile figure is divided into the subimage block of W M × N, each subgraph constitutes a feature vector xi=p, wherein i
=1,2,3 ..., W;P is crack most boundary rectangle long axis and horizontal angle;
(2) by manual method in xiMiddle selection a part can represent the feature vector of target area and nontarget area as
Trained feature vector, is expressed as (xj, yj), wherein j ∈ { 1,2 ..., W }, yjIt is class formative;
(3) A, B are set and respectively represents horizontal shape crack target area and vertical-shaped crack nontarget area.Then yjIt can indicate are as follows:
All training characteristics samples are all correctly classified, and linear discriminant function meets:
yj(wTxj+b)-1≥0 ②
Wherein, w is weight vectors, and b is constant, the gap size in the classification gap (M) of two class samples are as follows:
M=2/ | | w | |2 ③
At this point, optimal classification surface problem is converted into following optimization problem, i.e., under condition constraint 3., the minimum of the derivation of equation 4.
Value:
The minimum value of formula 4. is acquired by the constraint condition of formula 2., obtains globally optimal solution w*、b*, then linear optimal is classified
Decision function are as follows:
F (x)=sgn (w*x+b*) ⑤
Wherein, sgn is sign function, and x is sampling feature vectors.It is non-linear with one for the training sample of Nonlinear separability
The training sample space of points is mapped to higher-dimension sample space by function, and linear classification is carried out on the sample space of higher-dimension;
(4) by set of eigenvectors x to be sortedi(i=1,2,3 ..., W) is brought into 5., if f (xi) value be 1, then
Corresponding xiBelong to A class, otherwise xiBelong to B class, it can thus be appreciated that xiRepresentative profile belongs to horizontal shape crack or vertical-shaped crack.
10. a kind of CRTS II plate-type ballastless track Crack Detection classification side based on machine vision according to claim 1
Method, which is characterized in that in the Step8, fracture length be less than 5mm, be assessed as slight crack, with green wire tag,
With text prompt " this crack minor injury please keep continuing tracking ";Fracture length is greater than 5mm, is less than 10mm, is assessed as
Early warning crack, with the wire tag of yellow, with text prompt " this crack moderate lesion please send someone to be repaired ";Fracture length is big
In 10mm, it is assessed as high-risk crack, with red wire tag, " this fracture height is damaged, please immediately more with text prompt
It changes ".
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910065861.0A CN110175976A (en) | 2019-01-23 | 2019-01-23 | II plate-type ballastless track Crack Detection classification method of CRTS based on machine vision |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910065861.0A CN110175976A (en) | 2019-01-23 | 2019-01-23 | II plate-type ballastless track Crack Detection classification method of CRTS based on machine vision |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110175976A true CN110175976A (en) | 2019-08-27 |
Family
ID=67689105
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910065861.0A Pending CN110175976A (en) | 2019-01-23 | 2019-01-23 | II plate-type ballastless track Crack Detection classification method of CRTS based on machine vision |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110175976A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110717455A (en) * | 2019-10-10 | 2020-01-21 | 北京同创信通科技有限公司 | Method for classifying and detecting grades of scrap steel in storage |
CN111079747A (en) * | 2019-12-12 | 2020-04-28 | 哈尔滨市科佳通用机电股份有限公司 | Railway wagon bogie side frame fracture fault image identification method |
CN112862764A (en) * | 2021-01-26 | 2021-05-28 | 中国铁道科学研究院集团有限公司 | Method and device for identifying ballastless track bed gap damage and storage medium |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101915764A (en) * | 2010-08-10 | 2010-12-15 | 武汉武大卓越科技有限责任公司 | Road surface crack detection method based on dynamic programming |
CN103323526A (en) * | 2013-05-30 | 2013-09-25 | 哈尔滨工业大学 | Welding line defect detection and identification method based on ultrasonic phased array and support vector machine |
CN103486971A (en) * | 2013-08-14 | 2014-01-01 | 北京交通大学 | Subway tunnel crack width detecting and correcting algorithm |
CN106504246A (en) * | 2016-11-08 | 2017-03-15 | 太原科技大学 | The image processing method of tunnel slot detection |
CN106557784A (en) * | 2016-11-23 | 2017-04-05 | 上海航天控制技术研究所 | Fast target recognition methodss and system based on compressed sensing |
CN108226180A (en) * | 2018-01-11 | 2018-06-29 | 上海应用技术大学 | A kind of crack detection method for CRTS II plate-type tracks |
-
2019
- 2019-01-23 CN CN201910065861.0A patent/CN110175976A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101915764A (en) * | 2010-08-10 | 2010-12-15 | 武汉武大卓越科技有限责任公司 | Road surface crack detection method based on dynamic programming |
CN103323526A (en) * | 2013-05-30 | 2013-09-25 | 哈尔滨工业大学 | Welding line defect detection and identification method based on ultrasonic phased array and support vector machine |
CN103486971A (en) * | 2013-08-14 | 2014-01-01 | 北京交通大学 | Subway tunnel crack width detecting and correcting algorithm |
CN106504246A (en) * | 2016-11-08 | 2017-03-15 | 太原科技大学 | The image processing method of tunnel slot detection |
CN106557784A (en) * | 2016-11-23 | 2017-04-05 | 上海航天控制技术研究所 | Fast target recognition methodss and system based on compressed sensing |
CN108226180A (en) * | 2018-01-11 | 2018-06-29 | 上海应用技术大学 | A kind of crack detection method for CRTS II plate-type tracks |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110717455A (en) * | 2019-10-10 | 2020-01-21 | 北京同创信通科技有限公司 | Method for classifying and detecting grades of scrap steel in storage |
CN111079747A (en) * | 2019-12-12 | 2020-04-28 | 哈尔滨市科佳通用机电股份有限公司 | Railway wagon bogie side frame fracture fault image identification method |
CN111079747B (en) * | 2019-12-12 | 2020-10-09 | 哈尔滨市科佳通用机电股份有限公司 | Railway wagon bogie side frame fracture fault image identification method |
CN112862764A (en) * | 2021-01-26 | 2021-05-28 | 中国铁道科学研究院集团有限公司 | Method and device for identifying ballastless track bed gap damage and storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110175976A (en) | II plate-type ballastless track Crack Detection classification method of CRTS based on machine vision | |
CN109580657A (en) | A kind of crack detection method in bridge quality testing | |
CN104021574B (en) | Pavement disease automatic identifying method | |
CN110211101A (en) | A kind of rail surface defect rapid detection system and method | |
CN111476767B (en) | High-speed rail fastener defect identification method based on heterogeneous image fusion | |
CN107527067A (en) | A kind of Railway Roadbed intelligent identification Method based on GPR | |
CN102737252B (en) | Method for detecting faults caused by foreign body pollution between electrified railway insulator plates based on affine invariant moment | |
CN116757990A (en) | Railway fastener defect online detection and identification method based on machine vision | |
CN111260629A (en) | Pantograph structure abnormity detection algorithm based on image processing | |
CN110567680B (en) | Track fastener looseness detection method based on angle comparison | |
CN113112501B (en) | Vehicle-mounted track inspection device and method based on deep learning | |
CN111311567A (en) | Method for identifying fastener and steel rail diseases of track line image | |
CN104318582A (en) | Detection method for bad state of rotating double-lug component pin of high-speed rail contact network on basis of image invariance target positioning | |
CN104992429A (en) | Mountain crack detection method based on image local reinforcement | |
CN107798293A (en) | A kind of crack on road detection means | |
CN111402214A (en) | Neural network-based automatic detection method for breakage defect of catenary dropper current-carrying ring | |
CN104537651A (en) | Proportion detecting algorithm and system for cracks in road surface image | |
CN112419289B (en) | Intelligent detection method for urban subway rail fastener defects | |
CN108508023B (en) | Defect detection system for contact end jacking bolt in railway contact network | |
Resendiz et al. | Development of a machine-vision system for inspection of railway track components | |
CN104282018A (en) | Method for on-line detection of overall diameter of industrial product based on machine vision | |
CN109785301A (en) | A kind of rail corrugation automatic identifying method and appraisal procedure based on image procossing | |
CN108596909A (en) | Hollow cylinder Surface Flaw Detection system and method | |
CN117173661B (en) | Asphalt road quality detection method based on computer vision | |
CN113673614B (en) | Metro tunnel foreign matter intrusion detection device and method based on machine vision |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
WD01 | Invention patent application deemed withdrawn after publication | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190827 |