CN106355185B - A kind of method in rapidly extracting Rail Surface region under vibration condition - Google Patents
A kind of method in rapidly extracting Rail Surface region under vibration condition Download PDFInfo
- Publication number
- CN106355185B CN106355185B CN201610770769.0A CN201610770769A CN106355185B CN 106355185 B CN106355185 B CN 106355185B CN 201610770769 A CN201610770769 A CN 201610770769A CN 106355185 B CN106355185 B CN 106355185B
- Authority
- CN
- China
- Prior art keywords
- image
- rail
- straight line
- rail surface
- boundary
- 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.)
- Expired - Fee Related
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
- G06V10/462—Salient features, e.g. scale invariant feature transforms [SIFT]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a kind of methods in rapidly extracting Rail Surface region under vibration condition, comprising the following steps: S1 Image Acquisition and pretreatment；S2 carrying out image threshold segmentation；The connected component in S3 extraction Rail Surface region: by the area and perimeter threshold of 8 adjacent connected components of bianry image after setting Threshold segmentation, the connected component comprising Rail Surface region is extracted；S4, it extracts Rail Surface boundary straight line: straight-line segment present in rail level edge image being extracted using Hough transform, and deletes the biggish line segment of deflection angle；S5 extracts Rail Surface boundary reference point；The fitting amendment of S6 rail level boundary straight line；S7 carries out rotation affine transformation to image according to revised deflection angle.By comparing with existing several typical rail level extraction algorithms, the speed, precision and stability of this method all have clear superiority, are more suitable in high-speed automated rail level Method of Defect Segmentation.
Description
Technical field
The invention belongs to technical field of image processing, rapidly extracting Rail Surface region under a kind of vibration condition is referred specifically to
Method.
Background technique
With train high speed, heavy duty, highdensity operation, train can generate different degrees of hurt to rail.Work as hurt
More than train can be made to generate high vibration after certain limit, the heavy connection for not only resulting in train loosens, and will affect
Sensor life-time, can more likely make train wheel rough or Wheel deflection forms vicious circle in turn.Therefore, in time
Understand rail operation conditions and carry out corresponding maintenance and repair, for ensuring that safety, continuous railway operation are very necessary.
Currently, domestic steel rail defect detection method mainly has manual patrol method, electric vortex method, magnetic particle method, visual detection method.
Wherein, manual patrol method by external environment and employee's subjective factor because being affected, and detection efficiency is low, and detection accuracy is not high；Electricity
It is slow that eddy-current method detects speed, and personnel is needed to follow up in real time, and the degree of automation of detection is lower；Although magnetic particle method detects sensitivity
Height, but defect show it is not intuitive enough, and detect speed it is slower；And visual detection method can pass through image processing techniques and pattern-recognition
Algorithm is detected to rail surface defects region and is identified classification, has high speed, high-precision, the characteristic of high automation.Cause
This, defect automatic measurement technique of the research based on rail image has important practical significance.
During visual detection method, a large amount of unrelated regions are usually contained in the rail image using industrial camera acquisition
Domain, such as: tiny fragments of stone, coal, etc. stone, leaf, rail bottom, sleeper, not only amount of redundant data is big, and Accurate Segmentation for rail level defect and after
Continuous defect recognition is also easy to produce serious interference.Therefore, it usually requires first to extract rail table before dividing rail surface defects
Face region.Existing rail level extraction algorithm is all based on greatly rail level gray threshold, such as dragon builds force using rail level region ash in image
Angle value is high, and non-rail level area grayscale is worth feature low and that rail width is constant, proposes to rail level column grey level histogram minimum value
Method extracts steel rail area.Dimitris Samaras etc. then extracts rail level edge by edge detection algorithm first, and then positions
Rail Surface region；Orhan Yaman etc. first carries out Threshold segmentation to image, then passes through the side of image after extraction Threshold segmentation
Edge positions Rail Surface region.Above-mentioned rail level extracted region algorithm is in track checking car high vibration, camera deflection, rail bend etc.
Under the influence of environmental factor, exists and accidentally extract and leak the phenomenon that extracting, cannot accurately and fast extract Rail Surface area
Domain.In consideration of it, the present invention provides, a kind of complex condition is accurate, method of rapidly extracting Rail Surface region, above-mentioned to solve
Technical problem.
Summary of the invention
It is influenced when for Rail Surface extracted region vulnerable to factors such as vibration, camera deflection, rail bends, causes accidentally to mention
It takes, leaks extraction, extracts inaccurate problem, the present invention provides a kind of sides in rapidly extracting Rail Surface region under vibration condition
Method, it is ensured that under the influence of complex environment, quick, accurate, stable can extract Rail Surface region.
The present invention through the following technical solutions to achieve the above objectives:
A kind of method in rapidly extracting Rail Surface region under vibration condition, specifically includes the following steps:
S1, Image Acquisition and pretreatment: by image acquisition device rail image, and pass through filtering algorithm and figure
Image intensifying algorithm pre-processes the image of acquisition, to enhance the profile at rail edge and the edge of rail level defect in image；
S2, carrying out image threshold segmentation: analyzing the grey level histogram feature of rail image, determines that suitable threshold value carries out image
Binaryzation；
S3, the connected component for extracting Rail Surface region: pass through 8 adjacent connections point of bianry image after setting Threshold segmentation
The area and perimeter threshold of amount extract the connected component comprising Rail Surface region；
S4, it extracts Rail Surface boundary straight line: straight line line present in rail level edge image is extracted using Hough transform
Section, and delete the line segment that deflection angle is greater than 5 °；
S5, Rail Surface boundary reference point is extracted: equidistantly selection reference on each straight-line segment extracted in step s 4
Point, and determine coboundary with reference to point set and lower boundary with reference to point set according to the position of each line segment；
S6, the fitting amendment of rail level boundary straight line: the slope of fitting up-and-down boundary straight line obtains the deflection angle of boundary straight line
Degree is modified rail level boundary further according to deflection direction and deflection amplitude；
S7, image affine transformation: rotation transformation is carried out to image according to revised deflection angle.
As the prioritization scheme of this case, in above-mentioned steps S4, Hough transform is that the duality detection based on point and line is straight
Line, the vertical range of straight line l to coordinate origin is ρ in rectangular coordinate system, and the vertical line of the straight line and the angle of x-axis are θ, then should
Linear equation are as follows: x cos θ+y sin θ=ρ (1) extracts rail level boundary straight line using Hough transform, specifically includes following step
It is rapid:
1) it by (ρ, θ) space quantization, obtains two-dimensional matrix M (ρ, θ), M (ρ, θ) is accumulator, initial value 0；
2) to each point (x on image borderi,yi), all quantized values of θ are substituted into above-mentioned linear equation (1) and are fallen into a trap
It calculates corresponding ρ, and accumulator is added 1, i.e. M (ρ, θ)=M (ρ, θ)+1；
It 3) will whole (xi,yi) after processing, analyze M (ρ, θ), and M (ρ, θ) is ranked up, select first five maximum value work
For the pole coordinate parameter of rail level boundary sections；
4) by (ρ, θ) and (xi,yi) the common straightway determined in image, and line segment breaking portion is connected, as extract
Rail level boundary straight line.
As the prioritization scheme of this case, in above-mentioned steps S6, least square method is utilized to be fitted rail level boundary straight line.
As the prioritization scheme of this case, in above-mentioned steps S6, the amendment of rail level boundary straight line: upper and lower boundary is straight after fitting
The deflection direction of line be P, deflection amplitude Q, then
According to different P and Q value, transformation is modified to upper and lower boundary；The specific method is as follows:
1) as Q=1, θ is enabled1=θ2；
2) when working as Q=2, P=1 or P=2, θ1Rotate clockwise | θ1-θ2|/2, θ2Rotated along hour hands counterclockwise | θ1-θ2
|/2；When P=3 or P=4, θ1Along rotation counterclockwise | θ1-θ2|/2, θ2Rotated along hour hands clockwise | θ1-θ2|/2；P=5 or P=6
When, θ1Along rotation counterclockwise | θ1-θ2|/2, θ2Rotated along hour hands clockwise | θ1-θ2|/2；
3) as Q=3, θ is enabled1=0 or θ2=0；
4) as Q=4, transformation is not made an amendment.
As the prioritization scheme of this case, in above-mentioned steps S2, image grayscale threshold value is preferably that 40 pairs of images carry out binaryzations
Processing.
The beneficial effects of the present invention are: existing method mainly utilizes rail level gray scale and tiny fragments of stone, coal, etc. stone in rail level extracted region
And the gray scale difference of the extraneous areas such as sleeper is anisotropic, or is extracted using the feature of rail level column gray scale value mutation, the former not can solve
When rail is in hurts such as rail head abrasion, crushing of rail head, there are problems that the leakage of rail level region is extracted；The latter, which not can solve, to be acquired
Rail level region is not in horizontal or vertical relative to entire image in device vibration or rail bend, namely the rail image of acquisition
When direction, there are problems that the leakage of rail level region is extracted and accidentally extracted.The present invention has route length, shape specifically special using rail
Sign, the boundary sections of rail level are extracted using Hough transform principle, and are fitted line segment to extract rail using least square method
Then face region carries out rotation transformation to the rail level region of extraction, the rail level area image extracted is finally made to be in level side
To solving that the leakage of rail level region is extracted caused by by image collecting device vibration or the reasons such as rail bend and accidentally extraction is asked
Topic；By comparing with existing several typical rail level extraction algorithms, speed, precision and the stability of method all have obvious excellent
Gesture is more suitable in high-speed automated rail level Method of Defect Segmentation.
Specific embodiment
Embodiment:
In order to which the purpose of the present invention, content and advantage is more clearly understood, the acquisition of rail image acquisition device is selected
The rail image of 300 concrete sleepers and the rail image of 300 sleepers, specific implementation step of the invention are as follows:
One: image preprocessing
Gray level image is converted by the rail image of acquisition, and image is filtered using Bilateral filter
Pretreatment, to enhance the profile at rail edge and the edge of rail level defect in image, wherein Bilateral filter passes through filtering
Algorithm and algorithm for image enhancement pre-process the image of acquisition；
Two: the coarse extraction of Rail Surface region, including carrying out image threshold segmentation, binaryzation, connected component extract:
1) the rail image grey level histogram feature after gray processing is analyzed, determines that suitable threshold value carries out two-value to image
Change, image grayscale threshold value is preferably that 40 pairs of images carry out binary conversion treatment；Hood is arranged in this case on image collecting device, by
In the effect of hood, so that Image Acquisition visual field is not influenced substantially by available light, and picture edge characteristic is more prominent,
Therefore threshold value can be selected according to many experiments.
2) 8 adjacent connected components are successively marked to the image after binaryzation, and calculates area (the i.e. region of each connected component
Total pixel number) and perimeter (i.e. edges of regions pixel number), being extracted by setting area, the threshold value of perimeter includes Rail Surface region
Connected component；
3) opening operation carried out to the connected component region comprising rail level region of extraction, closed operation, take minimum circumscribed rectangle
Equal operations extract the rectangular area comprising Rail Surface, and carry out expansion appropriate to the up-and-down boundary of the rectangular area.
Three: rail level boundary straight line line segment being extracted using Hough transform, and deletes the line segment that deflection angle is greater than 5 °；
The basic thought of Hough transform detection straight line is duality of the point with line, i.e., there are one in rectangular coordinate system
The vertical range of straight line l, the straight line to origin are ρ, and the vertical line of the straight line and the angle of x-axis are θ, then the linear equation are as follows:
X cos θ+y sin θ=ρ
Specifically, using Hough transform extract rail level boundary straight line the following steps are included:
1) it by (ρ, θ) space quantization, obtains two-dimensional matrix M (ρ, θ), M (ρ, θ) is accumulator, initial value 0；
2) to each point (x on image borderi,yi), all quantized values of θ are substituted into (1) formula and calculate corresponding ρ,
And accumulator is added 1, i.e. M (ρ, θ)=M (ρ, θ)+1；
It 3) will whole (xi,yi) after processing, analyze M (ρ, θ), in order to the case where extracting rail level real border line segment
Under, reach the interference least effect of line segment and M (ρ, θ) is ranked up first herein by many experiments, it is final then to select
Pole coordinate parameter of first five maximum value as rail level boundary sections；
4) by (ρ, θ) and (xi,yi) the common line segment determined in image, and breaking portion is connected.
Four: choose rail level edge reference point:
It is lighted from line segment left end, chooses n on each line segmentsA ordinate point yls,yls+d,yls+2*d,yls+3*d,…,yls
+(ns- 1) d (wherein works as yls+(n-1)d≥yrsWhen, the last one point ordinate takes yrs), ordinate is substituted into corresponding line segment
Linear equation: y=ksx+bsIt can obtain the abscissa of each reference point.
Coboundary can be obtained with reference to point set according to the ownership situation of line segment: ((xu1,yu1),(xu2,yu2),…,(xum,
yum)) and lower boundary refer to point set: ((xd1,yd1),(xd2,yd2),…,(xdn,ydn))。
Five: being fitted rail level edge line method using least square method
It can be fitted to obtain rail level coboundary linear equation: y=k using least square methodux+buAnd lower boundary linear equation: y
=kdx+bd。
Six: the amendment transformation of rail level edge
According to the deflection angle of up-and-down boundary straight line and direction after fitting, defining P is deflection direction, and Q is deflection amplitude, tool
Shown in body is defined as follows:
For different P and Q, the strategy taken herein is first to be modified transformation to upper and lower boundary, and specific method is such as
Under:
1) as Q=1, θ is enabled1=θ2；
2) when working as Q=2, P=1 or P=2, θ1Rotate clockwise | θ1-θ2|/2, θ2Rotated along hour hands counterclockwise | θ1-θ2
|/2；When P=3 or P=4, θ1Along rotation counterclockwise | θ1-θ2|/2, θ2Rotated along hour hands clockwise | θ1-θ2|/2；P=5 or P=6
When, θ1Along rotation counterclockwise | θ1-θ2|/2, θ2Rotated along hour hands clockwise | θ1-θ2|/2；
3) as Q=3, θ is enabled1=0 or θ2=0；
4) as Q=4, transformation is not made an amendment.
Seven: the transformation of rail level image rotation
Using affine transformation image is rotated, wherein rotation transformation is defined as: the image on (w, z) coordinate system
F produces the image f being defined on (x, y) coordinate system by rotation transformation, which can indicate are as follows:
Rail Surface extracted region: to linear equation where the line segment AB of coboundary after the amendment of rail image edge are as follows: y=
kfx+buf, linear equation where lower boundary line segment CD are as follows: y=kfx+bdf, it can thus be concluded that coboundary line segment extreme coordinates: A (0,
buf), B (640,640kf+buf), lower boundary line segment extreme coordinates: C (0, bdf), D (640,640kf+bdf), by above four points
Coordinate substitutes into the extreme coordinates of Rail Surface boundary sections after being rotated in formula (4): A'(x1,y1),B'(x2,y1),C'
(x3,y2),D'(x4,y2).It, can be according to different direction of rotation again since rotation will be deleted rail level left and right edges part
The value of each endpoint is defined, specific value is shown below:
By Hough transform detect line segment endpoint be image pixel coordinates, value be integer, and use straight line fitting and
There may be decimals for line segment endpoint after affine transformation, in order to directly extract Rail Surface area pixel point coordinate, four ends
Point coordinate value be rounded directly up, as x or y be 1.1 when, value takes 2, it can thus be concluded that four of final Rail Surface region
Terminal A ' (x1,y1),E(x4,y2),F(x1,y3),D'(x4,y4) coordinate value.
Finally, using extraction time, extraction accuracy, extracting three standards of stability and existing two kinds of rail level extracting methods
Comparison, it was demonstrated that the method for the present invention at the extraction between it is close with existing method, but extraction accuracy, extract stability be substantially better than it is existing
There is method.Also, rail level extracting method proposed by the present invention is highly suitable in quick rail level defect inspection method.
It combines Threshold segmentation and morphological method such as binaryzation, connected component to extract first to carry out slightly Rail Surface region
It extracts, then extracts Rail Surface edge line segment and reference point by using Hough transform, recycle the method pair of straight line fitting
Rail Surface edge is accurately positioned, and is finally recycled affine transformation to extract horizontal, complete Rail Surface region, is passed through
It is compared with existing several typical rail level extraction algorithms, speed, precision and the stability of this paper all has clear superiority, more applicable
In high-speed automated rail level Method of Defect Segmentation.
Above embodiments are merely exemplary, and can't limit to the present invention, it should be pointed out that for those skilled in the art
For, the other equivalent modifications and improvement made under technical inspiration provided by the present invention are regarded as guarantor of the invention
Protect range.
Claims (3)
1. a kind of method in rapidly extracting Rail Surface region under vibration condition, which is characterized in that specifically includes the following steps:
S1, Image Acquisition and pretreatment: increase by image acquisition device rail image, and by filtering algorithm and image
Strong algorithms pre-process the image of acquisition, to enhance the profile at rail edge and the edge of rail level defect in image；
S2, carrying out image threshold segmentation: analyzing the grey level histogram feature of rail image, determines that suitable threshold value carries out two-value to image
Change；
S3, the connected component for extracting Rail Surface region: pass through 8 adjacent connected components of bianry image after setting Threshold segmentation
Area and perimeter threshold extract the connected component comprising Rail Surface region；
S4, it extracts Rail Surface boundary straight line: straight-line segment present in rail level edge image is extracted using Hough transform, and
Delete the line segment that deflection angle is greater than 5 °；
S5, it extracts Rail Surface boundary reference point: equidistantly selecting reference point on each straight-line segment extracted in step S 4,
And determine coboundary with reference to point set and lower boundary with reference to point set according to the position of each line segment；
S6, the fitting amendment of rail level boundary straight line: the slope of fitting up-and-down boundary straight line obtains the deflection angle of boundary straight line, then
Rail level boundary is modified according to deflection direction and deflection amplitude；
S7, image affine transformation: rotation transformation is carried out to image according to revised deflection angle；
In above-mentioned steps S 4, Hough transform is that the duality based on point and line detects straight line, the straight line in rectangular coordinate systemIt arrives
The vertical range of coordinate origin is, the vertical line of the straight line withThe angle of axis is, then the linear equation are as follows:(1), rail level boundary straight line is extracted using Hough transform, specifically includes the following steps:
1) willSpace quantization obtains two-dimensional matrix,For accumulator, initial value 0；
2) to each point on image border, willAll quantized values substitute into above-mentioned linear equation (1) and calculate
Accordingly, and accumulator is added 1, i.e.,；
It 3) will be wholeAfter processing, analysis, and it is rightBe ranked up, select first five maximum value as
The pole coordinate parameter of rail level boundary sections；
4) byWithThe common straightway determined in image, and line segment breaking portion is connected, the rail as extracted
Face boundary straight line.
2. the method in rapidly extracting Rail Surface region under vibration condition according to claim 1, it is characterised in that: above-mentioned
In step S 6, rail level boundary straight line is fitted using least square method.
3. the method in rapidly extracting Rail Surface region under vibration condition according to claim 1, it is characterised in that: above-mentioned
In step S 2, image grayscale threshold value is preferably that 40 pairs of images carry out binary conversion treatment.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610770769.0A CN106355185B (en) | 2016-08-30 | 2016-08-30 | A kind of method in rapidly extracting Rail Surface region under vibration condition |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610770769.0A CN106355185B (en) | 2016-08-30 | 2016-08-30 | A kind of method in rapidly extracting Rail Surface region under vibration condition |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106355185A CN106355185A (en) | 2017-01-25 |
| CN106355185B true CN106355185B (en) | 2019-07-16 |
Family
ID=57857530
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610770769.0A Expired - Fee Related CN106355185B (en) | 2016-08-30 | 2016-08-30 | A kind of method in rapidly extracting Rail Surface region under vibration condition |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106355185B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107392313B (en) * | 2017-06-12 | 2020-09-29 | 五邑大学 | Steel rail identification method based on deep learning |
| CN109752391B (en) * | 2018-12-25 | 2020-06-30 | 中国农业大学 | Carrot surface defect identification and quantification method based on machine vision |
| CN112233116B (en) * | 2020-12-11 | 2021-08-03 | 常州微亿智造科技有限公司 | Concave-convex mark visual detection method based on neighborhood decision and gray level co-occurrence matrix description |
| CN115661098B (en) * | 2022-10-31 | 2024-02-06 | 河海大学 | Submarine pipeline two-dimensional scouring profile image recognition and data extraction method |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101763512A (en) * | 2009-12-11 | 2010-06-30 | 西安电子科技大学 | Method for semi-automatically detecting road target in high-resolution remote sensing images |
| CN102628814A (en) * | 2012-02-28 | 2012-08-08 | 西南交通大学 | Automatic detection method of steel rail light band abnormity based on digital image processing |
| CN105095914A (en) * | 2015-08-13 | 2015-11-25 | 中国民航大学 | Airport runway detection method based on combination of h/q decomposition and Bayesian iterative classification |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP6299373B2 (en) * | 2014-04-18 | 2018-03-28 | 富士通株式会社 | Imaging direction normality determination method, imaging direction normality determination program, and imaging direction normality determination apparatus |
-
2016
- 2016-08-30 CN CN201610770769.0A patent/CN106355185B/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101763512A (en) * | 2009-12-11 | 2010-06-30 | 西安电子科技大学 | Method for semi-automatically detecting road target in high-resolution remote sensing images |
| CN102628814A (en) * | 2012-02-28 | 2012-08-08 | 西南交通大学 | Automatic detection method of steel rail light band abnormity based on digital image processing |
| CN105095914A (en) * | 2015-08-13 | 2015-11-25 | 中国民航大学 | Airport runway detection method based on combination of h/q decomposition and Bayesian iterative classification |
Non-Patent Citations (1)
| Title |
|---|
| "振动条件下钢轨表面区域快速提取";闵永智 等;《第27届中国过程控制会议（CPCC2016）摘要集》;20160731;第111页摘要 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106355185A (en) | 2017-01-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106355185B (en) | A kind of method in rapidly extracting Rail Surface region under vibration condition | |
| Wang et al. | Research on crack detection algorithm of the concrete bridge based on image processing | |
| CN103839268B (en) | Method for detecting fissure on surface of subway tunnel | |
| CN109754368A (en) | A kind of crack joining method in bridge quality testing | |
| CN102708351B (en) | Method for fast identifying Data Matrix two-dimensional bar code under complicated working condition background | |
| CN105387801B (en) | A kind of subway tunnel section of jurisdiction faulting of slab ends quantity measuring method | |
| CN102750531B (en) | Method for detecting handwriting mark symbols for bill document positioning grids | |
| CN103136528A (en) | Double-edge detection based vehicle license plate identification method | |
| CN107016394B (en) | Cross fiber feature point matching method | |
| CN105426868A (en) | Lane detection method based on adaptive region of interest | |
| CN109993154A (en) | The lithium sulfur type instrument intelligent identification Method of substation's simple pointer formula | |
| CN105608429A (en) | Differential excitation-based robust lane line detection method | |
| CN110567680A (en) | Track fastener looseness detection method based on angle comparison | |
| CN110009633A (en) | A kind of detection method of surface flaw of steel rail based on reversed difference of Gaussian | |
| Zheng et al. | Rail detection based on LSD and the least square curve fitting | |
| CN104850869A (en) | Automatic identification system and method for high-speed railway CPIII marker | |
| CN108226180A (en) | A kind of crack detection method for CRTS II plate-type tracks | |
| CN111311590B (en) | Switch point contact degree detection method based on image detection technology | |
| Liu et al. | A coarse-to-fine detection method of pantograph-catenary contact points using DCNNs | |
| Chen et al. | Vision-based position deviation measurement of rail corrugation chord measuring points under bi-linear laser assistance | |
| CN112857252B (en) | Tunnel image boundary line detection method based on reflectivity intensity | |
| CN104361344B (en) | A kind of seamless track steel rail defect identification method based on interval graph picture | |
| CN114419554A (en) | Monitoring and identifying system for violation of regulations on electric power operation site based on image identification | |
| Li et al. | A hough line detection based glass surface defects recognition algorithm | |
| Sun et al. | An automatic fault recognition method for side frame key in TFDS |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | 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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20190716 Termination date: 20200830 |