CN106570857A - High-speed railway overhead contact system lateral conductor fixation hook nut falling bad state detection method based on HOG features - Google Patents
High-speed railway overhead contact system lateral conductor fixation hook nut falling bad state detection method based on HOG features Download PDFInfo
- Publication number
- CN106570857A CN106570857A CN201610814223.0A CN201610814223A CN106570857A CN 106570857 A CN106570857 A CN 106570857A CN 201610814223 A CN201610814223 A CN 201610814223A CN 106570857 A CN106570857 A CN 106570857A
- Authority
- CN
- China
- Prior art keywords
- fixation hook
- hook
- image
- lateral conductor
- nut
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
- G06T7/0008—Industrial image inspection checking presence/absence
-
- 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
- G06T2207/10—Image acquisition modality
- G06T2207/10004—Still image; Photographic image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
Landscapes
- Engineering & Computer Science (AREA)
- Quality & Reliability (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a high-speed railway overhead contact system lateral conductor fixation hook nut falling bad state detection method based on HOG features. The method comprises the following steps: firstly, a positive and negative sample library for fixation hook parts is built, HOG features of samples are calculated, an SVM classifier is trained, and the classifier obtained through training is used for positioning the lateral conductor fixation hook; then, Hough transformation is adopted to realize extraction of a positioning tube inclination angle in a target image, the positioning tube inclination angle is rotated to a horizontal direction, and segmenting of the fixation hook part is further realized; then, the vertical direction pixel gray values of the fixation hook are accumulated, two ends of a U-type hook are determined, and screws are segmented; and finally, the relationship of ratios of the nut to the bolt in the case of normal operation and falling is concluded, and the working state of the nut is thus determined.
Description
Technical field
The present invention relates to applied to high-speed railway touching net field of fault detection, more particularly to a kind of high ferro based on HOG features connects
The detection method of net-fault lateral conductor fixation hook screw cap-falling defective mode.
Background technology
Limit in positioner in applied to high-speed railway touching net, lateral conductor fixation hook is support fixed Radix Saposhnikoviae bracing wire important zero
Part.In order to ensure the normal work of Radix Saposhnikoviae bracing wire, the cooperation of lateral conductor fixation hook and carrier cable base is fixed and makes Radix Saposhnikoviae bracing wire
In stress.Due to defective mounting, the hold-doun nut of part fixation hook is likely to be at the state for coming off, and makes fixation hook not tight
Gu on positioning pipe, so as to Radix Saposhnikoviae bracing wire does not stress or even drops, to the safe operation of EMUs hidden danger is brought.Therefore it is right to need
The failure of lateral conductor fixation hook screw cap-falling is detected and is taken measures to exclude hidden danger.The 4C systems technologies that the former Ministry of Railways promulgates
Specification, the monitoring of the high sharpness video comprising the suspended portion to contact net, bracket part, is related to based on digital image processing techniques
Fault detect to parts in contact net support and suspension arrangement.
For the detection of elements of contacting net state deficiencies, current China is mainly with traditional manual inspection mode, this field operation
Interior librarian use video camera shoots the supported and suspended installation drawing picture of contact net, and to the malfunction people of each part under off-line state
Work is recognized.But there is also problems with：Workload is big, efficiency is low, breakdown judge has larger hysteresis quality.Based on image procossing skill
The non-contact bow net detection technique research of art is capable of achieving not disturbing the bow net detection means of traffic safety to develop, and device therefor can
Expansion is strong, the automatic identification of bow net parameter and failure is capable of achieving, with numerous advantages.At present both at home and abroad based on image procossing
Existing some researchs of pantograph and catenary fault state-detection, Chen Weirong have studied the pantograph converted based on Morphological scale-space and Radon and slide
Board status are monitored.Zhang Guinan detects contact net insulator breakdown using pyramid neighbour average algorithm and wavelet singular value method, and
Have studied the anti-rotation for realizing insulator based on Harris angle points and spectral clustering to match and fault detect.Liu Yinqiu adopts normalizing
Change cross-correlation and local binarization method, extract and calculate the parameters such as contact net dynamic height and stagger.Due to collection in worksite
Contact net support and suspension arrangement image is universal more complicated, using image processing techniquess to as location hook nut this small zero
There is larger difficulty in the fault detect of part, research in this respect at present yet there are no relevant report.
The content of the invention
The invention provides a kind of high ferro contact net lateral conductor fixation hook screw cap-falling defective mode based on HOG features is examined
Survey method, realizes the detection being accurately positioned with its securing member screw cap-falling defective mode of lateral conductor location hook.
A kind of detection method of the high ferro contact net lateral conductor fixation hook screw cap-falling defective mode based on HOG features, tool
The job step of body includes：
A, special comprehensive row inspection car support to applied to high-speed railway touching net and suspension arrangement are carried out under certain speed of service
Picture, the high-definition image of uplink and downlink is respectively stored in two image libraries；
B, foundation fix the Sample Storehouse of hook part with regard to lateral conductor, and positive sample is the image that fixation hook occupies main body, bear sample
Originally it is not comprising other part diagram pictures of the contact net of fixation hook；
C, fix being accurately positioned for hook part using HOG features and implement the algorithm of support vector machine lateral conductor；
A, in the calculating of HOG features each detection window image spatially position is uniformly divided into several cell lists
Unit.For each pixel I (x, y), using simple single order template calculate in cell factory gradient magnitude m (x, y) and
Direction θ (x, y), such as formula 1,
In cell factory, by quantized interval statistical gradient rectangular histogram set in advance, gradient direction is by 0 degree～360 degree
It is divided into 9 direction blocks, each direction block size is 20 degree, one will be merged in the way of sliding per four adjacent cell factories
Individual block, adjacent block has cell factory overlap；HOG integration description are calculated each cell factory, by 4 in same
The histogram of gradients of cell factory is connected to together, forms the characteristic vector of 9 × 4=36 dimension；Become to eliminate illumination
The impact that brings such as change, in a block column hisgram normalization is entered, such as formula 2,
Wherein, ε is the constant of a very little, it is therefore an objective to avoid denominator from being 0；Characteristic vector v after normalization corresponds to one
HOG integration description of individual block；
B, assume for training sample set be (xi,yi), i=1 ..., l, and xi∈Rn, yi∈{1,-1}.L is sample
Sum, n is characterized dimension, then train the classifying face for obtaining to meet all samples：
Wherein, classifying face is determined that w is the weight vectors for adjudicating plane, and b is threshold value by parameter w with b；ξiLinearly can not
The lax item introduced in the case of point；C is penalty coefficient, for realizing the wrong point of balance between sample proportion and algorithm complex；
C is taken as into 0.01；The minima of φ (w, ξ) in following formula (4) is constrained using solution by iterative method formula (3), you can obtain optimal classification
Face [w*,b*] and optimal classification decision function, as shown in formula (5), sgn represents sign function in formula；For test sample, as long as
Its characteristic vector is substituted into into formula (5), sample class can determine that according to functional value；
F (x)=sgn (w*·x+b*) (5)
The segmentation of D, lateral conductor fixation hook and positioning pipe：
A, in order to exclude positioning pipe to detect interference, fixation hook and positioning pipe are partitioned into from the image for navigating to
Come；Because positioning pipe is linear structure, line is done using Hough transform and detects and link line segment, front 2 are extracted in Hough matrixes
Individual gray scale peak point, detects one group of less parallel line segment, takes its meansigma methods as the angle of inclination of positioning pipe, and positioning pipe is revolved
Go to horizontal direction；
B, using Canny operators to postrotational fixation hook image detection edge, and carry out pixel grey scale in the horizontal direction
Adding up for value, obtains statistic curve；The horizontal line section that the top edge of positioning pipe is located further is found, its place straight line is as solid
Determine the segmentation straight line of hook and positioning pipe that it is located；Intactly fixation hook is split；
The defective mode detection of E, fixation hook screw cap-falling：
The installment state of fixation hook, during screw cap-falling, only deposits bolt, to avoid in the contact net image of analysis collection in worksite
The interference of fixation hook, extracts nut and individually analyzes；
A. because nut is small parts, nut is extracted using the method extracted based on intensity profile law characteristic and is examined
Survey its defective mode；What the size of fixation hook was to determine, according to the length rule of the both sides of fixation hook, divide under same abscissa
The pixel grey scale statistical law of analysis fixation hook, the gray value curve obtained by analytic statisticss determines abscissa at hook bends two,
And then screw is extracted from fixation hook；
B. nut normal work and the two states that come off are observed, pixel grey scale statistical law is calculated under same abscissa：
Because head diameter is more than the diameter of bolt, the diameter d of the screw of normal work is more than diameter d when coming off；Shoot to eliminate
The interference of angle, from head diameter d and bolt length ratio as criterion.
Compared with prior art, the invention has the beneficial effects as follows：
1st, the present invention is directly examined by image processing method to the state that high ferro contact net lateral conductor fixes hook part
Survey, be given it is objective, true, accurately test and analyze result, overcome the defect of Traditional Man detection method.The invention is high ferro
The supported and suspended spacing positioning apparatus component failure detection of contact net provides a kind of preferable thinking；
2nd, the mounting structure feature according to lateral conductor fixation hook of the invention, dexterously by Hough transform and fixation hook gray scale
The regularity of distribution is combined, and simply and effectively judges whether nut comes off；
3rd, the fault detect of the contact net lateral conductor fixation hook being related in the present invention, its research yet there are no relevant report.
In sum, the method for the present invention can be carried out effectively for the failure of contact net lateral conductor fixation hook screw cap-falling
Detection.Correct verification and measurement ratio is higher, simplifies the difficulty of fault detect, and is first contact net lateral conductor fixation hook screw cap-falling
Defective mode detection proposes a kind of solution.
Description of the drawings
Fig. 1 is the processing procedure block diagram of the inventive method.
Fig. 2 is the lateral conductor fixation hook image in collection in worksite image of the present invention.
Fig. 3 is the computational methods schematic diagram of HOG Feature Descriptors.
Fig. 4 for lateral conductor fixation hook positive negative example base, (a) positive sample, (b) negative sample.
Fig. 5 is support vector machine classifier locating effect figure.
Fig. 6 is to contrast before and after lateral conductor fixation hook pretreatment.
Fig. 7 asks for positioning pipe inclination angle for Hough transform, and (a) Hough matrixes extract peak point, (b) Hough transform peak value
Corresponding line segment.
Fig. 8 is fixation hook image vertical direction gray value cartogram.
Fig. 9 is fixation hook and the cutting procedure of positioning pipe, (a) normal condition, (b) screw cap-falling state.
Figure 10 for lateral conductor fixation hook two kinds of installment states, (a) normal condition, (b) screw cap-falling state.
Figure 11 is the cutting procedure of nut and fixation hook, (a) splits bottom, (b) splits nut, and (c) cutting procedure is related
Coordinate determines.
Figure 12 is that nut working condition judges dependent coordinate determination process figure.
Specific embodiment
The present invention is realized by the following means：
1. special comprehensive row inspection car supports to applied to high-speed railway touching net and suspension arrangement is carried out under certain speed of service
Picture.The high-definition image of uplink and downlink is respectively stored in two image libraries.
2. the Sample Storehouse that hook part is positioned with regard to lateral conductor is set up, and positive sample is the figure that lateral conductor location hook occupies main body
Picture, negative sample is not comprising other part diagram pictures of the contact net of location hook.
3. the positioning of lateral conductor location hook and extraction.
Positive sample 100 is chosen, negative sample 500, sample-size normalization calculates the HOG features of positive negative sample, utilizes
HOG feature operators train SVM classifier, are slided in fact by training the grader for obtaining to treat fixed size window in detection image
The target recognition of existing lateral conductor location hook.
4. the segmentation of lateral conductor fixation hook.
In order to preferably analyze the structure of fixation hook, first the process of enhancing contrast ratio is done to extraction, then obtained by Hough transform
The edge of positioning pipe is taken, according to the inclination angle at edge and is rotated to level, followed by Canny operator edge detections, and to detection
Image level direction gray value afterwards adds up, and by statistic curve segmentation straight line is determined.
5. the detection of fixation hook screw cap-falling defective mode.
According to the characteristic of live fixation hook, the length at U-shaped fixation hook two ends is calculated, be partitioned into circular by fixation hook
Screw.Judge whether fixation hook nut comes off according to the ratio magnitude relationship of head diameter and bolt length.
Embodiments of the present invention are described in further detail below in conjunction with the accompanying drawings.
Fig. 1 is the processing procedure block diagram of the inventive method.Fig. 2 illustrates the position of lateral conductor fixation hook in collection in worksite image
Put.
A. the positioning of lateral conductor fixation hook and extraction
Spatially position is uniformly divided into several cell lists to each detection window image in the calculating of HOG features
Unit；For each pixel I (x, y), using simple single order template calculate in cell factory gradient magnitude m (x, y) and
Direction θ (x, y), such as formula 1,
In cell factory, by quantized interval statistical gradient rectangular histogram set in advance, gradient direction is by 0 degree～360 degree
It is divided into 9 direction blocks, each direction block size is 20 degree, one will be merged in the way of sliding per four adjacent cell factories
Individual block, adjacent block has cell factory overlap；HOG integration description are calculated each cell factory, by 4 in same
The histogram of gradients of cell factory is connected to together, forms the characteristic vector of 9 × 4=36 dimension；Become to eliminate illumination
The impact that brings such as change, in a block column hisgram normalization is entered, such as formula 2,
Wherein, ε is the constant of a very little, it is therefore an objective to avoid denominator from being 0；Characteristic vector v after normalization corresponds to one
HOG integration description of individual block；
HOG feature operators have the indeformable of the scaling to image, rotation and brightness flop.Due to can between adjacent block
With the repetition that there is cell factory, a width resolution is that 64 × 128 images include 7 × 14 blocks.Such as Fig. 3, will own in image
The characteristic vector of block links together and obtains the HOG characteristic vectors of entire image, and final HOG Feature Descriptors include 1764
Vector composition dimension.
In integrogram the value of any point (x, y) be defined as in original image the pixel at corresponding coordinate and zero it
Between in rectangle all pixels point gray value sum, i.e.,：
In formula, ii (x, y) is the value of (x, y) coordinate points in integrogram, and i (x', y') is that coordinate is (x', y') in original image
Pixel gray value.
The pixel value in four accessing operation calculating, one rectangular area can be passed through using integrogram, it will be apparent that reduce
The amount of calculation of HOG features.
The HOG features training SVM classifier positive samples for calculating sample are that diagonal brace sleeve is located at image center and occupies image
The image (shown in Fig. 4 (a)) of body position, intercepts 100；Negative sample is random packet containing other contacts unrelated with diagonal brace sleeve
Net parts (shown in Fig. 4 (b)), intercept 500, slide and generate 3000 windows.The size of positive negative sample is normalized to inspection
Survey the size (64 × 128 pixel) of window.
It is assumed that the sample set for training is (xi,yi), i=1 ..., l, and xi∈Rn, yi∈{1,-1}.L is that sample is total
Number, n is characterized dimension, then trains the classifying face for obtaining to meet all samples：
Wherein, classifying face is determined that w is the weight vectors for adjudicating plane, and b is threshold value by parameter w with b.ξiLinearly can not
The lax item introduced in the case of point.C is penalty coefficient, for realizing the wrong point of balance between sample proportion and algorithm complex.
C is taken as into 0.01.The minima of φ (w, ξ) in following formula (3) is constrained using solution by iterative method formula (2), you can obtain optimal classification
Face [w*,b*] and optimal classification decision function, as shown in formula (4), sgn represents sign function in formula.For test sample, as long as
Its characteristic vector is substituted into into formula (4), sample class can determine that according to functional value.
F (x)=sgn (w*·x+b*) (4)
Window sliding is done to contact net global and local image to be detected, the HOG features of sliding window is calculated, using instruction
The lateral conductor fixation hook grader for getting realizes the positioning of fixation hook, as a result as shown in Figure 5.
B. the segmentation of lateral conductor fixation hook
A. the lateral conductor fixation hook image first to extracting carries out the process such as accompanying drawing 6 of enhancing contrast ratio, makes image two-value
Positioning pipe both sides of the edge are closer to straightway during change.
B. the present invention makees line and detects and link line segment using Hough transform, and front 3 gray scale peaks are extracted in Hough matrixes
Value point, such as Fig. 7 a.One group of less parallel line segment is able to detect that, as shown in Fig. 7 b white bars.It is fixed to take its dip mean
The inclination angle of position pipe, positioning pipe is rotated to horizontal direction.
C. edge is detected to rotated image using Canny operators, and the cumulative of grey scale pixel value is carried out in vertical direction,
Obtain statistic curve such as Fig. 8.The horizontal line section of the corresponding greatest length in positioning pipe both sides of the edge is further found, in correspondence Fig. 8
Black is punctuated.Its top edge place straight line is the segmentation straight line of positioning pipe and fixation hook.So fixed hook portion can be complete
Individually split as shown in Figure 9.
C. the defective mode detection of fixation hook screw cap-falling
The installment state of fixation hook in the contact net image of analysis collection in worksite, the present invention is obtained as drawn a conclusion：Nut takes off
When falling, bolt is only deposited, for ease of avoiding the interference of fixation hook, need to extract nut, individually analysis.
A. because nut is small parts, the present invention extracts spiral shell using the method extracted based on intensity profile law characteristic
Cap simultaneously detects its defective mode.According to Chinese railway standard, what the size of fixation hook was to determine, the present invention is according to the two of fixation hook
The length rule of side, analyzes the pixel grey scale statistical law of fixation hook, the gray value obtained by analytic statisticss under same abscissa
Curve, determines abscissa at hook bends two, and then screw is extracted from fixation hook, cutting procedure such as Figure 10.
B. nut normal work and the two states that come off are observed, pixel grey scale statistical law is calculated under same abscissa：
Because head diameter is more than the diameter of bolt, the diameter d of the screw of normal work is more than diameter d when coming off.Shoot to eliminate
The interference of angle, the present invention selects the ratio of head diameter d and bolt length as criterion, such as Figure 11.
D. the defective mode detection of fixation hook screw cap-falling
The installment state of fixation hook in the contact net image of analysis collection in worksite, the present invention is obtained as drawn a conclusion：Nut takes off
When falling, bolt is only deposited, for ease of avoiding the interference of fixation hook, need to extract nut and solely analyze.
A. because nut is small parts, the present invention extracts spiral shell using the method extracted based on intensity profile law characteristic
Cap simultaneously detects its defective mode.According to Chinese railway standard, what the size of fixation hook was to determine, the present invention is according to the two of fixation hook
The length rule of side, analyzes the pixel grey scale statistical law of fixation hook, the gray value obtained by analytic statisticss under same abscissa
Curve, determines abscissa at hook bends two, and then screw is extracted from fixation hook.
B. nut normal work and the two states that come off are observed, pixel grey scale statistical law is calculated under same abscissa：
Because head diameter is more than the diameter of bolt, the diameter d of the screw of normal work is more than diameter d when coming off.Shoot to eliminate
The interference of angle, the present invention selects the ratio of head diameter d and bolt length as judgement reference value.
Claims (1)
1. a kind of detection method of the high ferro contact net lateral conductor fixation hook screw cap-falling defective mode based on HOG features, it is special
Levy and be, specific job step includes：
A, special comprehensive row inspection car support to applied to high-speed railway touching net and suspension arrangement are imaged under certain speed of service,
The high-definition image of uplink and downlink is respectively stored in two image libraries；
B, foundation fix the Sample Storehouse of hook part with regard to lateral conductor, and positive sample is the image that fixation hook occupies main body, and negative sample is
Contact net not comprising fixation hook other part diagram pictures；
C, fix being accurately positioned for hook part using HOG features and implement the algorithm of support vector machine lateral conductor；
A, in the calculating of HOG features each detection window image spatially position is uniformly divided into several cell factories；
For each pixel I (x, y), gradient magnitude m (x, y) and direction are calculated in cell factory using simple single order template
θ (x, y), such as formula 1,
In cell factory, by quantized interval statistical gradient rectangular histogram set in advance, gradient direction is divided into 9 by 0 degree～360 degree
Individual direction block, each direction block size is 20 degree, will merge into a block in the way of sliding per four adjacent cell factories,
Adjacent block has cell factory overlap；HOG integration description are calculated each cell factory, by 4 cells in same
The histogram of gradients of unit is connected to together, forms the characteristic vector of 9 × 4=36 dimension；In order to eliminate illumination variation etc.
The impact for bringing, in a block column hisgram normalization is entered, such as formula 2,
Wherein, ε is the constant of a very little, it is therefore an objective to avoid denominator from being 0；Characteristic vector v after normalization corresponds to a block
HOG integration description son；
B, assume for training sample set be (xi,yi), i=1 ..., l, and xi∈Rn, yi∈{1,-1}；L is total sample number,
N is characterized dimension, then train the classifying face for obtaining to meet all samples：
Wherein, classifying face is determined that w is the weight vectors for adjudicating plane, and b is threshold value by parameter w with b；ξiFor linearly inseparable situation
The lax item of lower introducing；C is penalty coefficient, for realizing the wrong point of balance between sample proportion and algorithm complex；C is taken as
0.01；The minima of φ (w, ξ) in following formula (4) is constrained using solution by iterative method formula (3), you can obtain optimal classification surface [w*,b*]
With optimal classification decision function, as shown in formula (5), sgn represents sign function in formula；For test sample, as long as by its feature
Vectorial substitution formula (5), according to functional value sample class is can determine that；
F (x)=sgn (w*·x+b*) (5)
The segmentation of D, lateral conductor fixation hook and positioning pipe：
A, in order to exclude positioning pipe to detect interference, fixation hook and positioning pipe are split from the image for navigating to；By
It is linear structure in positioning pipe, line is done using Hough transform and detects and link line segment, front 2 gray scales is extracted in Hough matrixes
Peak point, detects one group of less parallel line segment, takes its meansigma methods as the angle of inclination of positioning pipe, and positioning pipe is rotated to water
Square to；
B, using Canny operators to postrotational fixation hook image detection edge, and carry out grey scale pixel value in the horizontal direction
It is cumulative, obtain statistic curve；The horizontal line section that the top edge of positioning pipe is located further is found, its place straight line is fixation hook
With the segmentation straight line of positioning pipe that it is located；Intactly fixation hook is split；
The defective mode detection of E, fixation hook screw cap-falling：
The installment state of fixation hook, during screw cap-falling, only deposits bolt in the contact net image of analysis collection in worksite, to avoid fixing
The interference of hook, extracts nut and individually analyzes；
A. because nut is small parts, nut is extracted using the method extracted based on intensity profile law characteristic and it is detected
Defective mode；What the size of fixation hook was to determine, according to the length rule of the both sides of fixation hook, analyze solid under same abscissa
Determine the pixel grey scale statistical law of hook, the gray value curve obtained by analytic statisticss determines abscissa at hook bends two, and then
Screw is extracted from fixation hook；
B. nut normal work and the two states that come off are observed, pixel grey scale statistical law is calculated under same abscissa：Due to
Head diameter is more than the diameter of bolt, and the diameter d of the screw of normal work is more than diameter d when coming off；To eliminate shooting angle
Interference, from head diameter d and bolt length ratio as criterion.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610814223.0A CN106570857A (en) | 2016-09-11 | 2016-09-11 | High-speed railway overhead contact system lateral conductor fixation hook nut falling bad state detection method based on HOG features |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610814223.0A CN106570857A (en) | 2016-09-11 | 2016-09-11 | High-speed railway overhead contact system lateral conductor fixation hook nut falling bad state detection method based on HOG features |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106570857A true CN106570857A (en) | 2017-04-19 |
Family
ID=58531707
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610814223.0A Pending CN106570857A (en) | 2016-09-11 | 2016-09-11 | High-speed railway overhead contact system lateral conductor fixation hook nut falling bad state detection method based on HOG features |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106570857A (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107633267A (en) * | 2017-09-22 | 2018-01-26 | 西南交通大学 | A kind of high iron catenary support meanss wrist-arm connecting piece fastener recognition detection method |
| CN108305261A (en) * | 2017-08-11 | 2018-07-20 | 腾讯科技（深圳）有限公司 | Picture segmentation method, apparatus, storage medium and computer equipment |
| CN108846331A (en) * | 2018-05-30 | 2018-11-20 | 北京天亿时代科技有限公司 | The video frequency identifying method whether a kind of EMU chassis screw fastener falls off |
| CN109389566A (en) * | 2018-10-19 | 2019-02-26 | 辽宁奇辉电子系统工程有限公司 | Subway height adjusting valve fastening nut defective mode detection method based on boundary characteristic |
| CN111127448A (en) * | 2019-12-26 | 2020-05-08 | 西南交通大学 | Model and method for detecting air spring fault based on isolated forest |
| CN111145154A (en) * | 2019-12-25 | 2020-05-12 | 西北工业大学 | Machine vision-based serial steel wire anti-loosening structure detection method |
| CN111754508A (en) * | 2020-07-06 | 2020-10-09 | 中兴飞流信息科技有限公司 | Contact net limiting nail-stopping allowance state detection system |
| CN112102280A (en) * | 2020-09-11 | 2020-12-18 | 哈尔滨市科佳通用机电股份有限公司 | Method for detecting loosening and loss faults of small part bearing key nut of railway wagon |
| CN113705660A (en) * | 2021-08-25 | 2021-11-26 | 青岛睿维申信息科技有限公司 | Target identification method and related equipment |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104281838A (en) * | 2014-09-23 | 2015-01-14 | 西南交通大学 | Lug piece breakage detection method for high-speed rail overhead line system supporting device based on HOG features and two-dimensional Gabor wavelet transformation |
| CN104318582A (en) * | 2014-11-14 | 2015-01-28 | 西南交通大学 | Detection method for bad state of rotating double-lug component pin of high-speed rail contact network on basis of image invariance target positioning |
| CN104866865A (en) * | 2015-05-11 | 2015-08-26 | 西南交通大学 | DHOG and discrete cosine transform-based overhead line system equilibrium line fault detection method |
| CN105719305A (en) * | 2016-01-25 | 2016-06-29 | 成都国铁电气设备有限公司 | Assembly falloff defect identification method and system of overhead contact system |
-
2016
- 2016-09-11 CN CN201610814223.0A patent/CN106570857A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104281838A (en) * | 2014-09-23 | 2015-01-14 | 西南交通大学 | Lug piece breakage detection method for high-speed rail overhead line system supporting device based on HOG features and two-dimensional Gabor wavelet transformation |
| CN104318582A (en) * | 2014-11-14 | 2015-01-28 | 西南交通大学 | Detection method for bad state of rotating double-lug component pin of high-speed rail contact network on basis of image invariance target positioning |
| CN104866865A (en) * | 2015-05-11 | 2015-08-26 | 西南交通大学 | DHOG and discrete cosine transform-based overhead line system equilibrium line fault detection method |
| CN105719305A (en) * | 2016-01-25 | 2016-06-29 | 成都国铁电气设备有限公司 | Assembly falloff defect identification method and system of overhead contact system |
Non-Patent Citations (1)
| Title |
|---|
| 刘东升 等: "一种基于HOG的快速人体检测方法", 《电子设计工程》 * |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108305261A (en) * | 2017-08-11 | 2018-07-20 | 腾讯科技（深圳）有限公司 | Picture segmentation method, apparatus, storage medium and computer equipment |
| CN107633267A (en) * | 2017-09-22 | 2018-01-26 | 西南交通大学 | A kind of high iron catenary support meanss wrist-arm connecting piece fastener recognition detection method |
| CN108846331A (en) * | 2018-05-30 | 2018-11-20 | 北京天亿时代科技有限公司 | The video frequency identifying method whether a kind of EMU chassis screw fastener falls off |
| CN108846331B (en) * | 2018-05-30 | 2022-07-19 | 北京天亿时代科技有限公司 | Video identification method for judging whether screw fastener of motor train unit chassis falls off or not |
| CN109389566B (en) * | 2018-10-19 | 2022-01-11 | 辽宁奇辉电子系统工程有限公司 | Method for detecting bad state of fastening nut of subway height adjusting valve based on boundary characteristics |
| CN109389566A (en) * | 2018-10-19 | 2019-02-26 | 辽宁奇辉电子系统工程有限公司 | Subway height adjusting valve fastening nut defective mode detection method based on boundary characteristic |
| CN111145154A (en) * | 2019-12-25 | 2020-05-12 | 西北工业大学 | Machine vision-based serial steel wire anti-loosening structure detection method |
| CN111145154B (en) * | 2019-12-25 | 2022-04-01 | 西北工业大学 | Machine vision-based serial steel wire anti-loosening structure detection method |
| CN111127448A (en) * | 2019-12-26 | 2020-05-08 | 西南交通大学 | Model and method for detecting air spring fault based on isolated forest |
| CN111127448B (en) * | 2019-12-26 | 2022-05-10 | 西南交通大学 | Method for detecting air spring fault based on isolated forest |
| CN111754508A (en) * | 2020-07-06 | 2020-10-09 | 中兴飞流信息科技有限公司 | Contact net limiting nail-stopping allowance state detection system |
| CN112102280A (en) * | 2020-09-11 | 2020-12-18 | 哈尔滨市科佳通用机电股份有限公司 | Method for detecting loosening and loss faults of small part bearing key nut of railway wagon |
| CN113705660A (en) * | 2021-08-25 | 2021-11-26 | 青岛睿维申信息科技有限公司 | Target identification method and related equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106570857A (en) | High-speed railway overhead contact system lateral conductor fixation hook nut falling bad state detection method based on HOG features | |
| CN104318582B (en) | Detection method for bad state of rotating double-lug component pin of high-speed rail contact network | |
| CN106372667B (en) | A kind of high iron catenary diagonal brace sleeve part screw defective mode detection method | |
| CN106919978B (en) | Method for identifying and detecting parts of high-speed rail contact net supporting device | |
| CN106485694B (en) | A kind of high iron catenary double-jacket tube connector six-sided nut based on cascade classifier falls off defective mode detection method | |
| CN106340019A (en) | Method for detecting adverse state of high-speed rail overhead line system inclined cable fixing hook | |
| CN105719305B (en) | Component falls off defect identification method and system in contact net | |
| CN108009591B (en) | Contact network key component identification method based on deep learning | |
| CN105426905B (en) | Robot barrier object recognition methods based on histogram of gradients and support vector machines | |
| CN105158257B (en) | Slide plate measurement method and device | |
| CN104281838A (en) | Lug piece breakage detection method for high-speed rail overhead line system supporting device based on HOG features and two-dimensional Gabor wavelet transformation | |
| CN105741291A (en) | Method for detecting faults of equipotential lines of high-speed railway overhead line system suspension devices | |
| CN111402215B (en) | Contact net insulator state detection method based on robust principal component analysis method | |
| CN103745224A (en) | Image-based railway contact net bird-nest abnormal condition detection method | |
| CN105354589A (en) | Method and system for intelligently identifying insulator crack in catenary image | |
| CN104866865A (en) | DHOG and discrete cosine transform-based overhead line system equilibrium line fault detection method | |
| CN104281857B (en) | A kind of isolator detecting method projected based on orthogonal direction | |
| CN105718964B (en) | A kind of visible detection method of power transmission line damper | |
| CN111462045B (en) | Method for detecting defects of catenary support assembly | |
| CN106446934A (en) | High speed rail overhead contact system carrier cable seat support pedestal detection method based on HOG (Histogram) features | |
| CN106290393A (en) | A kind of high ferro contact net inclined cantilever pipe cap loss detection method | |
| CN105354827A (en) | Method and system for intelligently identifying clamp nut shedding in catenary image | |
| CN114030395B (en) | Foreign matter detection method and system for contact suspension dropper area | |
| CN111127409A (en) | Train component detection method based on SIFT image registration and cosine similarity | |
| CN106446929B (en) | Type of vehicle detection method based on edge gradient potential energy |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170419 |