CN113780464A - Method for detecting anti-loose identification of bogie fastener - Google Patents
Method for detecting anti-loose identification of bogie fastener Download PDFInfo
- Publication number
- CN113780464A CN113780464A CN202111126078.4A CN202111126078A CN113780464A CN 113780464 A CN113780464 A CN 113780464A CN 202111126078 A CN202111126078 A CN 202111126078A CN 113780464 A CN113780464 A CN 113780464A
- Authority
- CN
- China
- Prior art keywords
- image
- bogie
- classifier
- loose
- fastener
- 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 description 31
- 238000012545 processing Methods 0.000 claims abstract description 44
- 238000001514 detection method Methods 0.000 claims abstract description 18
- 238000007781 pre-processing Methods 0.000 claims abstract description 14
- 238000012549 training Methods 0.000 claims abstract description 13
- 238000004806 packaging method and process Methods 0.000 claims abstract description 4
- 230000005477 standard model Effects 0.000 claims abstract description 4
- 239000002131 composite material Substances 0.000 claims description 32
- 238000012937 correction Methods 0.000 claims description 10
- 238000007689 inspection Methods 0.000 claims description 6
- 238000001914 filtration Methods 0.000 claims description 4
- 238000005516 engineering process Methods 0.000 abstract description 4
- 238000004422 calculation algorithm Methods 0.000 description 4
- 238000013135 deep learning Methods 0.000 description 4
- 238000012423 maintenance Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 238000005286 illumination Methods 0.000 description 3
- 238000012216 screening Methods 0.000 description 3
- 238000011179 visual inspection Methods 0.000 description 3
- 150000001875 compounds Chemical class 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000009825 accumulation Methods 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000007635 classification algorithm Methods 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 238000011897 real-time detection Methods 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to a track traffic bogie overhauling technology, in particular to a bogie fastener anti-loose identification detection method. S1, acquiring standard model data of various fasteners in the bogie with the specified model and standard sample data of anti-loose marks of all the fasteners; s2, training a classifier with a three-layer cascade structure by using the standard sample data obtained in S1, and packaging the trained classifier in an image data processing system; s3, acquiring anti-loose identification images of all fasteners on the bogies to be inspected in the same model; s4, preprocessing the anti-loose identification image acquired in the step S3; s5, putting the step image preprocessing result into a classifier with a three-layer cascade structure for classification; and S6, comparing the classification result in the step S5 with the corresponding sample parameter, thereby judging whether the anti-loose mark is qualified. The anti-loosening mark recognition rate is improved, the image data processing and analyzing speed is improved, the learning capability is realized, and the anti-loosening mark recognition system is suitable for bogies of various models.
Description
Technical Field
The invention relates to a track traffic bogie overhauling technology, in particular to a bogie fastener anti-loose identification detection method.
Background
At present, the fixed connection parts on the railway train bogie are connected by screw threads in most cases, and the adopted screw threads are connected with various forms such as bolts and nuts, air valves and pipe joints and the like. The threaded connection has the advantage of convenient installation and disassembly.
However, it is difficult to avoid the vibration during the operation of the train, and as the train runs for a long time, the problem of looseness at the threaded connection part due to the accumulation of vibration can occur. Thereby causing the related equipment to be incapable of operating normally and seriously endangering the driving safety. Therefore, whether the threaded connection part (the fastener) is loosened or not is an important link for train overhaul and maintenance, and meanwhile, the check of the fastener anti-loosening mark is an important content of the clamping control inspection operation after the bogie is assembled. In order to easily identify whether the threaded connection is loosened, anti-loosening marks are made on the fastened threaded connection on the bogie, so that whether the anti-loosening marks are aligned and complete can be checked to judge whether the threaded connection on the bogie is loosened.
During train maintenance, the check of the anti-loose marks at the threaded connection part on the bogie comprises several methods such as manual visual inspection, automatic detection based on model matching, intelligent detection based on deep learning and the like. These three methods and their disadvantages will be described one by one.
Manual visual inspection: in the traditional method, a worker visually inspects the bogie and takes a picture of the threaded connection part of the bogie of the train and stores the data. When the maintenance is carried out, the staff is required to identify whether the anti-loosening marks are staggered or not at the bottom of the train through human eyes, and whether the anti-loosening marks are aligned or not is judged, so that the development of subsequent inspection and maintenance work is ensured. Still another manual visual inspection method is to take a picture of the position of the car bottom bogie where the anti-loosening mark is located by various automatic devices, and then the worker checks the position where the anti-loosening mark is abnormal by watching the picture. Hundreds of thousands of photos are required to be acquired by a train, and the mode is extremely labor-consuming and material-consuming. So that a team is caused to need more than ten people to work without interruption.
The automatic detection method based on model matching comprises the following steps: the method is mainly used for detecting the anti-loose mark at the hexagonal bolt by modeling and matching the same fastener, but if the bolt is just loosened and rotates by integral multiple of 60 degrees, the method can generate misjudgment; the method cannot complete the anti-loose identification detection of the circular pipe hoop and the like.
Intelligent detection method based on deep learning: the intelligent detection method performs sample training through a large amount of data, so that a high recognition rate can be obtained. For the intelligent detection method based on deep learning to improve the recognition rate, hundreds of millions of data are needed for training samples; however, for non-popular data sets, this is a time consuming and laborious and extremely difficult task to achieve. And the intelligent detection method based on deep learning lacks robustness, and the detection structure can be influenced by changes of environment, different illumination and the like.
Disclosure of Invention
In order to solve the problems that the detection of the anti-loose identifier of the fastening piece on the bogie is time-consuming and labor-consuming, the recognition rate is easily influenced by the environment and the like, the invention provides an intelligent method for detecting the anti-loose identifier of the fastening piece of the bogie.
In order to achieve the purpose, the invention adopts the following technical scheme: the method for detecting the anti-loose identifier of the bogie fastener is used for a composite robot with an autonomous positioning and navigation function and used for image acquisition, and an image data processing system, and comprises the following steps:
s1, acquiring standard model data of various fasteners in the bogie with the specified model and standard sample data of anti-loose marks of all the fasteners;
s2, training a classifier with a three-layer cascade structure by using the standard sample data obtained in S1, and packaging the trained classifier in an image data processing system;
s3, acquiring anti-loose identification images of all fasteners on the bogies to be inspected in the same model;
s4, preprocessing the anti-loose identification image obtained in the step S3 to obtain data used in the subsequent steps;
s5, putting the image preprocessing result of the step S4 into a classifier of the three-layer cascade structure for classification;
and S6, comparing the classification result in the step S5 with the corresponding sample parameter, thereby judging whether the anti-loose mark is qualified.
Compared with the prior art, the fastener anti-loose identification detection method provided by the invention has the beneficial effects that:
(1) the recognition rate of the anti-loosening mark is improved. The invention can carry out various classified identification on the complex anti-loose mark image data, greatly reduces the problems of non-identification, false identification and the like caused by the problems of illumination, angle, dirt covering and the like, and has the total correct identification rate of over 99.2 percent.
(2) The speed of image data processing and analysis is improved. Compared with the convolutional neural network, the speed is faster during training, and especially for a large training set, the recognition rate is slightly higher than that of the neural network. The method has the advantages of less required training samples and simpler and more convenient operation, greatly improves the software running speed, and can achieve the effect of real-time detection.
(3) The device has learning ability and can adapt to various bogies of different models. Compared with a model matching mode, the method is better in adaptability. The invention is not limited to a certain bogie, and can be suitable for any vehicle type by carrying out sample training aiming at national standard bolts of different types; by continuously training the classifier, the method can adapt to vehicle models with various specifications.
In order to obtain better technical effects, the invention can be further improved on the basis of the technical scheme.
Further, the step of acquiring the standard sample data in step S1 is as follows:
s11: selecting a bogie which is newly delivered or subjected to card control inspection to ensure that all images in the acquired standard sample data are clear and complete, and fixedly placing the specified bogie on a bogie positioning platform with a lifting function;
s12: teaching and photographing anti-loosening marks of the bogie by using a composite robot, performing model matching on photographed image data by using the standard fastener model data by using the image data processing system, determining the type and the model of each fastener on the bogie, and recording the position information of each fastener; recording a traveling route and an image acquisition point of the composite robot and position and posture information when an image is acquired, and simultaneously forming a teaching program corresponding to the specified bogie type;
s13: the image data processing system divides and identifies the shot image to acquire an image of the fastener anti-loosening mark.
Further, in step S3, S31: placing a bogie to be measured on the bogie positioning platform in the same mode and posture, and completing image acquisition by the composite robot in the same position and posture according to the teaching program of the corresponding type of bogie; s32: and the image data processing unit divides and identifies the acquired image to acquire the anti-loosening identification image to be detected.
Further, in step S3, after the composite robot travels to the taught image capturing point, the composite robot performs secondary positioning, and adjusts the position and posture of image capturing according to the secondary positioning result.
Further, in step S2, the three-layer cascade structure classifier includes 8 sub-classifiers; wherein the first layer of classifiers comprises an HSV classifier and a color component classifier; the second layer classifier comprises a brightness classifier and a contrast classifier; the third layer of classifiers includes a length classifier, a width classifier, an area classifier, and a morphology classifier.
Further, the image preprocessing in step S4 includes the following specific steps:
s41: filtering the anti-loose identification image data; carrying out white balance processing on the filtered image; then carrying out distortion correction processing on the image subjected to the white balance processing, and storing the image subjected to the distortion correction processing;
s42: converting the image after the distortion correction processing into an RGB color space image and storing the RGB color space image; decomposing the image of the RGB color space into a single-channel image, and respectively obtaining corresponding R, G and B values; and obtaining and storing a corresponding HSV color space image by using the values of R, G and B.
Further, when the bogies with various specifications need to be detected, a specification mark is pasted at the specified position of the bogie to be detected before detection, and the composite robot acquires the model number information of the bogie to be detected by collecting and identifying the specification mark; and selecting and executing a corresponding teaching program according to the bogie type number information composite robot to finish the image acquisition operation.
Drawings
FIG. 1 is a flow chart of a method of an embodiment of the present invention.
FIG. 2 is an image processing flow diagram of an embodiment of the present invention.
FIG. 3 is a diagram illustrating the structure relationship and flow of the cascade classifier according to an embodiment of the present invention.
Fig. 4 is a layout diagram of a two-dimensional code in a work area when the embodiment of the present invention is used.
Detailed Description
The main inventive concept of the invention is as follows: on the basis of an industrial robot and a high-definition imaging technology, shooting images of anti-loosening marks of all fasteners on a bogie in a proper pose; and then the shot (collected) images are segmented, processed, classified and identified by utilizing an image processing technology. The most important of the methods is the identification of the anti-loose identifier of the fastener through color identification and the classification of a cascade structure classifier based on a clustering algorithm; and comprehensively comparing the anti-loosening mark with the corresponding standard sample parameter, and judging whether the anti-loosening mark is qualified.
The composite robot applied by the invention mainly comprises an AGV, a base, a UR mechanical arm (6-axis series connection type), an image acquisition device (an industrial camera and an industrial lens), a controller, an industrial personal computer and the like.
The AGV is used as a walking part of the composite robot and is used for realizing the autonomous movement and navigation of the composite robot. The base is fixedly installed on the AGV, and the UR mechanical arm is fixed on the base. The image acquisition device is installed at UR arm end, utilizes the flexibility of UR arm to realize the multi-angle and shoot to satisfy the needs of shooing of the locking sign of different fasteners. The base is of a box body or frame structure, and the controller and the industrial personal computer are arranged in the base; the AGV and the UR mechanical arm are respectively electrically connected with the controller and in signal connection; the controller and the image acquisition device are respectively electrically connected and in signal connection with the industrial personal computer.
The image data processing system according to the present invention includes an image preprocessing module, an image analyzing and recognizing module, a data storing module, and the like.
The invention will be described in detail with reference to the accompanying drawings and examples.
The method for detecting the anti-loose identifier of the bogie fastener comprises the following steps:
and S1, acquiring standard model data of various fasteners in the bogie with the specified model and standard sample data of all fastener anti-loose marks.
And S2, training the classifier with the three-layer cascade structure by using the standard sample data acquired in the step S1, and packaging the trained classifier in an image data processing system.
And S3, acquiring anti-loose identification images of all fasteners on the bogies to be inspected in the same model.
S4, preprocessing the anti-loose identification image obtained in the step S3 to obtain data used in the subsequent steps.
And S5, putting the image preprocessing result of the step S4 into the classifier with the three-layer cascade structure for classification.
And S6, comparing the classification result in the step S5 with the corresponding sample parameter, thereby judging whether the anti-loose mark is qualified.
In practice, the method is used for checking the anti-loosening mark of the bogie, and the steps S1 and S2 are both preliminary preparation work. In practice, after the bogie locking identification card control inspection operation can be completed according to the flow shown in fig. 1.
When the detection method provided by the invention is used for identifying the anti-loose identifier, the following implementation scheme can be selected preferentially.
As a preferred embodiment, the detailed steps of acquiring the standard sample data in step S1 are as follows:
s11: selecting a bogie which is newly delivered or subjected to card control inspection to ensure that all images in the acquired standard sample data are clear and complete, and fixedly placing the specified bogie on a bogie positioning platform with a lifting function.
Namely, the bogie with complete and undamaged locking identifiers of all fasteners is selected for image acquisition so as to obtain high-quality standard sample data.
S12: teaching and photographing anti-loosening marks of the bogie by using a composite robot, performing model matching on photographed image data by using the standard fastener model data by using the image data processing system, determining the type and the model of each fastener on the bogie, and recording the position information of each fastener; and recording the traveling route, the image acquisition point and the position and posture information of the composite robot when the image is acquired, and simultaneously forming a teaching program corresponding to the specified bogie type number.
S13: the image data processing system divides and identifies the shot image to acquire an image of the fastener anti-loosening mark. The image is segmented to obtain and the complete anti-loosening identification image is reserved, so that the data processing amount of the following algorithm can be greatly reduced.
As a preferred embodiment, in step S3:
s31: and placing the bogie to be measured on the bogie positioning platform in the same mode and posture, and finishing image acquisition by the composite robot in the same position and posture according to the teaching program of the corresponding type of bogie.
S32: and the image data processing unit divides and identifies the acquired image to acquire the anti-loosening identification image to be detected.
In this process (S31), the compound robot executes the capturing operation of the image in accordance with the teaching program. In the embodiment, the composite robot travels and navigates in a two-dimensional code scanning mode. On the ground of the card control operation area, corresponding two-dimensional codes are laid around the bogie at fixed positions according to teaching positions, and the two-dimensional code layout of the embodiment is as shown in fig. 4. In the actual implementation process, a plurality of two-dimensional codes are additionally pasted on the running route of the composite robot, the more the two-dimensional codes are, the smaller the running error of the composite robot is, and the smaller the subsequent working error is. When the invention is implemented, the composite robot can also adopt other navigation modes, such as magnetic stripe navigation, laser navigation or visual navigation, according to the field situation and the actual needs. As long as the positioning requirements and the applicable field conditions can be met, the method can be used as a navigation mode of the AGV (namely a navigation mode of the composite robot).
As a preferred embodiment, in step S3, after the composite robot walks to the taught image capturing point, the composite robot performs secondary positioning, and adjusts the position and orientation of image capturing according to the secondary positioning result.
The purpose of secondary positioning is mainly to solve the problems of missed detection or incapability of shooting and the like caused by the deviation of the placing position of the bogie and the deviation of the position of the anti-loosening mark.
The secondary positioning method is mainly characterized in that the deviation between the shooting pose of the anti-loosening mark of the fastener and the teaching pose is calculated according to the pose of the fastener in the world coordinate system and the current position of the mechanical arm in the composite robot. Then the deviation compensation is completed through the movement of the UR mechanical arm in the compound robot, and then the image acquisition (photographing) is performed again on the fastener.
As a preferred embodiment, in step S2, the three-layer cascade structure classifier includes 8 sub-classifiers in total; wherein the first layer of classifiers comprises an HSV classifier and a color component classifier; the second layer classifier comprises a brightness classifier and a contrast classifier; the third layer of classifiers includes a length classifier, a width classifier, an area classifier, and a morphology classifier.
Screening an anti-loose mark image by using a color classifier, wherein the anti-loose mark is marked by red, and the red is changed into magenta after being faded after long-time use, so that the first-layer classifier performs color screening by setting a threshold value of a related color; in the second-layer classifier, the contrast classifier uses the contrast to enable the marking lines to be more prominent, on the basis, the hue is used for carrying out second judgment, and the brightness classifier screens the marking lines according to the brightness of the image; and the third layer of classifier synthesizes the length, width, area and form (shape and posture) of the anti-loose mark to carry out final comprehensive classification screening. The determination is made for the segmented mark lines using the relative pixel distance between the marks.
As a preferred embodiment, referring to fig. 2, the specific steps of the image preprocessing in step S4 are as follows:
s41: filtering the anti-loose identification image data; carrying out white balance processing on the filtered image; and then carrying out distortion correction processing on the image subjected to the white balance processing, and storing the image subjected to the distortion correction processing.
Job of image preprocessing: the image data processing system carries out filtering processing on the original image of the anti-loosening mark to remove noise in the image; white balance processing is carried out on the filtered image, and deviation of image colors caused by illumination problems is reduced; and finally, carrying out distortion correction processing on the image after the white balance processing, wherein the purpose is to reduce the distortion introduced by the deviation of the lens manufacturing precision and the assembly process, thereby solving the distortion problem of the original image caused by the distortion.
S42: converting the image after the distortion correction processing into an RGB color space image and storing the RGB color space image; decomposing the image of the RGB color space into a single-channel image, and respectively obtaining corresponding R, G and B values; and obtaining and storing a corresponding HSV color space image by using the values of R, G and B.
Converting the decomposed anti-loose identification image data from an RGB color space into an HSV color space, and extracting three components of H (chroma), S (saturation) and V (brightness), wherein the specific flow of the process is as follows:
Min := min([R, G, B])；
Max := max([R, G, B])；
V := Max；
S := (Max - Min) / Max；
if (R = = Max)
H = ((G-B)/(Max-Min)). pi/3;
if (G = = Max)
H = (2 + (B-R)/(Max-Min)). pi/3;
if (B = = Max)
Then H = (4 + (R-G)/(Max-Min)). pi/3.
In the above-mentioned flow scheme H is belonged to [0;2 pi ], S is belonged to [0;1], and V is belonged to [0;1 ].
After the image preprocessing is finished, the selection tree classifier algorithm is adopted to carry out classification algorithm recognition on the anti-loose identification image data of the RGB color space and the HSV color space, and the specific flow of the algorithm is shown in figure 3.
As a preferred embodiment, when bogies of various specifications need to be detected, a specification mark is pasted at a specified position of the bogie to be detected before detection, and the composite robot acquires the model number information of the bogie to be detected by collecting and identifying the specification mark; and selecting and executing a corresponding teaching program according to the bogie type number information composite robot to finish the image acquisition operation.
The following supplementary explanation is provided to better understand the present invention.
Firstly, through secondary positioning (UR mechanical arm photographing pose and photographing path planning), the anti-loose mark can be effectively and accurately positioned, and the problem of low recognition rate when a mark line is recognized in a large area is solved; the image data required to be collected and processed is greatly reduced, the image processing time is reduced, more accurate data are provided for the classification and identification of subsequent images, and the identification rate of the anti-loosening mark is further improved.
And storing the original image data of the anti-loosening mark acquired by the image acquisition device and the characteristic parameter data analyzed and recognized by the image data processing system in a corresponding storage module of the image data processing system. And identifying and extracting various characteristic parameter data of the classified anti-loosening marks, wherein the characteristic parameters comprise the maximum width and the minimum width of the anti-loosening marks, whether the anti-loosening marks are continuous or skew, and the positions of the anti-loosening marks in a world coordinate system.
Training principles of the classifiers and relations among the classifiers are as follows: the sets formed by the sample classes of each feature in each trainer sample are mutually exclusive pairwise, and the union of all the sets is all the sample classes. Based on such a rule, a set of corresponding classifiers is trained.
In training the classifier, for each feature classifier, the output result is determined according to the result obtained by comparing a certain feature value of the sample with the threshold value of the classifier.
Claims (7)
1. The method for detecting the anti-loose identifier of the bogie fastener is used for a composite robot with an autonomous positioning and navigation function and used for image acquisition and an image data processing system, and is characterized by comprising the following steps of:
s1, acquiring standard model data of various fasteners in the bogie with the specified model and standard sample data of anti-loose marks of all the fasteners;
s2, training a classifier with a three-layer cascade structure by using the standard sample data obtained in S1, and packaging the trained classifier in an image data processing system;
s3, acquiring anti-loose identification images of all fasteners on the bogies to be inspected in the same model;
s4, preprocessing the anti-loose identification image obtained in the step S3 to obtain data used in the subsequent steps;
s5, putting the image preprocessing result of the step S4 into a classifier of the three-layer cascade structure for classification;
and S6, comparing the classification result in the step S5 with the corresponding sample parameter, thereby judging whether the anti-loose mark is qualified.
2. The method for detecting the anti-loosening mark of the fastener of the bogie according to claim 1, wherein the step of acquiring standard sample data in step S1 is as follows:
s11: selecting a bogie which is newly delivered or subjected to card control inspection to ensure that all images in the acquired standard sample data are clear and complete, and fixedly placing the specified bogie on a bogie positioning platform with a lifting function;
s12: teaching and photographing anti-loosening marks of the bogie by using a composite robot, performing model matching on photographed image data by using the standard fastener model data by using the image data processing system, determining the type and the model of each fastener on the bogie, and recording the position information of each fastener; recording a traveling route and an image acquisition point of the composite robot and position and posture information when an image is acquired, and simultaneously forming a teaching program corresponding to the specified bogie type;
s13: the image data processing system divides and identifies the shot image to acquire an image of the fastener anti-loosening mark.
3. The method for detecting the anti-loosening mark of the bogie fastener according to claim 2, wherein: in the step S3, in step S3,
s31: placing a bogie to be measured on the bogie positioning platform in the same mode and posture, and completing image acquisition by the composite robot in the same position and posture according to the teaching program of the corresponding type of bogie;
s32: and the image data processing unit divides and identifies the acquired image to acquire the anti-loosening identification image to be detected.
4. The method for detecting the anti-loosening mark of the bogie fastener according to claim 3, wherein: in step S3, after the composite robot travels to the taught image capturing point, the composite robot performs secondary positioning, and adjusts the position and posture of image capturing according to the secondary positioning result.
5. The method for detecting the anti-loosening mark of the bogie fastener according to claim 1, wherein: in step S2, the three-layer cascade structure classifier includes 8 sub-classifiers in total; wherein the first layer of classifiers comprises an HSV classifier and a color component classifier; the second layer classifier comprises a brightness classifier and a contrast classifier; the third layer of classifiers includes a length classifier, a width classifier, an area classifier, and a morphology classifier.
6. The method for detecting the anti-loose mark of the bogie fastener according to claim 1, wherein the image preprocessing in the step S4 comprises the following specific steps:
s41: filtering the anti-loose identification image data; carrying out white balance processing on the filtered image; then carrying out distortion correction processing on the image subjected to the white balance processing, and storing the image subjected to the distortion correction processing;
s42: converting the image after the distortion correction processing into an RGB color space image and storing the RGB color space image; decomposing the image of the RGB color space into a single-channel image, and respectively obtaining corresponding R, G and B values; and obtaining and storing a corresponding HSV color space image by using the values of R, G and B.
7. The method for detecting the anti-loosening mark of the bogie fastener according to claim 1, wherein: when the bogies with various specifications need to be detected, before detection, specification marks are pasted at the specified positions of the bogies to be detected, and the composite robot acquires the type number information of the bogies to be detected by collecting and identifying the specification marks; and selecting and executing a corresponding teaching program according to the bogie type number information composite robot to finish the image acquisition operation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111126078.4A CN113780464A (en) | 2021-09-26 | 2021-09-26 | Method for detecting anti-loose identification of bogie fastener |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111126078.4A CN113780464A (en) | 2021-09-26 | 2021-09-26 | Method for detecting anti-loose identification of bogie fastener |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113780464A true CN113780464A (en) | 2021-12-10 |
Family
ID=78853348
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111126078.4A Pending CN113780464A (en) | 2021-09-26 | 2021-09-26 | Method for detecting anti-loose identification of bogie fastener |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113780464A (en) |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102324038A (en) * | 2011-09-06 | 2012-01-18 | 北京林业大学 | A kind of floristics recognition methods based on digital picture |
| CN104723338A (en) * | 2013-12-20 | 2015-06-24 | 蛇目缝纫机工业株式会社 | Robot, robot control method and robot control program |
| CN106372667A (en) * | 2016-08-31 | 2017-02-01 | 西南交通大学 | Method for detecting adverse state of inclined sleeve part screws of high-speed train overhead line system |
| CN106971182A (en) * | 2017-02-06 | 2017-07-21 | 王兴照 | Embedded electric power relay pressing plate is thrown and moves back state Intelligent Identify device and implementation method |
| CN107679508A (en) * | 2017-10-17 | 2018-02-09 | 广州汽车集团股份有限公司 | Road traffic sign detection recognition methods, apparatus and system |
| CN108121996A (en) * | 2017-11-21 | 2018-06-05 | 武汉中元华电软件有限公司 | A kind of hard pressing plate state identification method of power screen cabinet based on machine vision |
| CN208206469U (en) * | 2018-06-05 | 2018-12-07 | 成都精工华耀科技有限公司 | A kind of fastener loosening detection device |
| CN109190456A (en) * | 2018-07-19 | 2019-01-11 | 中国人民解放军战略支援部队信息工程大学 | Pedestrian detection method is overlooked based on the multiple features fusion of converging channels feature and gray level co-occurrence matrixes |
| US20190095739A1 (en) * | 2017-09-27 | 2019-03-28 | Harbin Institute Of Technology | Adaptive Auto Meter Detection Method based on Character Segmentation and Cascade Classifier |
| CN110009602A (en) * | 2019-03-08 | 2019-07-12 | 唐山百川智能机器股份有限公司 | Detection method, system storage medium and the computer equipment of fastener locking mark |
| CN111368852A (en) * | 2018-12-26 | 2020-07-03 | 沈阳新松机器人自动化股份有限公司 | Article identification and pre-sorting system and method based on deep learning and robot |
-
2021
- 2021-09-26 CN CN202111126078.4A patent/CN113780464A/en active Pending
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102324038A (en) * | 2011-09-06 | 2012-01-18 | 北京林业大学 | A kind of floristics recognition methods based on digital picture |
| CN104723338A (en) * | 2013-12-20 | 2015-06-24 | 蛇目缝纫机工业株式会社 | Robot, robot control method and robot control program |
| CN106372667A (en) * | 2016-08-31 | 2017-02-01 | 西南交通大学 | Method for detecting adverse state of inclined sleeve part screws of high-speed train overhead line system |
| CN106971182A (en) * | 2017-02-06 | 2017-07-21 | 王兴照 | Embedded electric power relay pressing plate is thrown and moves back state Intelligent Identify device and implementation method |
| US20190095739A1 (en) * | 2017-09-27 | 2019-03-28 | Harbin Institute Of Technology | Adaptive Auto Meter Detection Method based on Character Segmentation and Cascade Classifier |
| CN107679508A (en) * | 2017-10-17 | 2018-02-09 | 广州汽车集团股份有限公司 | Road traffic sign detection recognition methods, apparatus and system |
| CN108121996A (en) * | 2017-11-21 | 2018-06-05 | 武汉中元华电软件有限公司 | A kind of hard pressing plate state identification method of power screen cabinet based on machine vision |
| CN208206469U (en) * | 2018-06-05 | 2018-12-07 | 成都精工华耀科技有限公司 | A kind of fastener loosening detection device |
| CN109190456A (en) * | 2018-07-19 | 2019-01-11 | 中国人民解放军战略支援部队信息工程大学 | Pedestrian detection method is overlooked based on the multiple features fusion of converging channels feature and gray level co-occurrence matrixes |
| CN111368852A (en) * | 2018-12-26 | 2020-07-03 | 沈阳新松机器人自动化股份有限公司 | Article identification and pre-sorting system and method based on deep learning and robot |
| CN110009602A (en) * | 2019-03-08 | 2019-07-12 | 唐山百川智能机器股份有限公司 | Detection method, system storage medium and the computer equipment of fastener locking mark |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111259892B (en) | Inspection method, inspection device, inspection equipment and inspection medium for state of indicator lamp | |
| CN106709436B (en) | Track traffic panoramic monitoring-oriented cross-camera suspicious pedestrian target tracking system | |
| CN107945184B (en) | Surface-mounted component detection method based on color image segmentation and gradient projection positioning | |
| CN110246132A (en) | Rail vehicle bolt looseness detection method and system | |
| CN113469966B (en) | Train bolt looseness detection method based on anti-loosening line identification | |
| CN105373135A (en) | Method and system for guiding airplane docking and identifying airplane type based on machine vision | |
| CN108304785A (en) | Road traffic sign detection based on self-built neural network and recognition methods | |
| CN105469069A (en) | Safety helmet video detection method for production line data acquisition terminal | |
| CN113822840A (en) | Vehicle bottom inspection system and method | |
| CN113516629A (en) | Intelligent detection system for TFDS passing operation | |
| CN114758322B (en) | Road quality detection system based on machine identification | |
| CN114037703B (en) | Subway valve state detection method based on two-dimensional positioning and three-dimensional attitude calculation | |
| CN106447699A (en) | High-speed rail overhead contact line equipment object detection and tracking method based on Kalman filtering | |
| CN114119591A (en) | Display screen picture quality detection method | |
| CN114581760B (en) | Equipment fault detection method and system for machine room inspection | |
| CN112634269A (en) | Rail vehicle body detection method | |
| CN111539421A (en) | Deep learning-based railway locomotive number identification method | |
| CN111223081A (en) | Part hole opening recognition and detection method and system based on deep learning | |
| CN113269234B (en) | Connecting piece assembly detection method and system based on target detection | |
| CN105068139B (en) | A kind of characterization processes of piston cooling nozzle installment state | |
| Shajahan et al. | Automated inspection of monopole tower using drones and computer vision | |
| CN113639685A (en) | Displacement detection method, device, equipment and storage medium | |
| CN113780464A (en) | Method for detecting anti-loose identification of bogie fastener | |
| Zong et al. | Traffic light detection based on multi-feature segmentation and online selecting scheme | |
| CN115857040A (en) | Dynamic visual detection device and method for foreign matters on locomotive roof |
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 |