CN115049623A - Device for analyzing pantograph contour through visual segmentation - Google Patents
Device for analyzing pantograph contour through visual segmentation Download PDFInfo
- Publication number
- CN115049623A CN115049623A CN202210701498.9A CN202210701498A CN115049623A CN 115049623 A CN115049623 A CN 115049623A CN 202210701498 A CN202210701498 A CN 202210701498A CN 115049623 A CN115049623 A CN 115049623A
- Authority
- CN
- China
- Prior art keywords
- pantograph
- image
- analyzing
- unet
- outline
- 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
Images
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
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
- G06T7/001—Industrial image inspection using an image reference approach
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/12—Edge-based segmentation
-
- 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
- G06T2207/10016—Video; Image sequence
-
- 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
- G06T2207/10048—Infrared 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
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- 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
-
- 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
- G06T2207/30236—Traffic on road, railway or crossing
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a device for analyzing the outline of a pantograph by visual segmentation, which comprises an image acquisition device and an image processing device, wherein the image acquisition device acquires an image and transmits the image to the image processing device for analysis processing; the image processing device adopts a HI3559A processing board, which utilizes UNET neural network to realize pixel-level segmentation of image elements collected by the image collecting device, so as to obtain accurate contour information. The invention adopts a domestic processor HI3559A, utilizes UNET neural network semantic segmentation, and realizes full scene element structurization: the pantograph, the contact wire, the rod piece, the insulator and the like, and all the elements are segmented at the pixel level to obtain accurate contour information, so that effective monitoring is carried out.
Description
Technical Field
The invention relates to the technical field of pantographs in rail transit industry, in particular to a device for visually segmenting and analyzing the outline of a pantograph.
Background
The pantograph is one of the most important structures in a high-speed railway system, and is connected with a contact network and a high-speed train power supply system through the pantograph, so that the normal operation of a high-speed train is guaranteed. Since the pantograph is installed on the top of a train, the pantograph is in contact with a static pantograph-catenary wire for a long time in the high-speed running process of the train, so that the problems of abrasion of the pantograph, abnormal pulling value of the contact wire, intrusion of foreign matters and the like can occur.
The current scheme is to perform regular maintenance and repair on a high-speed train and manually find out problems. However, the pantograph condition cannot be effectively detected during the train operation, and the pantograph condition cannot be timely detected when a problem occurs. Accordingly, one skilled in the art provides a device for analyzing the contour of a pantograph by visual segmentation to solve the above-mentioned problems.
Disclosure of Invention
In order to solve the technical problem, the invention provides a device for analyzing the outline of a pantograph by visual segmentation, which comprises an image acquisition device and an image processing device, wherein the image acquisition device acquires an image and transmits the image to the image processing device for analysis processing;
the image processing device adopts a HI3559A processing board, which utilizes UNET neural network to realize pixel-level segmentation of image elements collected by the image collecting device, so as to obtain accurate contour information.
Preferably: the image acquisition equipment adopts a high-definition visible light camera and a thermal infrared camera, and the two cameras are matched for use, so that the image is more accurate.
Preferably: the image processing apparatus is externally powered.
Preferably: the image processing equipment has a network communication function and can be in network connection with external intelligent equipment.
Preferably: the system processing flow of the device is as follows:
(1) acquiring images, namely acquiring videos of train operation in different scenes by using a high-definition visible light camera and a thermal infrared camera;
(2) the collected video is transmitted to image processing equipment, the UNET segmentation network of a HI3559A processing board is used for processing the video, scene analysis is carried out on the video, and the whole scene elements in the video are structured;
(3) defect analysis, comparing different structural elements with normal state, and judging whether there is defect;
if yes, reporting an alarm, and continuously acquiring and processing the video;
if not, selecting whether to quit the analysis system: if yes, directly ending, otherwise, continuing the initial image acquisition.
Preferably: the video can be directly stored in a JPG picture format during video acquisition.
Preferably: the full scene element structuralization refers to the process of forming all parts of a pantograph in a video into different modules, wherein the modules comprise the pantograph, a contact wire, a rod piece and an insulator.
The UNET training process in the system processing process is as follows:
marking the contour of the pantograph, the contour of a contact wire and the contour of a point area at a special position of the pantograph by using special image marking software to obtain an original data set and a marked label data set, putting the original data set and the marked label data set in a UNET network, modifying network training parameters of the UNET network, and starting training a network model after the adjustment is finished;
and then converting the pantograph UNET model into an HI3559A model, converting the trained UNET neural network weight into a wk file by using an NNIE _ mapper tool of NNIE after the UNET neural network model is trained, loading the wk file into RuyiStudio simulation software, and transplanting the wk weight file to an HI3559A processing board platform for segmenting and extracting the pantograph outline after the test is passed.
Preferably: the simulation software used RuyiStaudi.
The invention has the technical effects and advantages that:
the invention adopts a domestic processor HI3559A, utilizes UNET neural network semantic segmentation, and realizes full scene element structurization: the pantograph, the contact wire, the rod piece, the insulator and the like, and all the elements are segmented at the pixel level to obtain accurate contour information, so that effective monitoring is carried out.
Drawings
Fig. 1 is a schematic diagram of an apparatus for analyzing a pantograph profile by visual segmentation according to an embodiment of the present disclosure;
fig. 2 is a flowchart of a processing of an apparatus for visually segmenting and analyzing a pantograph profile according to an embodiment of the present application;
fig. 3 is a flowchart of UNET training in an apparatus for visually segmenting and analyzing pantograph contours according to an embodiment of the present application;
fig. 4 is a UNET model diagram of an apparatus for visually segmenting and analyzing pantograph profile according to an embodiment of the present application;
FIG. 5 is a diagram illustrating the indoor operation effect of an apparatus for visually segmenting and analyzing the pantograph profile according to an embodiment of the present application;
fig. 6 is a diagram illustrating an effect of outdoor operation of an apparatus for visually segmenting and analyzing a pantograph profile according to an embodiment of the present application.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and specific embodiments. The embodiments of the present invention have been presented for purposes of illustration and description, and are not intended to be exhaustive or limited to the invention in the form disclosed. Many modifications and variations will be apparent to those of ordinary skill in the art. The embodiment was chosen and described in order to best explain the principles of the invention and the practical application, and to enable others of ordinary skill in the art to understand the invention for various embodiments with various modifications as are suited to the particular use contemplated.
Examples
Referring to fig. 1 to 3, in the present embodiment, a device for visually segmenting and analyzing a pantograph profile is provided, including an image acquisition device and an image processing device, where the image acquisition device acquires an image and transmits the image to the image processing device for analysis and processing;
the image acquisition equipment adopts a high-definition visible light camera and a thermal infrared camera, and the two cameras are matched for use, so that the image is more accurate;
the image processing equipment adopts a HI3559A processing board, and pixel-level segmentation is realized on image elements acquired by the image acquisition equipment by utilizing a UNET network to obtain accurate contour information;
the image processing equipment is supplied with power through the outside, and the image processing equipment has a network communication function and can be connected with an external intelligent device through a network.
The image processing equipment of HI3559A is used for processing videos of the image acquisition equipment, an internal accelerator is used for accelerating the hardware of the UNET network, each module of the pantograph is divided, and the defects are analyzed to perform alarm processing;
the system processing flow of the device is as follows:
(1) the method comprises the steps of collecting images, namely collecting videos of train running in different scenes by using a high-definition visible light camera and a thermal infrared camera, wherein the videos can be directly stored in a JPG picture format during collection, so that a UNET model can be conveniently trained;
(2) the collected video is transmitted to an image processing device, the UNET of a HI3559A processing board is used for dividing the network processing video, the scene analysis is carried out on the video, and the whole scene elements in the video are structured, namely, each part of a pantograph in the video is divided into different modules, wherein the modules comprise the pantograph, a contact wire, a rod piece, an insulator and the like;
(3) defect analysis, comparing different structural elements with normal state, and judging whether there is defect;
if yes, reporting an alarm, and continuously acquiring and processing the video;
if not, selecting whether to quit the analysis system: if yes, directly ending, otherwise, continuing the initial image acquisition.
The UNET training process in the system processing process is as follows:
marking the contour of the pantograph, the contour of a contact wire and the contour of a point area at a special position of the pantograph by using special image marking software to obtain an original data set and a marked label data set, putting the original data set and the marked label data set in a UNET network, modifying network training parameters of the UNET network, and starting training a network model after the adjustment is finished;
and then converting the pantograph UNET model into an HI3559A model, converting the trained UNET neural network weight into a wk file by using an NNIE _ mapper tool of NNIE after the UNET neural network model is trained, loading the wk file into RuyiStudio simulation software, and transplanting the wk weight file to an HI3559A processing board platform for segmenting and extracting the pantograph outline after the test is passed.
Referring to fig. 4, the advantage of using the UNET model in the present invention: the first half of the UNET network, the left side, is used for feature extraction and the second half, the right side, is upsampled. UNET employs a completely different feature fusion approach: and splicing, namely Concat, wherein the UNET is used for splicing the features together in channel dimension to form thicker features, and meanwhile, the UNET network is very simple, has high calculation speed and good extraction effect compared with other semantic segmentation neural network models.
Referring to fig. 5, which is an effect diagram of the indoor operation of the device, it can be seen from the test effect that the outline of the whole pantograph can be well divided in the operating state, the outlines of the upper and lower wires of the catenary can be always divided and identified, and 7 key position areas at the left and right of the pantograph can be well divided and extracted;
referring to fig. 6, which is an effect diagram of the outdoor operation of the device, it can be seen from the test effect that the outline of the entire pantograph can be perfectly divided and extracted in the high-speed operation state, the outlines of the upper and lower wires of the contact net can be always divided and extracted in the open air, and the areas of 7 key positions at the left and right of the pantograph can be well divided and extracted.
The method adopts a high-definition visible light camera and a thermal infrared camera to obtain videos, then uses image processing equipment adopting a domestic processor HI3559A to perform processing and analysis, utilizes UNET neural network semantic segmentation, and realizes full scene element structurization: the pantograph, the contact wire, the rod piece, the insulator and the like, and all the elements are segmented at the pixel level to obtain accurate contour information, so that effective monitoring is carried out.
It is to be understood that the described embodiments are merely a few embodiments of the invention, and not all embodiments. All other embodiments, which can be derived by one of ordinary skill in the art and related arts based on the embodiments of the present invention without any creative effort, shall fall within the protection scope of the present invention. Structures, devices, and methods of operation not specifically described or illustrated herein are generally practiced in the art without specific recitation or limitation.
Claims (9)
1. The device for analyzing the contour of the pantograph by visual segmentation is characterized by comprising an image acquisition device and an image processing device, wherein the image acquisition device acquires an image and transmits the image to the image processing device for analysis processing;
the image processing device adopts a HI3559A processing board, which utilizes UNET neural network to realize pixel-level segmentation of image elements collected by the image collecting device, so as to obtain accurate contour information.
2. The apparatus for visually segmenting and analyzing pantograph profile according to claim 1, wherein said image capturing device employs a high-definition visible light camera and a thermal infrared camera, and the two cameras are used in cooperation.
3. The device for visually segmenting and analyzing the outline of the pantograph according to claim 1, wherein the image processing equipment is powered externally.
4. The apparatus for visually segmenting and analyzing pantograph contour according to claim 1, wherein said image processing device has a network communication function, and is capable of being connected to an external intelligent device through a network.
5. The device for visually segmenting and analyzing the outline of the pantograph according to claim 1, wherein the system processing flow of the device is as follows:
(1) acquiring images, namely acquiring videos of train operation in different scenes by using a high-definition visible light camera and a thermal infrared camera;
(2) the collected video is transmitted to image processing equipment, the UNET segmentation network of a HI3559A processing board is used for processing the video, scene analysis is carried out on the video, and the whole scene elements in the video are structured;
(3) defect analysis, comparing different structural elements with normal state, and judging whether there is defect;
if yes, reporting an alarm, and continuously acquiring and processing the video;
if not, selecting whether to quit the analysis system: if yes, directly ending, otherwise, continuing the initial image acquisition.
6. The device for visually segmenting and analyzing the outline of the pantograph according to claim 5, wherein the video is directly stored in a JPG picture format during video acquisition.
7. The apparatus for visual segmentation and analysis of pantograph profile according to claim 5, wherein said full scene element structuring is to divide the pantograph's parts in video into different modules, including pantograph, contact wire, rod, insulator.
8. The device for visually segmenting and analyzing the outline of the pantograph according to claim 5, wherein the UNET training procedure in the system processing procedure is as follows:
marking out the pantograph outline, the outline of a contact wire and the outline of a point area at a special position of the pantograph by using special image marking software of the pantograph image to obtain an original data set and a marked label data set, putting the original data set and the marked label data set in a UNET network, modifying network training parameters of the UNET, and starting training a network model after the adjustment is finished;
and then converting the pantograph UNET model into an HI3559A model, converting the trained UNET neural network weight into a wk file by using an NNIE _ mapper tool of NNIE after the UNET neural network model is trained, loading the wk file into RuyiStudio simulation software, and transplanting the wk weight file to an HI3559A processing board platform for segmenting and extracting the pantograph outline after the test is passed.
9. The apparatus for visually segmenting and analyzing the outline of a pantograph according to claim 8, wherein the simulation software employs RuyiStudio.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210701498.9A CN115049623A (en) | 2022-06-20 | 2022-06-20 | Device for analyzing pantograph contour through visual segmentation |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210701498.9A CN115049623A (en) | 2022-06-20 | 2022-06-20 | Device for analyzing pantograph contour through visual segmentation |
Publications (1)
Publication Number | Publication Date |
---|---|
CN115049623A true CN115049623A (en) | 2022-09-13 |
Family
ID=83163540
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210701498.9A Pending CN115049623A (en) | 2022-06-20 | 2022-06-20 | Device for analyzing pantograph contour through visual segmentation |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115049623A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117408957A (en) * | 2023-10-13 | 2024-01-16 | 中车工业研究院有限公司 | Non-contact bow net deflection state monitoring method and device |
-
2022
- 2022-06-20 CN CN202210701498.9A patent/CN115049623A/en active Pending
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117408957A (en) * | 2023-10-13 | 2024-01-16 | 中车工业研究院有限公司 | Non-contact bow net deflection state monitoring method and device |
CN117408957B (en) * | 2023-10-13 | 2024-06-11 | 中车工业研究院有限公司 | Non-contact bow net deflection state monitoring method and device |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110826538B (en) | Abnormal off-duty identification system for electric power business hall | |
CN109879005B (en) | Belt tearing detection device and method | |
CN109271872B (en) | Device and method for judging on-off state and diagnosing fault of high-voltage isolating switch | |
AU2008286243B2 (en) | Pantograph damage and wear monitoring system | |
CN102759347B (en) | Online in-process quality control device and method for high-speed rail contact networks and composed high-speed rail contact network detection system thereof | |
CN113947731B (en) | Foreign matter identification method and system based on contact net safety inspection | |
CN112183313A (en) | SlowFast-based power operation field action identification method | |
CN110493574B (en) | Security monitoring visualization system based on streaming media and AI technology | |
CN111561967A (en) | Real-time online detection method and system for pantograph-catenary operation state | |
WO2024027009A1 (en) | Infrared thermal imaging defect inspection method and apparatus for substation insulator | |
CN106920240A (en) | A kind of insulator identification and method for diagnosing faults based on infrared image | |
CN115049623A (en) | Device for analyzing pantograph contour through visual segmentation | |
CN113487821A (en) | Power equipment foreign matter intrusion identification system and method based on machine vision | |
CN112508911A (en) | Rail joint touch net suspension support component crack detection system based on inspection robot and detection method thereof | |
KR102169211B1 (en) | apparatus and method for automatically detecting bird's cast | |
CN113673614B (en) | Metro tunnel foreign matter intrusion detection device and method based on machine vision | |
CN106803254A (en) | The picture automatic testing method of insulator defect analysis in a kind of transmission line of electricity | |
CN112381031B (en) | Real-time online pantograph and horn detection method based on convolutional neural network | |
CN111597939B (en) | High-speed rail line nest defect detection method based on deep learning | |
CN109934172B (en) | GPS-free full-operation line fault visual detection and positioning method for high-speed train pantograph | |
CN117372677A (en) | Method for detecting health state of cotter pin of fastener of high-speed railway overhead contact system | |
CN111160224B (en) | High-speed rail contact net foreign matter detection system and method based on FPGA and horizon line segmentation | |
CN114724091A (en) | Method and device for identifying foreign matters on transmission line wire | |
CN116124771A (en) | Method for identifying loosening faults of central anchoring rope of contact net | |
CN112528827B (en) | Automatic detection method for crack loss of high-speed rail contact network power supply equipment |
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 |