CN108090912A - Track detection method and system based on image recognition - Google Patents

Abstract

The invention belongs to the technical field of track detection, and particularly relates to a track detection method and system based on image recognition, which comprises the following steps of S1, acquiring fastener images, steel rail light band images, track bed plate images and mileage data; s2, analyzing and identifying the fastener image, the steel rail light band image and the track bed board image through image processing, and obtaining fastener data, steel rail light band data and track bed board data by combining mileage data; s3, respectively storing the fastener data, the steel rail light band data and the track bed board data into a steel rail light band data chart library, a fastener data chart library and a track bed board data chart library of the database; and S4, comparing the data in each data chart database with the pre-stored track normal data, drawing and analyzing to generate an analysis chart and an analysis report. According to the method, the analysis chart and the analysis report form of the stability of the track are obtained through the detection of the track fastener, the detection of the steel rail light belt and the detection of the crack of the track bed plate, and the method is efficient, accurate and intelligent.

Description

A track detection method and system based on image recognition

technical field

The invention belongs to the technical field of track detection, and in particular relates to a track detection method and system based on image recognition.

Background technique

With the development of high-speed railways, trains run faster and faster, and the requirements for track stability continue to increase. However, affected by factors such as railway construction, geographical environment, and train operation, it is inevitable that there will be problems with the stability of the track. Since the stability of the track will directly affect the safety of the train, the detection of the track stability is related to the safety of people's lives and property. The detection of track stability includes the detection of rail fasteners, rail light strips, and track bed boards.

The traditional detection method is to detect by manual observation, which has potential safety hazards and is not accurate. With the development of science and technology, there are currently ballastless track crack detection equipment on the market, which can be used to detect single cracks. However, due to the small detection area, it is necessary to detect specified cracks, and it is impossible to check the cracks of the track bed (caulk) , not suitable for high-speed rail ballastless track testing. At present, there is no equipment for ballastless track fastener specification (label) identification.

Contents of the invention

Aiming at the defects in the prior art, the present invention provides a track detection method and system based on image recognition, through the detection of track fasteners, the detection of rail light strips and the detection of track bed cracks, the analysis of track stability is obtained Graphs and analysis reports are efficient, accurate and intelligent.

In order to achieve the above object, a kind of track detection method based on image recognition provided by the invention comprises the following steps:

S1, respectively collect fastener images, rail light belt images, track bed board images and mileage data;

S2, analyzing and identifying the fastener image, the rail light belt image and the ballast bed plate image respectively through image processing, and combining the mileage data to obtain fastener data, rail light belt data and ballast bed plate data respectively;

S3, storing the fastener data, the rail light belt data and the track bed board data into the rail light belt data chart library, fastener data chart library and ballast bed board data chart library of the database respectively;

S4, compare, draw and analyze the data in the fastener data chart database, the rail light belt data chart database and the ballast bed plate data chart database with the pre-stored track normal data, and generate analysis charts and analysis reports.

Preferably, in the S2, the image of the fastener is analyzed and identified through image processing, combined with the mileage data, the fastener data is specifically:

Carry out backing plate edge recognition on the fastener image, the edge recognition includes upper and lower edge recognition and left and right edge recognition, extract the longitudinal centerline by recognizing the left and right edges, and obtain the longitudinal centerline between the upper and lower edges by recognizing the upper and lower edges distance, and then get backing plate thickness data;

Carrying out baffle image recognition on the fastener image, further identifying the baffle model in the baffle image, and then obtaining baffle model data;

Combining the mileage data to obtain fastener data, the fastener data includes the mileage data, the backing plate thickness data and the baffle model data.

Preferably, in the S2, the image of the rail light band is analyzed and identified through image processing, combined with the mileage data, the rail light band data is specifically:

Performing rail edge recognition on the rail light strip image to obtain a rail surface image of a preset length;

Carry out light band identification at every set distance on the surface image of the rail to obtain N sets of preliminary light band width values and distance values from the center of the light band to the edge, and then sum the average value to obtain the final light band width data and the center of the light band Distance data from edge;

Combined with the mileage data, rail light strip data is obtained, and the rail light strip data includes the mileage data, the light strip width data, and the distance data between the center of the light strip and the edge.

Preferably, in S2, image processing is used to analyze and identify the image of the track bed slab, and combined with the mileage data, the track slab bed data is specifically:

Carrying out edge recognition on the image of the ballast bed slab, and extracting the crack image of the caulking glue;

Recognizing the crack image of the caulking glue to obtain crack length data and crack maximum width data;

Combined with the geographical mileage data, the track slab bed data is obtained, and the track slab bed data includes the mileage data, the crack length data and the crack maximum width data.

Preferably, before acquiring the images of fasteners, rail light belts, and track bed boards in S1, it is also included to determine whether the position of the fastener is reached through the switch signal output by the proximity switch, and if so, execute step S1.

Preferably, the analysis diagram includes a fastener analysis diagram, a rail light belt analysis diagram and a ballast bed slab analysis diagram; and the analysis report includes a fastener analysis report, a rail light belt analysis report and a ballast bed slab analysis report.

A track detection system based on image recognition, comprising a detection vehicle moving on the track, the detection vehicle is provided with a plurality of image acquisition devices fixed by a plurality of adjustable support frames, and the detection vehicle is also equipped with a mileage meter, a proximity switch and a controller, the image acquisition device, the odometer, and the proximity switch are electrically connected to the controller respectively;

The odometer is used to collect mileage data;

The proximity switch is used to detect whether it is close to a metal substance and output a switch signal;

The image acquisition device includes a first image acquisition device, a second image acquisition device and a third image acquisition device located at different positions on the inspection vehicle;

The first image acquisition device is used to acquire fastener images;

The second image acquisition device is used to acquire rail light strip images;

The third image acquisition device is used to acquire the image of the track bed;

The controller includes:

An image acquisition module, configured to control the image acquisition device to acquire the fastener image, the rail light belt image and the ballast bed plate image through a preset time interval or a switch signal output by the proximity switch;

The image processing module is used to analyze and identify the fastener image, the rail light strip image and the ballast bed plate image respectively through image processing, and combine the mileage data to obtain fastener data, rail light strip data and Road bed data;

The data storage module is used to store the fastener data, the rail light strip data and the track bed plate data into the rail light strip data chart library, fastener data chart library and track bed plate data chart library of the database respectively;

The result analysis module is used to compare, draw and analyze the data in the fastener data chart library, the rail light strip data chart library and the ballast bed plate data chart library with the pre-stored normal track data, generate analysis charts and analyze report.

Preferably, the controller further includes a wireless transmission module, the wireless transmission module is used to send the fastener data, the rail light belt data, the ballast bed board data, the analysis diagram and the analysis report to to the remote device.

The beneficial effects of the invention are: through the detection of rail fasteners, rail light strips and ballast slab cracks, an analysis diagram and analysis report of track stability can be obtained, which is efficient, accurate and intelligent.

Description of drawings

In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings that need to be used in the description of the specific embodiments or the prior art. Throughout the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, elements or parts are not necessarily drawn in actual scale.

Fig. 1 is the flowchart of the track detection method based on image recognition in the present embodiment;

Fig. 2 is a block diagram of the electrical connection structure of the track detection system based on image recognition in the present embodiment;

Fig. 3 is the track detection system based on image recognition in the present embodiment;

Reference signs:

1-Testing car, 2-Adjustable support frame, 3-Image acquisition device, 4-Controller, 5-Track

Detailed ways

Embodiments of the technical solutions of the present invention will be described in detail below in conjunction with the accompanying drawings. The following examples are only used to illustrate the technical solutions of the present invention more clearly, and therefore are only examples, rather than limiting the protection scope of the present invention. It should be noted that, unless otherwise specified, the technical terms or scientific terms used in this application shall have the usual meanings understood by those skilled in the art to which the present invention belongs.

Embodiment one:

Embodiment one provides a kind of track detection method based on image recognition, as shown in Figure 1, comprises the following steps:

S1, respectively collect fastener images, rail light belt images, track bed plate (caulking glue) images and mileage data (before collecting images, the image acquisition equipment must be calibrated in advance, that is, use the image acquisition equipment to shoot the calibration plate, so as to calibrate the image corresponding to pixel size);

Specifically, the detection vehicle using this method moves on the track, judges when to collect the fastener image according to the detection result of the proximity switch, and controls the image acquisition device through the preset time interval or the switch signal output by the proximity switch Collect images of fasteners, rail light strips, ballast bed panels (caulking glue) and mileage data. Because the fasteners are usually made of metal materials, the detection of the metal material by the proximity switch is used to judge the position of the fastener. During the movement of the detection vehicle, the switch signal output by the proximity switch is used to judge whether it has reached the position of the fastener. If so, then Execute image acquisition in step S1.

S2, analyze and identify the fastener image, the rail light belt image and the track bed plate (caulk) image respectively through image processing, and combine the mileage data to obtain fastener data and rail light belt data respectively And the bed board (caulk) data;

Wherein, in the S2, the fastener image is analyzed and identified through image processing, and the fastener data is obtained in combination with the mileage data as follows:

Carry out backing plate edge recognition on the fastener image, the edge recognition includes upper and lower edge recognition and left and right edge recognition, extract the longitudinal centerline by recognizing the left and right edges, and obtain the longitudinal centerline between the upper and lower edges by recognizing the upper and lower edges distance, and then get backing plate thickness data;

Carrying out baffle image recognition on the fastener image, further identifying the baffle model in the baffle image, and then obtaining baffle model data;

Combining the mileage data to obtain fastener data, the fastener data includes the mileage data, the backing plate thickness data and the baffle model data.

Wherein, in the S2, the image of the rail light strip is analyzed and identified through image processing, and the rail light strip data is obtained in combination with the mileage data as follows:

Performing rail edge recognition on the rail light strip image to obtain a rail surface image of a preset length;

Carry out light band identification at every set distance on the surface image of the rail to obtain N sets of preliminary light band width values and distance values from the center of the light band to the edge, and then sum the average value to obtain the final light band width data and the center of the light band Distance data from edge;

Combined with the mileage data, rail light strip data is obtained, and the rail light strip data includes the mileage data, the light strip width data, and the distance data between the center of the light strip and the edge.

Wherein, in the S2, the image of the track bed (caulk) is analyzed and identified through image processing, combined with the mileage data, the track bed data is specifically:

Carry out edge recognition to the image of the track bed plate (caulking glue), and extract the crack image of the caulking glue;

Recognizing the crack image of the caulking glue to obtain crack length data and crack maximum width data;

Combined with the geographical mileage data, the track slab bed data is obtained, and the track slab bed data includes the mileage data, the crack length data and the crack maximum width data.

Specifically, the accuracy of the detection data in this embodiment is greater than 0.5 mm.

S3, storing the fastener data, the rail light belt data and the track bed board (caulk) data into the rail light belt data graph library, fastener data graph library and track bed board (caulk) data respectively in the database Glue) data chart library;

S4, comparing the data in the fastener data chart library, the rail light strip data chart library and the track bed plate (caulking glue) data chart library with the pre-stored normal data of the track, and generating an analysis chart and Analyze reports. The analysis diagram includes a fastener analysis diagram, a rail light strip analysis diagram and a ballast bed board (caulk) analysis diagram; the analysis report includes a fastener analysis report, a rail light belt analysis report and a ballast bed board (caulk glue) Analyze reports.

Specifically, the structure of several inspections is recorded in order in the analysis report. The fastener analysis report includes multiple categories: serial number, mileage, backing plate thickness, baffle type, etc.; the rail light belt analysis report includes multiple categories: serial number, Mileage, width of light band, distance from center of light band to edge, etc.; said road bed board (caulk) analysis report includes multiple categories: serial number, mileage, crack length, crack maximum width, etc. Generating analysis diagrams and analysis reports can intuitively show the stability of the track, which is convenient for the staff to view and find problems.

Embodiment two:

Embodiment 2 provides a track detection system based on image recognition, applicable to the track detection method of Embodiment 1, as shown in Figure 2 and Figure 3, including a detection vehicle 1 moving on the track, and the detection vehicle 1 is equipped with There are a plurality of image acquisition devices 3 fixed by a plurality of adjustable support frames 2, and the detection vehicle 1 is also provided with an odometer, a proximity switch and a controller 4, the image acquisition device 3, the odometer, the The proximity switch is electrically connected with the controller 4 respectively;

The odometer is used to collect mileage data; other devices capable of collecting mileage data, such as GPS locators, can also be used.

The proximity switch is used to detect whether it is close to a metal object and output a switch signal; other devices that can determine whether it is close to a metal fastener, such as a metal detector, a metal sensor, etc., can also be used.

The image acquisition device 3 includes a first image acquisition device, a second image acquisition device and a third image acquisition device located at different positions on the inspection vehicle 1;

The first image acquisition device is used to acquire fastener images;

The second image acquisition device is used to acquire rail light belt images;

The third image acquisition device is used to acquire the image of the road bed board (caulking glue);

Each image acquisition device 3 in this embodiment may include one or more cameras, each camera is fixed on the inspection vehicle through the adjustable support frame, and the shooting angle and shooting direction of the camera are adjusted through the adjustable support frame 2 .

Described controller 4 comprises:

The image acquisition module is used to control the image acquisition device to acquire the fastener image, the rail light belt image and the ballast bed (caulking) through a preset time interval or a switch signal output by the proximity switch glue) image;

The image processing module is used to analyze and identify the fastener image, the rail light belt image and the ballast bed plate (caulk) image respectively through image processing, and combine the mileage data to obtain fastener data, Rail light strip data and ballast bed plate (caulk) data;

The data storage module is used to store the fastener data, the rail light strip data and the track bed plate (caulk) data into the rail light strip data chart library, the fastener data chart library and the track track data chart library respectively in the database. Bed board (caulk) data chart library;

The result analysis module is used to compare and analyze the data in the fastener data chart library, the rail light belt data chart library and the ballast bed plate (caulking glue) data chart library with the pre-stored normal track data, Generate analysis graphs and analysis reports.

A wireless transmission module, the wireless transmission module is used to send the fastener data, the rail light belt data, the ballast bed plate (caulking glue) data, the analysis diagram and the analysis report to the remote device .

In this embodiment, the detection vehicle moves forward on the track, and collects the rail light belt image, track bed plate (caulk) image and mileage data according to the preset time interval, and collects the fastener every time it moves to the fastener position. Image: After the controller identifies, analyzes and processes the collected images, it obtains relevant data and generates analysis charts and analysis reports, and then sends the relevant data, analysis charts and analysis reports to a remote server for easy data storage and later viewing. The wireless transmission module in this embodiment includes a wifi module, a 3G/4G module and the like.

Comprehensive embodiment 1 and embodiment 2, method and system of the present invention obtain the analysis diagram of track stability and The analysis report is efficient, accurate and intelligent, and intuitively shows the stability of the track.

Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present invention, rather than limiting them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: It is still possible to modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the technical solutions of the various embodiments of the present invention. All of them should be covered by the scope of the claims and description of the present invention.

Claims (8)

1. a track detection method based on image recognition, is characterized in that, comprises the following steps: S1, respectively collect fastener images, rail light belt images, track bed board images and mileage data; S2, analyzing and identifying the fastener image, the rail light belt image and the ballast bed plate image respectively through image processing, and combining the mileage data to obtain fastener data, rail light belt data and ballast bed plate data respectively; S3, storing the fastener data, the rail light belt data and the track bed board data into the rail light belt data chart library, fastener data chart library and ballast bed board data chart library of the database respectively; S4, compare, draw and analyze the data in the fastener data chart database, the rail light belt data chart database and the ballast bed plate data chart database with the pre-stored track normal data, and generate analysis charts and analysis reports. 2. A kind of track detection method based on image recognition according to claim 1, it is characterized in that, in described S2, analyze and identify described fastener image by image processing, combine described mileage data, obtain fastener data Specifically: Carry out backing plate edge recognition on the fastener image, the edge recognition includes upper and lower edge recognition and left and right edge recognition, extract the longitudinal centerline by recognizing the left and right edges, and obtain the longitudinal centerline between the upper and lower edges by recognizing the upper and lower edges distance, and then get backing plate thickness data; Carrying out baffle image recognition on the fastener image, further identifying the baffle model in the baffle image, and then obtaining baffle model data; Combining the mileage data to obtain fastener data, the fastener data includes the mileage data, the backing plate thickness data and the baffle model data. 3. A kind of track detection method based on image recognition according to claim 1, it is characterized in that, in described S2, analyze and identify described rail light band image through image processing, combine described mileage data, obtain rail light band image The band data is specifically: Performing rail edge recognition on the rail light strip image to obtain a rail surface image of a preset length; Carry out light band identification at every set distance on the surface image of the rail to obtain N sets of preliminary light band width values and distance values from the center of the light band to the edge, and then sum the average value to obtain the final light band width data and the center of the light band Distance data from edge; Combined with the mileage data, rail light strip data is obtained, and the rail light strip data includes the mileage data, the light strip width data, and the distance data between the center of the light strip and the edge. 4. A kind of track detection method based on image recognition according to claim 1, it is characterized in that, in described S2, analyze and identify described track bed slab image through image processing, combine with described mileage data, obtain track slab bed data Specifically: Carrying out edge recognition on the image of the ballast bed slab, and extracting the crack image of the caulking glue; Recognizing the crack image of the caulking glue to obtain crack length data and crack maximum width data; Combined with the geographical mileage data, the track slab bed data is obtained, and the track slab bed data includes the mileage data, the crack length data and the crack maximum width data. 5. A kind of track detection method based on image recognition according to claim 1, it is characterized in that, before collecting fastener image, rail light strip image, ballast bed plate image respectively in described S1, also include, output by proximity switch It is judged whether the switch signal reaches the position of the fastener, and if so, step S1 is executed. 6. A kind of track detection method based on image recognition according to claim 1, is characterized in that, described analysis diagram comprises fastener analysis diagram, rail light strip analysis diagram and and ballast bed board analysis diagram; Described analysis report includes Fastener analysis report, rail light strip analysis report and ballast bed slab analysis report. 7. A track detection system based on image recognition, applicable to a kind of track detection method based on image recognition according to claim 1, characterized in that, it includes a detection vehicle moving on the track, and the detection vehicle is equipped with Through a plurality of image acquisition devices fixed by a plurality of adjustable support frames, the detection vehicle is also provided with an odometer, a proximity switch and a controller, and the image acquisition device, the odometer, and the proximity switch are respectively connected to the The controller is electrically connected; The odometer is used to collect mileage data; The proximity switch is used to detect whether it is close to a metal substance and output a switch signal; The image acquisition device includes a first image acquisition device, a second image acquisition device and a third image acquisition device located at different positions on the inspection vehicle; The first image acquisition device is used to acquire fastener images; The second image acquisition device is used to acquire rail light strip images; The third image acquisition device is used to acquire the image of the track bed; The controller includes: An image acquisition module, configured to control the image acquisition device to acquire the fastener image, the rail light belt image and the ballast bed plate image through a preset time interval or a switch signal output by the proximity switch; The image processing module is used to analyze and identify the fastener image, the rail light strip image and the ballast bed plate image respectively through image processing, and combine the mileage data to obtain fastener data, rail light strip data and Road bed data; The data storage module is used to store the fastener data, the rail light strip data and the track bed plate data into the rail light strip data chart library, fastener data chart library and track bed plate data chart library of the database respectively; The result analysis module is used to compare, draw and analyze the data in the fastener data chart library, the rail light strip data chart library and the ballast bed plate data chart library with the pre-stored normal track data, generate analysis charts and analyze report. 8. A kind of track detection system based on image recognition according to claim 7, it is characterized in that, described controller also comprises wireless transmission module, and described wireless transmission module is used for described fastener data, described rail The light strip data, the ballast bed data, the analysis graph and the analysis report are sent to the remote device.