KR102330188B1 - Mobile rail facility inspection apparatus using hyperspectral camera - Google Patents

Abstract

The present invention relates to a mobile railway facility inspection device using a hyperspectral camera, which is to check the condition of the railway through the image taken while driving along the railway, classifies rails, sleepers, and fasteners constituting the railway through image processing obtained through the hyperspectral camera, and performs precise analysis and acquires accurate position coordinates according to movement so that inspection and management are performed. The mobile railway facility inspection device includes a main body (100); a position calculating unit (110); a photographing unit (120); an image processing unit (140); and a control unit (150).

Description

{Mobile rail facility inspection apparatus using hyperspectral camera}

The present invention relates to a railway facility inspection device, and in detail, the rails, sleepers, and fasteners constituting the railway through image processing obtained through a hyperspectral camera while driving along the railway and inspecting the state of the railway through captured images. It relates to a mobile railway facility inspection device using a hyperspectral camera that performs detailed analysis by dividing and acquires accurate position coordinates according to movement so that inspection and management are performed.

1 is a plan view showing the structure of a conventional railway.

Railroad, which is one of the important means of transport for transporting people and goods, includes sleepers 202 arranged at regular intervals on a ballast 201 made of gravel or concrete, and rails 203 installed in parallel and perpendicular to the sleepers 202. ), and the rail 203 and the sleeper 202 are moved along the rail consisting of a fastener 204 connecting the.

These railroads cause various types of damage such as abrasion, cracks, and peeling of each component over time as the railroad car runs and time passes. In particular, due to its characteristics, even a small defect in the railway can lead to a major accident, so safety management of the railway is very important.

In particular, due to the characteristics of high-speed trains operating recently, regular inspections are essential to early detection and corrective measures for railway defects.

For this reason, the facility manager patrols along the track for a set time when railway operation is suspended, and inspections are carried out at regular intervals such as regular inspection, detailed inspection, and precise safety diagnosis. Not only this time is required, but there is a limit to the precise inspection of the damaged area, and difficulties and risks follow due to the problem of time and place when people are rare.

In order to solve this problem, a track patrol inspection vehicle has been proposed that can check whether the railway is damaged by visual or video while moving the rail with inspection personnel on board or equipped with a camera. There is still a limit to precisely checking the condition of the railway through which this is inevitable through the naked eye or video.

Republic of Korea Patent Registration No. 10-1833484 (2018.02.22)

The present invention was created to solve the above problems, and an object of the present invention is to analyze the rails, sleepers, and fasteners constituting the railway through image processing obtained through a hyperspectral camera and display them in different colors. It is easy and is to provide a mobile railway facility inspection device using a hyperspectral camera that can be inspected and managed by specifying accurate location coordinates according to movement.

For the above purpose, the present invention is an inspection device for moving on a railroad track consisting of a sleeper installed on a road, rails and fasteners. A wheel is formed in the lower portion to move along the rail, and a driving unit for rotating the wheel is provided. one body; a position calculation unit for generating position information according to the movement of the main body; a photographing unit installed on the main body and configured as a hyperspectral camera for photographing a railway; an image processing unit configured to process the hyperspectral image taken by the photographing unit, a classification unit for classifying the sleeper, rail, and fastener as inspection targets through a spectral wavelength, and a processing unit for color processing the classified inspection targets; a control unit having a position determining unit for sequentially connecting the images photographed by the photographing unit by using the position information, but extracting and designating the coordinates of the selected inspection target in the connected image; characterized in that it consists of

At this time, a grip structure to which the photographing unit is coupled, a connection structure having one end connected to the grip structure and a first connecting plate formed at the other end, and a second connection having a shape corresponding to the first connecting plate and coupled to the main body a damper part connected between the plate and the first connecting plate and the second connecting plate and configured as a vibration-proof structure having a vibration absorbing structure; It is preferable to include

In addition, the position calculation unit includes a GPS module, an encoder that counts the number of rotations of the wheels and converts them into a movement distance, and an interlocking module that outputs the GPS signal and the converted movement distance in connection with regional geographic information including railroad tracks. It is preferred to be constructed.

In addition, the photographing unit further includes an RGB camera for photographing the railway, and the control unit further matches the hyperspectral image and the RGB image obtained by the photographing unit and aligns them with a matching unit for giving the same coordinates according to the same location may include

In addition, the control unit comprises: an analysis unit that determines defects of the railway through color-processed hyperspectral images and RGB images and generates defect information along with coordinate information; and an evaluation unit that receives feedback information on the defect information; The method may further include a learning unit that analyzes the defect information and the feedback information as they are accumulated and stored and generates reference information of the analysis unit.

Through the present invention, even if inspection personnel are not put into the field, it is possible to check the condition of the railway through the images taken while driving along the railway. In particular, the rails, sleepers, and By dividing the fasteners, detailed analysis can be performed for each configuration.

In addition, by acquiring accurate location coordinates according to movement and writing them as a management map, the inspection and management of the railway can be performed efficiently.

1 is a plan view showing the structure of a conventional railway;
2 is a conceptual diagram of the present invention;
3 is a block diagram showing the configuration and connection relationship according to an embodiment of the present invention;
4 is a concept showing the principle of an encoder according to an embodiment of the present invention;
5 is a view showing a structure of a damper according to an embodiment of the present invention.

Hereinafter, with reference to the accompanying drawings, the configuration of the mobile railway facility inspection device using the hyperspectral camera of the present invention will be described in detail.

2 is a conceptual diagram of the present invention, in which the present invention moves along a railway 200 consisting of a sleeper 202 installed on a ballast 201, a rail 203, and a fastener 204, and photographing the railway 200 The condition is checked through one image, and each component of the railway can be inspected separately.

The ballast 201 is a floor composed of a gravel ballast and concrete, in which sleepers 202 are installed at regular intervals on the upper side thereof, and a pair of rails 203 are parallel to the upper side of the sleeper through a fastener 204. Inspection is made through the main body moving along the conventional rail 200 is installed to

For this purpose, a wheel 101 is formed on the lower portion of the body 100 , and a driving unit 102 for rotating the wheel 101 is provided.

The wheel 101 is configured not to deviate from the rail 203 during operation in the same shape as the wheel of a railway vehicle, and the driving unit 102 may be made of an internal combustion engine such as an engine as a power means, but has low vibration characteristics and speed, etc. It is preferable to use a motor that is easy to control. In addition, it is natural that a power supply system including a battery for this should be provided in the main body 100 .

Since various known driving means can be applied in configuring the driving unit 102, a detailed description will be omitted in order to prevent the gist of the present invention from being obscured.

3 is a block diagram showing the configuration and connection relationship according to an embodiment of the present invention, in which the main body 100 moves the rail 203 and is configured to check the railway through an image, In the embodiment, a position calculating unit 110 , a photographing unit 120 , an image processing unit 140 , a control unit 150 , and a database 160 are provided as main components for such a series of functions.

The position calculating unit 110 is configured to generate position information according to the movement of the main body 100, so that the exact position of the part of the railway in the photographed image can be known. In the embodiment of the present invention, the GPS module 111 and the encoder 112 are used, and the location information is generated by a combination of the two methods to increase the accuracy, and the interworking module 113 for linking the geographic information system is also provided together.

The GPS module 111 is configured to receive a signal from a Global Positioning System (GPS) satellite and calculate the user's current location. 200) so that the current location of the image can be identified. Recently, the accuracy of GPS has increased and the error range has been greatly reduced. However, as errors still exist, precise location information can be calculated by using RTK-GPS (Real Time Kinematic GPS).

In the conventional general method, location information is calculated through communication between the GPS satellite, the control station on the ground, and the GPS module 111 provided in the main body 100, that is, the GPS receiver, but the RTK-GPS is an antenna fixed on the ground. By adding the reference station, the relative distance and angle of the main body 100 are detected in real time and the position obtained by GPS is corrected, thereby reducing the error in meters to centimeters.

In addition, in the present invention, as the sleepers 202 are arranged at relatively narrow intervals, the signal of the encoder 112 may be used together to accurately calculate the position of the sleepers.

4 is a conceptual diagram illustrating the principle of an encoder according to an embodiment of the present invention.

The encoder 112 is configured to count the number of revolutions of the wheel in conjunction with the wheel 101 , and as the diameter of the wheel is determined, the number of revolutions of the wheel may be converted into a moving distance. Specifically, a slot plate 1123 having a plurality of slots 1124 penetrating at equal intervals is mounted on a shaft rotating together with the wheel, and a light emitting element 1121 and a light receiving element 1122 with the slots 1124 interposed therebetween is mounted. ) is arranged in an optical structure, and the number of rotations of the wheel can be accurately measured by composing the encoder.

This encoder 112 can obtain signals of various waveforms according to the number and shape of slots provided in the slot plate, so that not only the number of rotations but also the rotation angle of the wheels can be grasped, and these waveform signals can be photographed as described below. It can also be used as a trigger signal for

After all, in the state where the position of the primary road is grasped through the GPS module 111, the number of revolutions of the wheels 101 is calculated based on the initial position of the main body 100 or a designated reference position, and the number of revolutions is converted into a moving distance, so that the railway 200 is Position information indicating the exact position of the upper body 100 may be generated.

The interlocking module 113 is configured to manage inspection and maintenance status in connection with a Geographic Information System (GIS), and is converted in the GPS signal and encoder 112 of the GPS module 111 . It enables to display and manage the distance traveled in connection with regional geographic information including railroads, especially railroad information.

The photographing unit 120 is installed in the main body and configured to photograph the railway, and in the present invention, the inspection is performed using the hyperspectral camera 121 . The hyperspectral camera 121 is configured to greatly improve the detection or classification ability of a target by acquiring information using the unique spectral wavelength information possessed by the photographed object, and the hyperspectral image obtained through this is very narrow. As the electronic system is subdivided again within a bandwidth, the sleepers 202, rails 203, and fasteners 204, which are the main inspection targets, are made of different materials, so that they can be clearly distinguished and analyzed.

At this time, the photographing unit 120 may shoot a continuous image, that is, a moving image, but as the camera supports a larger resolution in a still image than a moving image, it is preferable to take a snapshot in the form of a snapshot and then reconstruct it as a panoramic image as will be described later.

In this process, by using the signal generated from the encoder 112 as a trigger signal for shooting, shooting can be made at an accurate timing, and it is easy to connect the images taken at the correct period even in the process of reconstructing a panoramic image thereafter. will be done

To this end, a reduction gear that lowers the number of revolutions of the wheel 101 at an appropriate ratio is installed and a separate encoder is mounted on it, or when an encoder signal for generating the position information is repeated during a setting cycle, the trigger for driving the photographing unit 120 It can be configured to take a picture by generating a signal.

The image processing unit 140 is configured to process the hyperspectral image captured by the photographing unit 120, and classifies the sleeper 202, the rail 203, and the fastener 204 as inspection targets through the spectral wavelength. It includes a classification unit 141 and a processing unit 142 for processing the classified inspection target in different colors.

That is, in the hyperspectral image, the sleeper 202, the rail 203, and the fastener 204 made of different materials can be easily distinguished by using the unique spectral wavelength information of the object. By converting each color into a designated color, the administrator can easily distinguish each inspection target in the image.

In classifying the sleepers, rails, and fasteners through these spectral wavelengths, the image processing unit 140 can learn the spectral wavelengths of the sleepers, rails, and fasteners by applying deep learning. can be done As mentioned above, in hyperspectral images, unique spectral wavelengths of each object can be identified, so it is possible to find and learn the characteristic points of the spectral wavelengths of sleepers, rails, and fasteners appearing in hyperspectral images through deep learning. classification can be made.

The control unit 150 is a configuration corresponding to the MCU, and controls the driving unit 102, the position calculating unit 110, the photographing unit 120, and the image processing unit 140 to capture an image along with the movement of the body. The detailed configuration of the location determination unit 151, the matching unit 152, the analysis unit 153, the evaluation unit 154, the learning unit 155, and the map creation unit 156 to provide

The position determining unit 151 connects the images photographed by the photographing unit 120 in order by using the position information obtained through the position calculating unit 110, but extracts and holds the coordinates of the selected inspection target in the connected image decide

That is, the images captured by the photographing unit 120 are arranged according to the photographing order, but are reconstructed into a panoramic image in which the overlapping portion is excluded. At this time, by combining the images corresponding to the rotational speed of the wheel calculated through the encoder 112, that is, the speed of the moving body 100, the actual results of the sleeper 202 and fastener 204 appearing in a repeated form in the image. It is possible to organize images that are naturally connected according to the interval, and as mentioned above, when the signal of the encoder 112 is used as the trigger signal of the photographing unit 120, the image is acquired according to a predetermined period, so it is easy to combine them to form a panorama of image reconstruction can be made.

In addition, the coordinates of each inspection target extracted from each photographed image are reassigned according to the set criteria as multiple images are reconstructed in a panoramic form, so that accurate coordinates in the reconstructed image can be obtained.

At this time, if there is a problematic part in the image, it is necessary to apply the location information to the image so that the exact location for re-inspection and repair can be found. The location determination unit 151 applies the location information generated through the location calculation unit 110 to the image, and utilizes the previously re-designated coordinates, so that the inspection personnel can accurately identify the location of the part where the abnormality occurs.

In addition, inspection personnel can see each component of the railway that has been processed by color, that is, the sleeper 202, rail 203, and fastener 204 by a designated color, and can easily check whether they are cracked, deformed, or separated. In addition to being able to grasp the inspection target, it is possible to improve the management convenience by creating a separate image of the inspection target and operating it in a form that reflects the replacement and maintenance schedule and operates the inspection cycle differently.

As an auxiliary means along with the configuration using the hyperspectral image, the photographing unit 120 may further include an RGB camera 122 for photographing the railway in the same way as the hyperspectral camera 121 . This helps to make more accurate judgments in various situations by reconfirming the part that is judged to have an abnormality through the hyperspectral image through the RGB image taken with the original color.

Correspondingly, the matching unit 152 provided in the control unit 150 matches the hyperspectral image and the RGB image obtained by the photographing unit 120 with each other and aligns them in the same position, but gives the same coordinates according to the same position. It makes it easy to compare and analyze hyperspectral images and RGB images for locations.

In addition, the processed images as described above are collected for inspection, and through visual confirmation by inspection personnel, the same level of inspection can be made even if the personnel are located on the site and do not move, but the image analysis algorithm automatically determines the defective part. This may be done.

To this end, the analysis unit 153 provided in the control unit 150 determines the defect of the railway through the color-processed hyperspectral image and the RGB image, and generates defect information together with coordinate information. That is, the main analysis targets of the captured image are rails, sleepers, and fasteners. As their shapes are constant, they are compared with each other, compared with a reference image, or, if there is an existing captured image, it is compared with the previous image of the same location. Defects can be determined by extracting differences by comparison.

The evaluation unit 154 is configured to receive feedback information on the defect information calculated through the analysis unit 153, and a detailed inspection and Maintenance is made, and in this process, the inspector inputs whether the judgment result is correct and whether the judgment is accurate as feedback information.

The learning unit 155 is configured to generate the reference information of the analysis unit by analyzing the defect information and the feedback information as they are accumulated and stored. If it is determined through the feedback information that it is not an actual defect, the analysis unit 153 does not make the same determination later, and, conversely, as the management personnel inputs feedback information as having an actual defect for the part determined to have no defect, Learning is performed so that accurate judgment results are derived by reflecting these inputs in the future.

The database 160 basically stores the captured and processed images according to the date and time, and the defect information and the feedback information of the analysis unit 153 and the evaluation unit 154 are accumulated and stored. As they increase, they can be used as big data to improve the railway inspection function in the future.

The map creation unit 156 creates and outputs a map capable of managing the inspection conditions for each route by region in connection with the geographic information on which the railroads are displayed, so that complex railroad inspections can be systematically performed.

5 is a view showing the structure of a damper according to an embodiment of the present invention.

In addition, the photographing unit 120 is coupled to the main body 100 through the damper 130 so as to prevent the photographing unit 120 from shaking due to vibrations or small shocks generated during movement of the main body and from lowering the image quality. It is preferable to be

Specifically, the damper unit 130 is connected to a grip structure 131 to which the photographing unit 120 is coupled, one end connected to the grip structure 131 and a first connecting plate 133 formed at the other end. The structure 132, the second connecting plate 135 having a shape corresponding to the first connecting plate 133 and coupled to the main body 100, the first connecting plate 133 and the second connecting plate It is composed of a vibration-proof structure 134 that connects between the 135 and has a vibration-absorbing structure, so that vibration or shaking from the main body 100 is applied to the photographing unit 120 and a clearer image can be obtained. .

The rights of the present invention are not limited to the above-described embodiments, but are defined by the claims, and those of ordinary skill in the art can make various modifications and adaptations within the scope of the claims. it is self-evident

100: body 101: wheel 102: driving unit
110: position calculator 111: GPS module 112: encoder
1121: light emitting element 1122: light receiving element
1123: slot plate 1124: slot
113: interlocking module
120: photographing unit 121: hyperspectral camera 122: RGB camera
130: damper part 131: grip structure 132: connection structure
133: first connection plate 134: vibration-proof structure
135: second connection plate
140: image processing unit 141: classification unit 142: processing unit
150: control unit 151: location determination unit 152: matching unit
153: analysis unit 154: evaluation unit
155: learning unit 156: map creation unit
160: database
200: railroad 201: roadway 202: sleeper
203: rail 204: fastener

Claims (5)

As an inspection device for moving on a railroad 200 consisting of a sleeper 202 installed on a ballast 201, a rail 203, and a fastener 204,
a body 100 having a wheel 101 formed at a lower portion to move along the rail 203, and a driving unit 102 for rotating the wheel 101;
A GPS module 111, an encoder 112 that counts the number of revolutions of the wheel 101 and converts it into a moving distance, and a linkage that outputs the GPS signal and the converted moving distance in connection with regional geographic information including the railway a position calculating unit 110 having a module 113 and generating position information according to the movement of the main body 100;
A hyperspectral camera 121 installed in the main body 100 and using a signal generated from the encoder 112 as a trigger signal to photograph the railway 200 in the form of a snapshot, and an RGB camera 122 for photographing the railway ), the photographing unit 120 consisting of;
A classification unit 141 that processes the hyperspectral image photographed by the photographing unit 120, and classifies the sleeper, rail, and fastener by a specified color through the spectral wavelength into inspection targets, and the classified inspection targets, respectively. an image processing unit 140 including a color processing unit 142;
Using the location information, the images taken by the photographing unit 120 are sequentially connected and reconstructed into a panoramic image, and a location determination unit 151 for extracting and designating the coordinates of the selected inspection target in the connected image, and the photographing unit A matching unit 152 that matches and aligns the hyperspectral image and the RGB image obtained in 120 but gives the same coordinates according to the same location, and determines the defect of the railway through the color-processed hyperspectral image and the RGB image, An analysis unit 153 that generates defect information together with coordinate information, an evaluation unit 154 that receives feedback information for the defect information, and the analysis unit ( A control unit having a learning unit 155 for generating the reference information of 153), and a map creation unit 156 for creating and outputting a map that can manage the inspection conditions for each route by region in connection with geographic information on which railways are displayed ( 150); A mobile railway facility inspection device using a hyperspectral camera, characterized in that consisting of.
According to claim 1,
A grip structure 131 to which the photographing unit 120 is coupled, a connection structure 132 having one end connected to the grip structure 131 and a first connecting plate 133 formed at the other end, and the first A second connecting plate 135 having a shape corresponding to the connecting plate 133 and coupled to the main body 100 is connected between the first connecting plate 133 and the second connecting plate 135 to absorb vibration. a damper unit 130 composed of a vibration-proof structure 134 having a structure; A mobile railway facility inspection device using a hyperspectral camera, characterized in that it comprises a.
delete delete delete

Images