Assembly and method for optically capturing a trajectory - Patents.com

Abstract

The invention relates to an assembly (11) for optically capturing a track (2), comprising a support device (12), a camera (13) and two light sources (14), which can be fixed to a transport vehicle (1) movable on a rail (3), in which the orientation of the central axes of the two light sources (14) and thus the emitted light beams or light waves comprises an acute angle (β1, β2) with the rail longitudinal axis (15), the light sources (14) being coupled to the camera (13) via a control signal cable (16), the triggering or activation of the light sources (14) being carried out using a predetermined control signal via the control signal cable (16). This achieves detailed and sharp image contours, which allows for highly accurate image evaluation and object recognition at the highest possible working speed.

Description

[Technical field]

[0001] The present invention relates to an assembly for optically capturing a trajectory, the assembly including a support device, a camera and two light sources, and which can be fixed to a transport vehicle movable on rails. The present invention further relates to a method for operating the assembly. [Background technology]

[0002] In modern automated image capture and image evaluation, which focuses on object recognition, a wide variety of methods and procedures are used. For this purpose, the principle of optical object capture with dark field illumination is known. In order to capture objects quickly and with high contrast using a camera, the light from a light source is illuminated flat. If irregularities are present on the object surface, which may be bumps or depressions, such as defects, beveled edges, embossings, etc., the light path is obstructed. This intentionally results in shading, which has a significant effect on the downstream image evaluation and object recognition. By orienting the light beam approximately parallel to the object to be recorded, the light is refracted accordingly only at the edges and reflected towards the camera. This refraction of light leads to a dark image background, in front of which the structures to be observed stand out brightly. This creates a high-contrast image and highlights the irregularities. In known applications, cameras with one color channel as well as one light source are usually used.

[0003] The applicant's AU 518692 A1 discloses a method and a system for the maintenance of rail tracks for railway vehicles. The measuring system described therein comprises a line laser scanner 14 arranged on each rail 6 in order to capture the shape of the respective rail surface. This also captures further features such as rolling marks, embossments on the sleepers, fastening means, etc. In the aforementioned method, a so-called 3D point cloud is generated from the objects captured by the line laser scanner. By means of corresponding calculation and evaluation algorithms, the geometry and the state of the captured track can be deduced from the data of the 3D point cloud. Summary of the Invention [Problem to be solved by the invention]

[0004] The present invention is based on the problem of providing an improvement over the prior art for an assembly of the type mentioned at the beginning, in order to increase the performance and thus the length of the track to be acquired per unit of time. Furthermore, the operation for acquiring the track can be carried out with shorter downtimes and with a higher quality of the data and image material obtained. Furthermore, it is an object of the present invention to provide a corresponding method for operating the assembly. [Means for solving the problem]

[0005] According to the invention, this problem is solved by an assembly having the features of independent claim 1 and by a method according to the features of independent claim 9. The dependent claims give preferred embodiments of the invention.

[0006] Here, the orientation of the central axes of the two light sources and therefore the emitted light beams or waves includes an acute angle (β1, β2) with the rail longitudinal axis, the light sources are coupled to the camera via a control signal cable, and triggering or activation of the light sources is performed using a predetermined control signal via the control signal cable.

[0007] By orienting the two light sources at the smallest possible acute angles β1, β2 compatible with the predefined frame and construction conditions relative to the rail longitudinal axis, the greatest possible shading can be provided for rail rolling marks, lettering and other objects to be captured. This is a prerequisite for a subsequent highly accurate image evaluation and object recognition at high working speeds. If the light sources are oriented in the immediate vicinity of the rails and parallel to the respective rails, with angles β1, β2 equal to zero, the greatest possible shading will be obtained. However, such a spatial arrangement is excluded in order to comply with the available clearance profiles predefined by the infrastructure operator and to prevent collisions with objects on the track.

[0008] The two light sources form a pair and are arranged symmetrically with respect to the direction perpendicular to the rail longitudinal axis. This results in opposing opening directions with angles β1 and β2, which are therefore also symmetric with respect to the direction perpendicular to the rail longitudinal axis. In order to be able to capture the entire lateral rail contour and the nearby areas, one light source pair is arranged on each side of each rail track, surrounding the camera. For a better understanding of this, please also refer to Fig. 2 and Fig. 3. In addition to rail rolling marks, embossings, lettering, small irons and fastening materials, rail joints can also be captured. The latter are usually designed as welded connections or as non-welded multi-part fishplate connections.

[0009] By coupling the two light sources to the camera via a control signal cable, the triggering of the light sources and the image capture of the camera simultaneously are ensured via a pre-set control signal. This control signal is set as a path-based information of, for example, one pulse per 25 cm of travel distance. Alternatively, the control signal may be a time signal with, for example, a frequency of 160 Hz. This allows optimal exposure and detailed image contours with very short shutter / exposure times of the camera. With the current state of the art, working speeds and therefore capture speeds of more than 100 km per hour are feasible.

[0010] Preferably, the camera is positioned approximately midway between the light sources when viewed in the direction of the rail longitudinal axis, the camera axis is oriented approximately perpendicular to the rail longitudinal axis, and includes an acute angle (α1, α2) between the central axis of the light source and the horizontal floor surface.

[0011] A camera oriented centrally between the two light sources allows the projection of different intensities from raised / recessed features on the object to be captured to be visible. The additional inclination of the light sources by angles α1 and α2 results in an additional third angle in the projection diagonally downwards in the direction of the rail foot in addition to the above mentioned shadows.

[0012] Furthermore, it is an advantage that the camera is designed as an RGB video camera having three color channels: red, green and blue.

[0013] This allows separate capture and reading of image data for the individual color channels. Here, one grayscale image each is obtained for the three color spectrums: red, green and blue. For digital image processing in the RGB color space, this means 8 bits per channel, and therefore a maximum color depth of 3x8 bits = 24 bits overall for all color channels. For the disclosed invention, it is advantageous to read image information from two color channels, in this case modulated with emitted light beams, in particular with two different wavelengths of two light sources. Compared to known methods in which several cameras are used, the use of three or two separate color channels from one camera is a major advantage, since it makes the synchronization of the channels obsolete. In contrast, three independent cameras with only one channel output would require a complex synchronization of the image content.

[0014] In one embodiment of the invention, it is envisaged that a camera, in particular a camera lens, is assigned an optical colour filter which allows only a certain region or a certain sub-region of the optical spectrum or a certain bandwidth of the wavelength region of the incident light to enter the optical system of the camera.

[0015] The color filters, also called attachment filters, are also adjusted to the two wavelengths of the light sources used in this specification. In addition to the already described above adjustment of the wavelengths of the emitted light of the two light sources, the use of the color channels of the RGB camera results in a significant improvement of the light / shadow ratio. This allows the incidence of undesirable light spectra and scattered light, such as, for example, sunlight or artificially generated task lights, to be largely blocked. The color filters may be designed as multiply acting combination filters or offset from one another in order to allow combinations of different wavelengths (for example, red and green or red and blue) to pass unhindered.

[0016] Also preferably arranged in front of the camera is a device for generating a so-called air curtain, which is a particularly uniform air flow.

[0017] This guarantees high quality and problem-free image material by being particularly less susceptible to external weather influences such as, for example, splashes of water, particle deposits and other contamination.

[0018] It is envisaged that the light source is designed as an LED lighting means. LED lighting technology is characterised by a very robust and long-life construction type with high beam power.

[0019] The light source is clocked via a control signal, similar to a flashlight. The frequency of this control is 160-200 Hz. Higher frequencies are also possible in the workplace. The clocked flash / trigger of the light source in this frequency range is imperceptible to the human eye. The frequency of the light source is adjusted to the camera to achieve the desired flash time. Both the high luminous flux of the LED light source and the high light intensity allow high relative speeds between a moving camera and a stationary orbit, without compromising the quality and sharpness of the image recording, even in the case of very short exposure times. In addition, the high light intensity is advantageously able to wipe out most of the external light, such as direct sunlight.

[0020] In a preferred variant of the invention, it is envisaged that each light source has a fixed, pre-set wavelength and therefore a colour spectrum corresponding thereto.

[0021] To facilitate the adjustment of the light source to the camera, it is convenient if the two light sources each emit a light beam at one specific wavelength or in a specific spectral region. These LED light sources are then governed by specifications that are precisely defined for this application and directly integrated into the acquisition process. Thus, for example, to capture and evaluate image information via the red color channel of the camera for shading, a light source with a wavelength around 660 nm is preferably mounted. A light beam with this wavelength is perceived as red by the human eye. Furthermore, for example, blue with a wavelength around 460 nm and green with a wavelength around 530 nm can be quantified.

[0022] In a further development of the invention it is provided that the camera is coupled to a computing unit, the computing unit having an integrated internal memory.

[0023] This ensures a direct transmission of the data, in particular the image information, to the computing unit, where, depending on the configuration, it is either fed to the processing and evaluation algorithms or stored and buffered for further processing.

[0024] In the method for operating the assembly according to the present invention, with the start of the work process, the camera and the light source are activated, where the light source is triggered synchronously using the provided control signal, and at the same time, the trajectory point to be captured is recorded by the camera and the image data obtained therefrom are captured by the computing unit.

[0025] By jointly controlling the driving of the light source and the camera via a control signal, optical capture of the trajectory is achieved with the best possible illumination and shortest exposure times. Clocked activation of the light source in a flash manner allows maximum light emitting efficiency without any overloading or damage to the light source in the process.

[0026] In this case, it is advantageous if the image data captured by the computing unit is stored in a buffer memory for later or time-shifted use or is directly read in by the computing unit for further processing.

[0027] This allows flexible access to a wide variety of processing, evaluation and decision processes of the captured image data, whereby data processing can be performed almost in real time on the respective transport vehicle at the work site, depending on the task, course and requirements of the activity, or at any other time, for example after the end of the work site, by accessing the memory of the computing unit. The captured image data and information can be transferred by means of a standard interface to other stationary computing units or to mobile notebooks, for example, for simple observation or extensive evaluation.

[0028] More preferably, the image data read into the calculation unit, i.e. the raw data from the camera, is separated into individual color channels, whereupon the image information of the image data or of two color channels contained therein, preferably the red and green or red and blue color channels, is subtracted from one another.

[0029] Since a camera, preferably an RGB camera, outputs the colors (red, green, blue) as an overall image signal, by separating the RGB color information into three separate channels, it is possible to obtain one image for each of the corresponding color spaces, i.e. - Image 1 containing red light image content (wavelength, approximately 660 nm), Image 2 containing green light image content (wavelength, approximately 530 nm); -Image 3 containing blue light image content (wavelength, approximately 460 nm) In the case of an application according to the invention with two light sources defined as red and green in their emission coloration, this means that Image content produced by a red light source, in particular an image 1 containing a projection of a red light source; -Image content produced by green light sources, especially images containing the projection of a green light source 2 The same is true for two light sources defined as red and blue, i.e. Image content produced by a red light source, in particular an image 1 containing a projection of a red light source; -Image content produced by blue light sources, especially images containing the projection of a blue light source2 This also applies to.

[0030] At each camera recording time, these two described images are subtracted in a calculation unit in order to clearly display the shading and contours. According to the invention, by taking into account the two color channels and thus the image information of two differently created shadings generated from two approximately opposite directions contained therein, a doubling of image information relevant for further use is provided. In this way, after subtraction, a particularly high-contrast image material is obtained with a pronounced relief representation of the object captured on track. The use of two separate color channels makes it possible to display almost all contours extending in different directions, even in the case of subjects that are difficult to capture. In this respect, for example, cursive writing as well as diagonally arranged letters / symbols may be mentioned.

[0031] It is of secondary importance whether the colouring of the two light sources is chosen to be red and green, red and blue or green and blue. This choice depends on the rail infrastructure operator and also on possible legal settings and regulations. In some countries, the colour green may be important in the rail industry. In that case, a blue-emitting light source can be used instead.

[0032] In one embodiment of the present invention, it is envisaged that the image information obtained from the subtraction of the two colour channels is evaluated, interpreted and interpreted in a computational unit using pattern and / or text recognition algorithms, OCR software or artificial intelligence "AI" or neural network "CNN" techniques.

[0033] The automated evaluation and assessment of pre-prepared image material allows for a versatile and further use of the information obtained at the work sites along the track. Well-established and known methods are used here. Purely textual objects can be recognized using OCR software "Optical character recognition", while more complex image objects such as symbols, special characters, foreign objects, welded seams (to name just a few) have to be detected using complex methods. The latter can also be interpreted and assessed using artificial intelligence methods.

[0034] It is further envisaged that the control signals are generated by a controller unit integrated in the camera and output via a control signal cable.

[0035] This achieves an efficient use of the components used in the assembly. Furthermore, a separate unit for generating the control signals is not required.

[0036] The control signals can also be generated by a computing unit or a signal generator and output via a control signal cable.

[0037] In the case of a particular application, depending on the type and functional extent of the camera, it may be advantageous if the control signal is generated externally by a signal generator. This signal generator may represent an independent unit coupled to the control signal cable. Another possibility is that the signal generator is integrated as a functional module in the computing unit or in a higher-level management / open-loop control / closed-loop control system. Likewise, distance-based control signals are possible, for example generated by a distance measuring wheel.

[0038] The invention will now be described in an exemplary manner with reference to the accompanying drawings, in which: [Brief description of the drawings]

[0039] [Figure 1] FIG. 1 is a schematic side view of a transport vehicle equipped with an assembly for optically capturing a trajectory. [Diagram 2] FIG. 2 is a schematic cross-sectional view of a track body with the above assembly. [Diagram 3] FIG. 3 is a schematic plan view of FIG. 2. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0040] FIG. 1 shows a transport vehicle 1, which is designed and illustrated diagrammatically as a universally applicable maintenance or track-laying machine. In the application case of the invention, it is used in its function as a measuring vehicle for optically capturing the track 2. For this purpose, the assembly 11 according to the invention is fixed to a machine frame 7. In other use cases, not shown, the assembly 11 can also be mounted on any respective track-laying machine with corresponding accommodation means, for example a tamping machine, a ballast plough, an inspection and crane vehicle. The transport vehicle 1 is mounted on a rail carriage 6 via the machine frame 7 and can move on the track 2. The machine frame 7 is designed in the simplest case as a vertical beam, which is mounted on the rail carriage 6 at its end side. The track 2 comprises here rails 3 on sleepers 4 embedded in ballast 5. The frequently used and widespread term rail track is considered equivalent to the term rails 3. Furthermore, the management / open-loop control / closed-loop control system 10 as well as the calculation unit 19 are integrated in a protected area or booth 8 on the transport vehicle 1. Furthermore, the current rail position is captured by means of measuring systems 9 arranged in the end regions of the machine frame 7 .

[0041] The assembly 11 comprises a support device 12, a camera 13 and two light sources 14, where the camera 13 and the light sources 14 are arranged on the support device 12. The camera 13, the light sources 14 and the calculation unit 19 are connected via control signal cables 16, which, in addition to transmitting the control signals, also serve to transmit image data. The calculation unit 19 is integrated into the higher-level management / open-loop control / closed-loop control system 10. Alternatively, the calculation unit 19 can be mounted as a separate, independent unit. In a design variant, it is also possible to transmit the control signals wirelessly in a correspondingly configured and communicating transmitting / receiving unit. The working or travel direction 20 is independent of the process of optical detection of the track 2. This is equally possible in both directions.

[0042] FIG. 2 shows a cross-section of the track 2 and a cross-section of the machine frame 7 of the transport vehicle 1 as well as a quadruple embodiment of the assembly 11. In the variant embodiment shown in FIG. 2 and FIG. 3, a total of four assemblies 11 are used to capture the two rails (rail tracks) 3 from both sides. As a result, a total of four cameras 13 and eight light sources 14 are applied. For the purpose of better visibility, the camera 13 is shown in FIG. 2 tilted by an angle γ. Usually, the camera 13 and the light source 14 are arranged one behind the other, as seen in the direction of the rail longitudinal axis 15. This means that both the camera axis 17 as well as the central axis of the light source 14 have approximately the same inclination or angular position with respect to the horizontal underground surface 18. The angles α1, α2 and the angle γ of the two light sources 14 are therefore approximately the same. These angles are usually between 35° and 45° for constructional reasons. Here, compliance with the available clearance profile and therefore the possible construction space is decisive for the arrangement.

[0043] The support devices 12 of each assembly 11 are attached to the transport vehicle 1 by means of additional support structures installed between the machine frames 7. Although not shown in the illustrated embodiment, there is a color filter in front of the camera lens of the camera 13, and a color filter similarly arranged in front of the light source 14. The depiction of an apparatus for generating an air curtain, which would be considered to belong to the camera 13, is likewise omitted.

[0044] Furthermore, in the variant of Fig. 2, a mechanical device 21 for blocking incoming direct sunlight is provided. This device is arranged below the machine frame 7 and has an adjustable inclination. This inclination can be variably adjusted by a drive (not shown) about an axis parallel to the longitudinal axis of the machine frame 7. This results in a significant reduction in the incoming sunlight and other scattered light. This device 21 is not shown in the remaining figures.

[0045] FIG. 3 shows the cross section of FIG. 2, here in plan view. The outer assembly 11, assigned to the right rail track 3, is highlighted with a dotted outline for better distinction. The light sources 14 are arranged such that their associated central axes have acute angles β1 and β2 with the rail longitudinal axis 15. These two angles are of equal magnitude in their absolute value in order to obtain a uniform projection from both sides of the object to be captured. The angles β1 and β2 are typically between 10° and 30°, which ensures a flat light incidence with maximum shading effect.

Claims

1. An assembly (11) for optically capturing an orbit (2), comprising a support device (12), a camera (13) and two light sources (14), wherein the assembly (11) is fixable to a vehicle (1) movable on a rail (3), in the assembly (11), the orientation of the central axes of the two light sources (14), and thus the emitted light beams or light waves, includes acute angles (β1, β2) with the longitudinal axis (15) of the rail, the light source (14) is coupled to the camera (13) via a control signal cable (16), characterized in that the triggering or activation of the light source (14) is performed using a predetermined control signal via the control signal cable (16). Assembly (11).

2. The camera (13) is arranged approximately in the center between the light sources (14) when viewed in the direction of the longitudinal axis (15) of the rail, the camera axis (17) is oriented approximately perpendicular to the longitudinal axis (15) of the rail, and has acute angles (α1, α2) between the central axis of the light source (14) and the horizontal floor (18). The assembly according to claim 1.

3. The camera (13) is designed as an RGB video camera having three color channels of red, green and blue. The assembly according to claim 1.

4. An optical color filter is assigned to the camera (13), in particular to the camera lens, and the color filter allows only a predetermined region or a predetermined sub-region of the optical spectrum, or a predetermined bandwidth of the wavelength region of the incident light, to enter the optical system of the camera (13). The assembly according to claim 1.

5. A device for generating a so-called air curtain of a particularly uniform air flow is arranged in front of the camera (13). The assembly according to claim 1.

6. The light source (14) is designed as LED lighting means. The assembly according to claim 1.

7. Each light source (14) has a preset fixed wavelength and thus a corresponding color spectrum. The assembly according to claim 1.

8. The camera (13) is coupled to a computing unit (19), which has an integrated internal memory. The assembly according to claim 1.

9. A method for optically capturing an orbit using the assembly (11) according to any one of claims 1 to 8, at the start of the working process, the camera (13) and the light source (14) are activated, and the light source (14) is triggered synchronously using the provided control signal, wherein the orbit location to be captured is simultaneously recorded by the camera (13), and the resulting image data is captured by the computing unit (19).

10. The image data captured by the computing unit (19) is stored in a buffer memory for subsequent use or time-shifted use, or is directly read by the computing unit (19) for further processing. The method according to claim 9.

11. The image data read into the computing unit (19), i.e., the raw data from the camera (13), is separated into individual color channels, and subsequently, the image information of the image data or two color channels contained therein, preferably the red and green, or red and blue color channels, is subtracted from each other. The method according to claim 9.

12. The image information obtained from the subtraction of the two color channels is evaluated, interpreted, and judged within the computing unit (19) using a pattern and / or text recognition algorithm, OCR software, or techniques of artificial intelligence "AI" or neural network "CNN". The method according to claim 11.

13. The control signal is generated by a controller unit integrated in the camera (13) and output via a control signal cable (16). The method according to claim 9.

14. The control signal is generated by the computing unit (19) or a signal generator and output via the control signal cable (16). The method according to claim 9.

Images