WO2018104462A1

WO2018104462A1 - Method, device, and track-bound vehicle, in particular rail vehicle, for a track-based image analysis in track-bound traffic, in particular for a rail-based image analysis in rail traffic

Info

Publication number: WO2018104462A1
Application number: PCT/EP2017/081845
Authority: WO
Inventors: Andreas Schaefer-Enkeler; Andreas Schönberger; Christopher Drexler; Wolfgang Trumler
Original assignee: Siemens Aktiengesellschaft
Priority date: 2016-12-07
Filing date: 2017-12-07
Publication date: 2018-06-14
Also published as: EP3523175A1; RU2720303C1; CN110248858A

Abstract

The aim of the invention is to improve the automatic (autonomous) or supported travel of track-bound vehicles (BFZ) or rail vehicles (SFZ) without additional infrastructure along a route. This is achieved in that on the basis of multiple images (BI_FSB, BI_GLB) that represent a track/rail region (FSB, GLB) in an image region (BIB), which is produced in each image and in which captured tracks/rails (FS, GL) lie substantially in the same location relative to the graphically displayed track/rail region (FSB, GLB), the course of a track/rail (FS, GL) used by the rail vehicle through a track/rail image section which changes in the captured image relative to the total captured image is detected on the basis of the sections of the captured tracks/rails in the image region (BIB) using edge detection algorithms and is compared with stored known meta-information (MI), wherein the meta-information contains both track-related or rail-related primary metadata as well as track route-related or rail route-related secondary metadata.

Description

description

Process, apparatus and rail vehicle, in particular sliding ^¬ nenfahrzeug, the lane based image analysis in the dispatch rail traffic, in particular for track-based image analysis slide ^¬ nenverkehr

The invention relates to a method for lane-based image analysis in rail traffic, in particular for track-based image analysis in rail traffic according to the preamble of patent claim 1, an apparatus for lane-based image analysis in rail traffic, especially for track-based image analysis in rail traffic according to the preamble of claim 11 and a Railway vehicle for lane-based image analysis in rail traffic, in particular rail vehicle for track-based image analysis in rail traffic, according to the preamble of claim 26.

Rail vehicles are part of a modern traffic infrastructure track-bound transport and transport, for example, rolling on or under of one or two rails (tracks), floating above or below a magnetic field or hanging on steel cables move. Among the mentioned rail-bound transport and transport means are rail vehicles based on a wheel-rail system which are either pulled or pushed by their own traction drive (railcar) or by a locomotive and predominantly steel wheels with a wheel flange on two steel rails railways are the most widespread.

The underlying the invention object is to provide a method, a device and a railway vehicle, insbesonde ^¬ re rail vehicle, to specify the lane based image analysis in rail traffic, in particular for track-based image analysis in rail transport, with or automated (autonomous) or assisted driving the railway vehicles Rail vehicles without additional infrastructure along a route is improved.

Carrying out a lane-based image analysis in railway traffic, in particular a track-based image analysis in railway traffic

Railway transport, which is the subject of the present International Patent Application (Application No. PCT / ...; Publication No. WO ...) and the priority patent application DE (Application No. 102016224331.4), is with regard to a future automated (autonomous) or assisted driving of rail vehicles in rail traffic or rail vehicles in rail transport meaningful improvement for the stated purpose. So it is necessary for automated or assisted driving of the train / rail vehicles without additional infrastructure along the route, evaluate images from the vehicle, in particular ^¬ sondere from the perspective of railcars leader automatic ^¬ Siert. These images may be from at least one imaging device, but preferably from multiple devices, such as video cameras, laser sensors, radar devices, thermal imaging cameras, or other image acquisition devices.

The problem of automated driving has so far been realized through costly additional investments in the route infrastructure such as induction loops, computers along the route and communication systems between train and route components. However, it is not only the aspect of lane / track-based image analysis that is important for future automated (autonomous) or assisted driving, but also the following technical aspects, all more or less in a technical context with the present invention. and therefore listed and their contents have to be taken into account against this background and may even be included. These are the aspects:

1) The automatic recognition of signals in rail / rail traffic according to International Patent Application (Application No. PCT / EP2016 / 057804, Publication No. WO 2017/174155 Al) and the technical teaching disclosed therein.

2) The automatic recognition of hazardous situations in rail / rail traffic according to DE patent application (application No. 102016224358.6) and International patent application (application No. PCT / ..., publication No. WO ...) and the therein respectively disclosed technical teaching.

3) The automatic detection of obstacles in rail / rail traffic according to DE patent application (application No. 102016224344.6) and International Patent Application (application No. PCT / ..., Publication No. WO ...) and the therein respectively disclosed technical teaching. 4) The alternative determining positions in the rail when a conventional satellite-based Positionsbe ^¬ humor fails or is insufficient, (in accordance with the DE patent application Application No. 102016224355.1) and the International patent application (application number PCT / ...,.. Veröffentli - Chungs No. WO ...) and the technical teaching disclosed therein.

5) The automatic recognition of lanes / tracks in railway / rail traffic according to DE patent application (application No. 102016224335.7) and International patent application (application No. PCT / ...; publication No. WO ...) and the technical teaching disclosed therein.

The above-mentioned context-related object is achieved on the basis of the image analysis method defined in the preamble of patent claim 1 by the features specified in the characterizing part of patent claim 1. In addition, the above-mentioned contextual object is achieved on the basis of the image analysis device defined in the preamble of patent claim 11 by the features specified in the characterizing part of claim 11.

Furthermore, the above-mentioned contextual Aufga ^¬ be, starting from the defined in the preamble of claim 26 railway vehicle, in particular rail vehicle, achieved by the features specified in characterizing portion of claim 26 features.

The idea underlying the invention according to the independent claims 1, 11 and 26 is based on several images representing a traffic lane / track area in an image area respectively identified in the images, in relation to the depicted lane / track Lanes / tracks detected in the area essentially lie in the same place, starting from the proportions of the detected lanes / tracks in the image area with the aid of

Edge detection algorithms to recognize the course of a lane / track, which is used by the railway vehicle by a changing in the captured image portion of the lane / track to the entire picture captured and compared with stored known image metadata, these both lane-related or track ^¬ related primary metadata as well as railway-related res ^¬ pective rail-track related secondary metadata enthal ^¬ th.

In the literal sense, the image meta information includes feature and characteristic data of the lane / track area recorded in the images. The basic principle of the invention is, by means of an automated evaluation of the images, a part of the traffic lane / traffic lane that is visible in the respective image Identify tracks so that preferably (in advantageous developments) the following objectives are achieved:

I. Detection, where in the picture pre and main signals are to be found for a driving operation.

II. Detection of whether a track continues to LEFT, STRAIGHT, RIGHT.

III. Detecting if people are on a lane / track.

IV. Detecting how far a railway vehicle is from an object.

V. Detection of impermissible objects on a lane / track. The goal in each case is to allow a contribution to the fully automatic ^¬ terraced driving without additional investment in the track infrastructure.

The automated evaluation of images that are captured from the vehicle, particularly from the perspective of the motor car guide and / or of a stationary fahrspurobservierenden position in or on the vehicle, was performed, so far only rate by analysis of the "front of the vehicle lying preparation ^¬ ches" without detailed reference to the lane / track in the picture. However, it can advantageously be achieved at least partially by the following steps:

1. In a first step, (a, for example, or more Bildak- quisitionsgeräte such as video cameras, laser sensors, Infrarotka ^¬ ra, thermal imaging cameras, radar devices, other Bildak- quisitionsgeräte, etc.) with the aid of at least one image recording apparatus a plurality of images from the web ^¬ vehicle from, for example, from the perspective of the railcar driver, he ^¬ summarized or recorded.

The multiple execution is mainly relevant for redundancy purposes. a. In a modification or further development, extension of the invention but also images of several image recording devices or image acquisition devices of the same type (eg, two video cameras) for mutual validation and synthesis of the results can be used. b. In a further modification, or further, extension of the invention but also images of a plurality of image recording devices and image acquisition devices of different type (eg, a video camera and a Wärmebildka ^¬ ra) may be used for mutual validation and synthesis of the results. c. In a further modification or development, extension of the invention, non-imaging sensors such as radar, ultrasound or laser can be used to validate the image information.

2. In a second step, the area at the near bottom of the screen is considered, because the lanes / tracks there are within We ^¬ sentlichen at the same location, while it is preferably the case that the image recording apparatus is fixedly mounted in the railway vehicle / rail vehicle. ^¬ by edge detection algorithms, which are used in a standard image processing, the profile of the lane traveled / busy of the track can be detected on the proportions lane / the track at the bottom of starting. a. In an optimization, the maximum degree of bending of tracks / tracks can be taken into account in order to avoid false identifications of tracks).

b. In a further optimization, the traveled lane / track can also be detected across points, in which the continuous guide rail of the track is detected.

3. In a third step, metadata about the lane / track known is used, in particular the distance between the parallel laid rails and the width of the rail head.

4. In a third step, metadata about the route, which are known, are used, for example at which geo ^¬ graphical positions railroad crossings, points or signal systems are.

5. On the basis of the above data and steps, the above five objectives I ... V.

I. Detection, where in the picture pre and main signals are to be found for a driving operation

i. Due to the current geographical position [for the railway vehicle / rail car are the relevant Ortskoor ^¬ ordinates based on GPS data or Positionsbestimmungsda ^¬ th according to the German patent application (application no. ...) known] can first of area Detect image in which a signal is detected.

ii. Then it is determined, where in this area the busy track performs (result of the data from step 2)

iii. Metadata about the distance of the signal system can then be used as follows.

1. In the relevant section of the picture, it is possible to know on the basis of the lane / track width how far the signal system is horizontally and vertically away from the track.

2. Furthermore, the outer edges of the signal system are known.

3. Now, at short distances from the vehicle using a standard edge detection algorithm, it is attempted to detect the outer edges of the signal system at a known distance from the lane / track, until a good coverage has been found. There is then the searched signal system. II. Detection of whether a route continues to the LEFT or STRAIGHT AHEAD or RIGHT

i. The busy lane / the busy track is detected from step 2). ii. An imaginary lane / a planned track GERADEAUS can easily be generated artificially, since the well-known busy lane / the well-known track at the bottom of the picture must be updated only linearly.

iii. Then the traveled lane / track can be compared with the imaginary "STRAIGHT SEE lane" / with the imaginary "STRAIGHT OFF track".

1. Are the traffic lane / track and the imaginary "STRAIGHT SEE lane" / imaginary "STRAIGHT ELEMENT"

Track "are identical, the traffic lane / the befah ^¬ rene track leads straight.

2. If the busy lane / track LINKS is from the imaginary "STRAIGHT SEE lane" / the imaginary "STRAIGHT SEE track", then it is a left turn. From the Ab ^¬ state of the traffic track of the intended "straight ahead lane" / a "STRAIGHT-track" and the distance from the vehicle (imaginary be calculated from the width of the lane / the track at the lower edge and the width of the lane / the track in each relevant image section), the degree of left curvature can be calculated.

3. In the event that the traffic lane / track is located RIGHT from the imaginary "STRAIGHT SEA Lane" / imaginary "STRAIGHT EAST track", see previous point.

In addition to detecting whether a track continues to LEFT or STRAIGHT or RIGHT, there is also the case that switches are in the picture and these, of course, influence the further track. In this connection, reference is made to performing automatic recognition of lanes / tracks in rail / rail traffic according to German Patent Application (Application No. 102016224335.7) and International Patent Application (Application No. PCT / ...; Publication no. WO ...) and the technical teaching disclosed therein.

III. Detecting if people are on a lane / track i. The usual height of a human (from toddler (70 cm) to tall man (220 cm)) can be related to the lane / track width.

ii. Then the busy lane / track in the picture and a critical area to the left and right of the lane / track can be divided into several analysis sections with increasing distance from the vehicle. (The critical loading can ^¬ rich depending on the environment of the lane / the track (town ^¬ area vs. Country) different be broadly defined). Then, in the respective analysis section, a standard

Pattern algorithm used to identify persons who ^¬ .

1. In a further optimization, the analysis sections of several consecutive images can be related to detect movement of persons and track their movement to predict potential collisions. IV. Detecting how far a railway vehicle is from an object

i. Suppose an object is detected on a lane / track or in the critical area next to the lane / track.

ii. Then the pixel width of the lane / track has to

Height of the object to the pixel width of the lane / track at the bottom of the image. From the ratio and the known width of the lane / track, the distance to the object can be calculated.

V. Detecting if prohibited objects are on a lane / track

Objects that can be characterized ^¬ he recognized that in Step 2 of the algorithm for Kantenerkennungsalgo- lanes / tracks fails lying on the lane / track.

In a modification or extension, the regular pattern of the lane / between the tracks, the is formed by a lane carrier / track carrier, used as a basis for the detection of irregularities. If an irregularity is detected (eg Balise or

playing child), then in a second step the permissibility of the irregularity can take place by comparison with an image database. For example, In an initialization run, all irregularities can be recorded over all images of a route and then provided for adjustment.

By outlined above recognition of the lane / Glei ^¬ ses can be achieved that:

- Pre and main signals are more easily detected.

- The left or right curvature of a lane / a track can be detected.

- The detection of people on a lane / track is facilitated.

- The distance of a railway vehicle from objects / persons on a lane / track or in the critical area to the left or next to the lane / track can be calculated.

- The tracking of objects / people (tracking) in the critical image area the calculation of possible collisions is simplified.

- Analysis objects under unfavorable visibility conditions can be detected more reliably than by train drivers.

- driver no longer needed for the recognition of analysis items described so that it can be driven by the availability inde ^¬ pending.

In the course of an advantageous development of the invention, the following additional components - a) to c) for the image recording device (eg the image acquisition device) - may be used in relation to the image analysis device according to claim 11: a. A correction component according to claim 21, weather ^¬ and brightness data for the evaluation of the photographic material with includes. Thus, for example, in heavy fog, the evaluation of video images can be limited to the first 50 meters in front of the rail vehicle or rail vehicle and the speed of the vehicle can be throttled accordingly. b. A focal length variation component according to claim 22, depending on the environment (eg station, urban history ^¬ Biet, country, etc.) and the speed selector right up acceptance angle, so as to support the evaluation of the image perfectly. For example, both shooting scenes on the open road can then (need pictures from a distance to view reagie ^¬ ren due to the speed to be able to) and shooting situations in Bahnhofsbe- rich (need images with a high width) be operated suitable. c. A lighting component according to claim 23, ^¬ example, a headlight of the inside or outside of the human visible range lent operates, through which the Qua ^¬ formality of onsgerät from the image recording device or Bildakquisiti- at night or in bad weather captured image material improves. Moreover, it is possible that an image analysis device is formed as a virtual machine in the sense of "Software Defined signal recognition of Rail Traffic Systems" ^¬ and working. Other advantages of the invention will be apparent from the nachfol ^¬ constricting description of an embodiment of the invention with reference to FIGS 1 to 3. These show:

FIG. 1 shows a lane-based image analysis for the recognition of route installations on a railway line and route, FIGURE 2 is a lane-based image analysis for object detection of a residing on a lane of a railway person, FIGURE 3 shows a basic structure of an image analysis device for the lane-based according to FIGS 1 and 2 ^¬ image analysis.

FIG. 1 shows a lane-based image analysis for the recognition of a route system SAL, such as e.g. a railroad crossing BÜG,

Turnouts or a signal AL giving or leading system AL in railway traffic BVK, when on a railway section BST shown in sections of a railway network BNE a Bahnfahr ^¬ tool BFZ on a lane FS of the railway line BST of the signal system SI, AL approaches. In this lane based image ^¬ analysis for detection of the path system SAL and the signal SI-giving or leading plant AL, it is also, although not explicitly shown in FIGURE 1, a route, that the course of the lane FS what obligations RICH the Lane FS, as GERADEAUS, LEFT, RIGHT, anbe ^¬ applies to recognize.

According to the present embodiment, the lane ^¬ related railway line BST of the rail network BNE is a rail track SST of a rail network SNE, on which a rail vehicle SFZ on a rail GL and traveling in the rails ^¬ traffic SVK for a track-based image analysis to detect the track system SAL is the track system SAL on the track GL approaches. In place of the illustrated rail traffic SVK with that on the railway line SST of

Rail network SNE moving rail vehicle SFZ is here again due to the discussion at the outset any other arbitrary short- or long-distance based rail transport system as a further embodiment of the invention conceivable and conceivable. For example, a maglev system (Stw .: Transrapid, Maglev, etc.) would also have a comparable infrastructure. consisting of railway network, railway line and railway vehicle, in question.

In the rail transport system shown in FIG. 1, a picture analysis device BAV is provided in a railcar TRW of the rail vehicle SFZ with a driver's cab TFS and an integrated display ^device AZE, in which the workstation of the vehicle driver FZF is located, for the track-based image analysis for recognizing the route system SAL accommodated. The image analysis device BAV includes for this purpose a preferably as a sensor formed image on ^¬ drawing device BAZG, which for example is formed as an ordinary video camera, laser sensor, thermal imager, radar equipment, infrared camera, etc., and because of acquisition of pic- tures is also referred to as an image acquisition device.

With the image recording device BAZG, when the rail vehicle SFZ traveling on the track GL approaches the signal system SI, AL, it starts from the rail vehicle SFZ, eg from the perspective of the railcar driver FZF in the driver's seat TFS of the railcar TRW and / or from a stationary, lane-based position in or on the vehicle SFZ, from a track of the rail vehicle SFZ, thereby preferably oriented to the speed of the rail vehicle SFZ, track area GLB a plurality of the track area GLB representing images BI _GLB detectable.

Formulated more generally: With regard to the rail vehicle BFZ traveling on the lane FS, a plurality of images BI _FSB representing the traffic lane area _FSB can be detected by a lane area FSB which is located upstream of the rail vehicle BFZ and preferably orientates itself to the speed of the rail vehicle BFZ. The images BI _FSB , BI _{GLB of} the traffic lane area FSB, GLB each contain an image area BIB, in which lanes FS, GL detected in relation to the depicted traffic lane area FSB, GLB are substantially at the same location based on these portions of the detected lanes FS, GL with the aid of edge detection algorithms, the course of a lane FS, GL used by the railway vehicle BFZ, SFZ is detected by an image portion of the lane FS, GL changing in the acquired image to form the captured overall image.

In addition, a partial picture region BTB in the images BI _FSB, BI _GLB is for a currently determined geographical position of the railway vehicle BFZ, SFZ, in which the signal SI of the plant AL can be seen as a striking abutment feature of the routes ^¬ plant SAL. Again, this partial picture region BTB has a first image portion BAS1, by the used of the railway vehicle BFZ, SFZ lane FS, GL throughput leads, and a second image portion BAS2 wherein letz ^¬ more excellent for the route-detection, that is, the respective detecting a STRAIGHT-EAUS direction, LEFT-direction or

RIGHT direction of the lane FS, GL is relevant. As now based on the images BI _FSB , BI _glb of _Fahrspurbe ^¬ richs FSB, GLB with the image area contained therein BIB and the image subregion BTB with the two image sections BAS 1, BAS 2, the image analysis for distance system detection and the detection of the signal system SI, AL and track detection is explained below with the description of FIG.

FIGURE 2 shows a lane-based image analysis for Objekter ^¬ identifier in the rail BVK if on the off again cut as shown railway line BST of the rail network BNE the railway vehicle BFZ on the lane FS of the railway line BST at least one object OBJ, preferably a per ^¬ son, an animal, a movable or stationary object, is located in the area of the lane foreign and actually have to look for anything, so are there inadmissible.

The lane-related railway BST of the railway network BNE is here again the rail line SST of the rail network SNE, on which rail transport company SVK travels on track GL for rail-based image analysis for object recognition and approaches the object OGJ. In place of rail transport SVK illustrated with the vehicle running on the railway line SST of the rail network SNE rail vehicle SFZ any other x-any short or long distance based web transport system is here also conceivable as a white ^¬ teres embodiment of the invention, because of a ^¬ gangs guided discussion and imaginable. For example, a maglev system (Stw .: Transrapid, Maglev, etc.) with a correspondingly comparable infrastructure consisting of rail network, railway line and railway vehicle would also be suitable. In the rail transport system shown in FIG. 2, the image analysis device BAV is again provided in the railcar TRW of the rail vehicle SFZ with the driver's seat TFS and the integrated display device AZE, in which the workstation of the vehicle driver FZF is located, for the track-based image analysis for object recognition accommodated. The image analysis apparatus includes BAV this again preferably formed as a sensor Bil ^¬ daufzeichnungsgerät BAZG which, for example as ordinary Videoka ^¬ ra, laser sensor, thermal imager, radar device Inf- rarotkamera etc., is formed.

Are connected to the image recording apparatus BAZG when the running on the track GL rail vehicle SFZ nä ^¬ Hert the OBJ object of the rail vehicle SFZ from, for example, from the perspectives tive of the railcar leader FZF in the engine cab TFS the railcar TRW and / or from the fixed, lane ^¬ servierenden ^¬ position in or on the vehicle SFZ, of a the rail vehicle SFZ upstream, thereby preferably to the speed of the rail vehicle SFZ oriented, track area GLB a plurality of the track area GLB representing images BI _GLB detectable. More generally, are back in relation to the traveling on the lane FS railway vehicle BFZ from an upstream the railway vehicle BFZ, is preferably orienting the Ge ^¬ speed of the railway vehicle BFZ, lane area FSB a plurality of the lane area FSB representing images BI _FSB detectable.

In the images BI _FSB , BI _{GLB of} the traffic lane area FSB, GLB, the image area BIB is again contained, in which lanes FS, GL detected in relation to the depicted traffic lane area FSB, GLB lie substantially at the same location and starting from these proportions of the detected lanes FS, GL using the Kantenerkennungsalgo ^¬ algorithms of the course of the loading took advantage of the railway vehicle BFZ, SFZ lane FS, GL is detected by a changing in the captured image image portion of the lane FS, GL to the detected picture ,

In addition, a third image section BAS3 is contained in the images BI _FSB , BI _GLB , in which the object OBJ is ^¬ .

As now based on the pictures BI _FSB, BI _glb of Fahrspurbe ^¬ Reich FSB, GLB image analysis Objekter ^¬ detection is carried out with the contained image area BIB, and the image portion BAS 3, is explained below with the Be ^¬ scription of FIG. 3

FIGURE 3 shows the basic structure of the Bildanalysevor- direction BAV for the lane-based image analysis for Stre ^¬ ckenanlage- and route-detection according to the FIGURE 1 and for object recognition according to the FIGURE 2, when the rail vehicle BFZ, SFZ, which on the lane FS , GL is on the road, according to the FIGURE 1 of the track system SAL or the signal giving or leading system AL on the railway line

BST, SST or, according to FIG. 2, approaches the object OBJ located on the lane FS, GL. The starting point for the respective detection forms according to the embodiments of FIGS 1 and 2, the image recording apparatus BAZG that captures the images BI _FSB, BI _glb of Fahrspurbe ^¬ Reich FSB, GLB for these respective recognition.

The image recording device BAZG is preferably designed to be pivotable for alignment with the image object.

Furthermore, it is possible and possibly also useful for detection reasons that several image recording devices BAZG the same design, eg multiple video cameras, or devices un ^¬ ferent type, eg multiple video cameras, laser sensors, RADAR-based, based on radio-based location and distance ^¬ measurement sensors, infrared cameras and / or thermal imaging cameras, are included in the image analysis device BAV, which receive the images BI _FSB, BI _GLB. Such a multiple execution of the image recording or Bildakquirie- tion may be relevant for redundancy purposes, among others. To further improve the quality of the image recording device BAZG drawn on ^¬ or acquired images, preferably following items are included in the image recording device BAZG: 1. A correction component KOK be included with the weather and Hellig ^¬ keitsdaten for the evaluation of the photographic material. With this component it is possible, for example, to limit the evaluation of video images to the first 50 meters in front of the rail vehicle in heavy fog and to throttle the speed of the rail vehicle accordingly.

2. A focal length variation component BVK, in depen ^¬ dependence on the environment (eg station, city, country, etc.) and the speed chooses the right angle, so the evaluation of the image optimally under support alarm ^¬ zen. This allows both recording situations on the open road (need images from a long distance to respond in time due to the speed) as well as recording situations in the station area (need images with a high width) are suitably operated. ^¬ to additionally particularly interesting areas along the railway track in the rail network SST SNE can be focused by fusion of image data and link data, such as a railway crossing.

3. A lighting component BLK, which is formed for example as a spotlight operating inside or outside the human visual range, and by which the quality of the captured from the image recording device or the image acquisition apparatus BAZG at night or in bad Witte ^¬ tion picture material improves , The images thus captured are stored by the image recording device BAZG in an image storage device BSPE. This image storage device BSPE is either connected according to option "A" as a component of the image analysis device BAV with the image recording device BAZG or according to option "B" outside the image analysis device BAV, eg as a storage database, in the railcar or in a data cloud associated with the image recording device BAZG or .with this connectable. For the lane-based image analysis of the recorded or acquired images for distance system and route detection according to FIGURE 1 and for object detection according to FIGURE 2, the image recorder BAZG is connected to a calculation / evaluation BAWE, which is also a component of the image analysis device BAV is. For this purpose, the calculation / Auswerteeinrich- tung BAWE as the image recording apparatus BAZG either ge ^¬ Mäss with said image memory means BSPE option "A" connected to or assigned according to option "B" of the image memory means BSPE or connectable therewith. In this way, a functional subunit of the calculation / evaluation BAWE, the image recording device BAZG and the image storage device BSPE, in which the Components of the image analysis device BAV for the calculation- / evaluation-assisted and lane-based image analysis for track system / Streckenverlauf- / Obj ect recognition partially functional co-operate.

For the formation of a complete functional unit, in which the subunits involved functionally cooperate, said functional subunit is extended by a further subunit, an information database IDB. The information database IDB can thereby be beispielswei ^¬ se integrated with the image storage device BSPE as a structural unit in a common storage device. This is not explicitly shown in the Figure 3 memory ^¬ apparatus may in turn, as the image storage device BSPE, either in accordance with option "A" as a component of the image analysis device BAV with the image recording apparatus BAZG and the calculation / evaluation BAWE connected accordingly or in accordance with option " B "outside the image analysis device BAV in the railcar or in a data cloud the image recording device BAZG and the calculation / evaluation device BAWE assigned to or be connected to this. In this context, reference is made to the information storage device in DE patent application (application No. 102016224355.1) and the corresponding international patent application (application no. PCT / ...; publication No. WO alternatively determining positions in rail traffic when conventional satellite positioning is failing or inadequate.

In the information database IDB, image meta information BMI, which in the literal sense includes feature and property data of the traffic lane area FSB, GLB detected in the images BI _FSB , BI _GLB , is stored.

In this image meta-information BMI, primary metadata related both to the traffic lane FS, GL, and preferably the distance from parallel traffic lanes FS, GL and the width of a lane head, in particular a sliding ^¬ nenkopfes specify, as well as to the railway line BST, SST contain be ^¬ plated secondary metadata preferably specify which geographical positions or Ortskoordina- to plants th on the railway line BST, SST, the so-called track systems SAL, such as the level crossing BÜG in Figures 1 and 2, points or the signal SI giving or leading system AL are. As shown in FIGURE 3, the Informationsda ^¬ tenbank IDB of the image analysis apparatus BAV is assigned in such a way and with this connected, as the calculation / evaluation BAWE auswertegestützte for the calculation / and lane-based image analysis for Streckenanlage- / Streckenverlauf- / Obj ect recognition accesses the image meta information BMI stored in the information database IDB. The information database IDB is preferably outside the Bildanalysevorrich ^¬ tion BAV, eg as a database, arranged in the railcar or is designed as a data cloud.

For the calculation / evaluation-based and lane-based image analysis for route / route / object recognition, the calculation / evaluation device BAWE preferably has a non-volatile, readable memory SP in which processor-readable control program instructions of a picture analysis for route / route / Obj ect recognition-controlling program module PGM are stored, and a processor PZ, which executes the control program commands of the program module PGM for the calculation / evaluation-based and lane-based image analysis for track / route / Obj ekt recognition performs. PZ, the processor accesses in addition - in addition to the accesses to the image-BMI meta information in the information database IDB - for control purposes and for reading out data to the image recording device and the image BAZG spoke pure Rich ^¬ tung BSPE to. The calculation / evaluation device BAWE or the program module PGM with which the control program commands of the program module PGM for the calculation / evaluation-based and driving ^¬ track-based image analysis for Streckenanlage- / Streckenverlauf- / Obj ect-executing processor PZ is auswertegestützte for the calculation / and lane based Bildana ^¬ analysis for Streckenanlage- / Streckenverlauf- / Obj first formed ect-detection in the circumference of the functional part of the unit formed of the ^¬ art that in the images BI _FSB, BI _glb each of the image area BIB is identified, in with respect to the illustrated lane area FSB, GLB the detected lanes FS, GL lie substantially in the same place. In addition, the calculation / evaluation device BAWE or the program module PGM with the control program instructions of the program module PGM executing processor PZ within the scope of the functional unit formed such that, starting from the proportions of the detected lanes FS, GL in the image area BIB detected with the aid of Kantenerkennungsal ^¬ algorithms the course of a lane FS, GL used by the rail vehicle BFZ, SFZ by a changing in the captured image image portion of the lane FS, GL to the detected picture and compared with data stored in the information database IDB known image meta information BMI becomes.

For trackside detection of the track system SAL or the signal SI giving or leading system AL the lane-based image analysis in the calculation / evaluation BAWE advantageously such that in the image BI _FSB , BI _{GLB is} recognized where the Signaling system SI, AL or the track system SAL can be found by

1) for a currently determined geographical position of the railway vehicle BFZ, SFZ the image subregion BTB in the image

BI _FSB , BI _{GLB is} identified, in which a distinctive Anlagenmerk ^¬ times the track system SAL or the signal SI of the system AL can be seen, 2) the first image section BAS1 is determined in the image subregion BTB, through which the traffic lane FS, GL used by the railway vehicle BFZ, SFZ performs,

3) is determined by means of a first known metadata about the distance of the route installation SAL or the signal installation SI, AL of the secondary metadata in the stored image meta information BMI, as far as on the basis of a width of the traffic lane FS, GL in the Image section BAS1 line system SAL or the signal system SI, AL is separated from the lane FS, GS,

4) by means of a second known metadata item on the Au ^¬ ßenkanten the track system SAL or the signal conditioning Si, Al secondary metadata in the stored image

Meta information BMI and a standard edge detection algorithm is attempted to detect the outer edges of the track system SAL or the signal system SI, AL in the distance 3) determined by the lane FS, GL until the known and recognized outer edge substantially überde ^¬ bridges ,

For route course recognition, whether a travel path leads to the left or right or straight ahead, that is to say in which direction the traffic lane FS, GL recognized from the recognized route leads, the calculation / evaluation device BAWE in the lane-based image analysis continues to be such in an advantageous manner designed that in the image BI _{FSB /} BI _{GLB is} detected, in which direction, in particular LEFT, RIGHT or STRAIGHT ^¬ OFF, the railway vehicle BFZ, SFZ drives by for the detected from the detected course lane FS, GL

1) a pseudo-GERADEAUS lane, in particular a pseudo-GERADEAUS track, is artificially generated by linear updating of the lane FS, GL used by the railway vehicle BFZ, SFZ in accordance with the image area BIB,

2) the pseudo-straight-ahead lane is compared with the used lane FS, GL in such a way that

2.1) is detected a straight-ahead direction when the lanes were comparing ^¬ NEN FS, GL are identical, 2.2) a LEFT direction, eg a LEFT curve, is recognized when the used lane FS, GL is to the left of the pseudo-GERADEAUS lane,

2.3) a RIGHT direction, e.g. a right-hand turn, is recognized when the used lane FS, GL is to the right of the pseudo-straight lane.

In this case, it is possible in particular and the calculation / evaluation device BAWE designed accordingly that from the distance of the used lane FS, GL from the pseudo

STRAIGHT LEVEL lane and from the railway vehicle BFZ, SFZ, which consists of a width of the lane FS, GL in the image area BIB and a width of the lane FS, GL respectively in the relevant for the respective detection of the direction straight direction, LEFT direction or RIGHT direction the second image section BAS2 is calculated, a degree of curvature of the LEFT direction / curve or the RIGHT direction / curve is calculated.

For object recognition, calculation / evaluation device BAWE in the case of lane-based image analysis is furthermore advantageously designed in such a way that in the image

BI _FSB , BI _{GLB is} detected, whether at least one object OBJ, which may be, for example, a person, an animal, a movable or stationary object on the used lane FS, GL by

1) a personal size of the person and / or a ge ^¬ genstandbezogen size of the movable object in the Ver ^¬ ratio to a known width of the lane FS, GL is set from a third known metadata on the width of the lane FS, GL of the primary Metadata is obtained in the stored image meta information BMI,

2) in the image BI _FSB , BI _GLB the used lane FS, GL and a critical area, for example, according to an environment of the lane FS, GL, as there are country or city, of different widths, left and right of the used Lane FS, GL are divided into several analysis sections with increasing Ab ^¬ stand for railway vehicle BFZ, SFZ, ^¬ uses will be 3) for each analysis section an environment-specific pattern recognition algorithm to detect the object OBJ. In particular, it is possible and the calculation

/ Evaluation device BAWE trained accordingly that when a moving object OBJ, for example, a person, an animal, other oncoming or preceding train vehicles, etc., the analysis sections of several successive images BI _FSB , BI _{GLB be} set to the movement for to recognize and to follow the precalculation of a possible collision with the object OBJ.

In addition, it is preferably possible and the calculation / evaluation device BAWE designed accordingly that in the image BI ^ _SB ^ BI _{GLB is} detected, how far the railway vehicle BFZ, SFZ is spaced from the object OBJ on the used lane FS, GL has been recognized by the one of a pixel width of the lane FS, G) in the image area BIB and a pixel width of the lane FS, GL in the third

Image section BAS3, in which the object OBJ is located, are set in relation to each other, from this ratio and the obtained from the known third metadata width of the lane FS, GL of said distance is calculated.

With the above-described image analyzing device BAV an automated (autonomous) or assisted driving of the railway vehicle or the railway vehicle BFZ SFZ without additional infrastructure along a route as may ^¬ be suspended or even realized. This is especially true if the image analysis device BAV is realized as a virtual machine which is designed and functions in the sense of a " ^software- defined signal recognition of rail traffic system".

Claims

claims

1. A method for lane-based image analysis in railway traffic (BVK), in particular for track-based image analysis in rail traffic (SVK),

characterized in that

a) from a railway vehicle (BFZ), in particular a rail vehicle (SFZ), in particular from the perspective of a railcar driver (FZF, TFS, TRW) and / or from a stationary, lane-servicing position in or on the vehicle

(BFZ, SFZ), a plurality of a lane area (FSB), in particular a track area (GLB), representing images (BI _FSB , BI _glb ) is detected,

b) in the images (BI _FSB , BI _GLB ) in each case an image area (BIB) is identified, in which in relation to the pictured ^¬ lane area (FSB, GLB) detected lanes (FS), in particular tracks (GL), im Essentially in the same place,

c) starting (the proportions of the detected lanes FS, GL) (in the image region BIB) used with the aid of edge detection ^¬ algorithms the course of a (of the railway vehicle BFZ, SFZ) lane (FS, GL) by a in the captured image detected changing image portion of the lane (FS, GL) for sensing ^¬ th picture and having stored known picture meta information (BMI) is matched to the traffic lane (FS, GL) related primary metadata, in particular the distance of parallel lanes ( FS, GL) and the width of a lane head, in particular a rail head, or as well as to a railway line (BST), in particular to a rail line (SST), related secondary metadata, in particular ^¬ at which geographical positions or location coordinates are plants on the railway line (BST, SST), so-called track systems (SAL), such as railroad crossings, switches or signaling or leading facilities, are included.

2. The method according to claim 1, characterized in that in the image (BI _FSB , BI _GLB ) is detected where the track system (SAL) can be found by

a) for a currently determined geographical position of the railway vehicle (BFZ, SFZ) an image subregion (BTB) in the image (BI _FSB , BI _GLB ) is identified, in which a distinctive feature ^¬ genmerkmal, in particular a signal (SI) of the signaling system, it can be seen

b) a first image section (BAS1) is determined in the image subregion (BTB) through which the traffic lane (FS, GL) used by the railway vehicle (BFZ, SFZ),

c) is determined using a first known metadata about the Ab ^¬ stand of the track system (SAL) of the secondary metadata in the stored image meta-information (BMI) how far on the basis of a width of the lane (FS, GL) in the image section (BAS1) the track system (SAL) from the Fahr ^¬ track (FS, GL) is spaced,

d) using a second known metadata item on the Au ^¬ ßenkanten the track system (SAL) of the secondary metadata (in the stored image meta information BMI) and a standard edge detection algorithm is attempted, the outer ^¬ edges of the path system (SAL) in the in c ) detected distance from the lane (FS, GL) so long until the known and recognized outer edge substantially überde ^¬ bridges.

3. The method according to claim 1 or 2, characterized in that

in the image (BI _FSB , BI _GLB ) it is detected in which direction, in particular LEFT, RIGHT or STRAIGHT, the railway vehicle (BFZ, SFZ) travels, by checking the lane (FS, GL) identified from the detected course

a) a pseudo-GERADEAUS lane, in particular a pseudo-GERADEAUS track, is artificially generated by linear updating of the lane (FS, GL) used in accordance with the image area (BIB) by the railway vehicle (BFZ, SFZ),

b) the pseudo-straight-ahead lane is compared with the used lane (FS, GL) in such a way that bl) is detected a straight-ahead direction when the lanes were comparing ^¬ NEN (FS, GL) are identical,

b2) a LEFT direction, in particular a LEFT curve, it is ^¬ known is when the used lane (FS, GL) is to the left of the pseudo-GERADEAUS lane,

b3) a RIGHT direction, in particular a RIGHT curve, it is ^¬ known is when the used lane (FS, GL) to the right of the pseudo-GERADEAUS lane.

4. The method according to claim 3, characterized in d a s s

from the distance of the used lane (FS, GL) from the pseudo-GERADEAUS lane and from the railway vehicle (BFZ, SFZ), which consists of a width of the lane (FS, GL) in the image area (BIB) and a width of the lane (FS , GL) in each case for a respective second image section relevant for the respective recognition of the STRAIGHT-EAUS direction, LEFT-direction or RIGHT-direction

(BAS2) is calculated, a degree of curvature of the LEFT direction / curve or the RIGHT direction / curve is calculated.

5. The method according to any one of claims 1 to 4, characterized in that

is recognized in the image (BI _FSB, BI _G LB) whether at least one object (OBJ), preferably a person, an animal, a Move ^¬ Barer or fixed object on the used driving lane (FS, GL) is by

a) a personal size of the person and / or a ge ^¬ genstandbezogen size of the movable object in relation to a known width of the lane (FS, GL) is set, from a third known metadata across the width of the lane (FS, GL) of the primary metadata is obtained in the stored image meta-information (BMI), b) in the image (BI _FSB , BI _GLB ) the used lane (FS, GL) as well as a critical area, preferably according to an environment of the traffic lane (FS, GL), for example country or city is different widths sized left and right of the ^¬ be used lane (FS, GL) into several sections with analysis increasing distance to the railway vehicle (BFZ, SFZ),

c) an environment-specific pattern recognition algorithm for recognizing the object (OBJ) is used for each analysis section.

6. The method according to claim 5, characterized in that a s s

at a moving object (OBJ), e.g. Moving person, moving animal, other oncoming or preceding train vehicles, etc., the analysis sections of several successive images (BI FSB / BIGLB) are related to each other in order to calculate the movement for the precalculation of a possible collision with the object (OBJ). to recognize and pursue.

7. The method according to claim 5 of 6, characterized in that

in the image (BI _FSB , BI _GLB ) it is detected how far the railway vehicle (BFZ, SFZ) is from the object (OBJ), which has been recognized on the used lane (FS, GL), by the a pixel width of the lane (FS, GL) in the Bildbe ^¬ rich (BIB) and a pixel width of the lane (FS, GL) in a third image section (BAS3), in which the object (OBJ) is located in relation to each other, wherein from said ratio and the width of the lane (FS, GL) obtained from the known third metadata, said distance is calculated.

8. The method according to any one of claims 1 to 7, characterized in that

in the image (BI _FSB , BI _GLB ) is detected, whether on the used lane (FS, GL) of the railway line (BST, SST) is at least one foreign object (OBJ) by

a) the detection of the course of the lane (FS, GL) used by the railway vehicle (BFZ, SFZ) using the edge detection algorithms fails or b) an irregularity detection is carried out on the basis of regular patterns formed by lane markers of the traffic lane (FS) or track carriers between the parallel tracks (GL) in such a way that when an irregularity is detected, preferably when an object (OBJ), for example, a beacon or a playing child, on the lane used (FS, GL), this irregu ^¬ temperance is compared with reference information used previously recorded and in driving stretch-Initialisierungsläufen route images.

9. The method according to any one of claims 1 to 8, characterized in that

the images (BI _FSB, BI _G LB) having a plurality of image recording devices (BAZG) of the same type, for example with video cameras, or ^¬ schiedlicher type, such as video cameras, laser sensors, RADAR-based, based on radio-based positioning and distance measuring sensors, infrared cameras and / or Thermal Imaging ^¬ ras, are recorded.

10. The method according to any one of claims 1 to 9, characterized in that

is assisted by an automated process (autonomous) or assisted ^¬ driving the railway vehicle (BFZ, SFZ) without additional infrastructure along a route.

11. Apparatus (BAV) for lane-based image analysis in rail traffic (BVK), in particular for track-based image analysis in rail traffic (SVK),

marked by

a) at least one image recording device (BAZG) with which of a railway vehicle (BFZ), in particular a rail vehicle (SFZ), in particular from the perspective of a railcar ^¬ driver (FZF, TFS, TRW) and / or from a stationary, driving track-occupying position in or on the vehicle (BFZ, SFZ), a plurality of images representing a traffic lane area (FSB), in particular a track area (GLB), (BI _FSB , BI _GLB ) can be detected and stored in an image storage device (BSPE),

b) a calculation / evaluation device (BAWE) connected and functionally cooperating with the image recording device (BAZG) and the image memory device (BSPE) in such a way, in particular with a non-volatile, readable memory (SP), in the processor-readable control program instructions the lane-based image analysis controlling program module (PGM) are stored, and a processor (PZ), which executes the programmer module (PGM) control program instructions for calculation / evaluation-assisted and lane-based image analysis, that in the images ( BI _FSB , BI _GLB ) je ^¬ Weils an image area (BIB) is identified, in which in relation to the illustrated lane area (FSB, GLB) detected lanes (FS, GL) are located substantially in the same place,

c) the calculation / evaluation device (BAWE), the connected and functionally cooperating with a Informationsdaten ^¬ bank (IDB), preferably both, the image storage device (BSPE) and the information database (IDB), as a structural unit in a shared memory device are integrated, is formed such that, starting from the proportions of the detected lanes (FS, GL) in the image area (BIB) by means of edge detection algorithms of a lane used by the railway vehicle (BFZ, SFZ) lane (FS, GL) by a in the detected image changing image portion of the lane (FS, GL) is detected to the captured total image and compared with stored in the information database (IDB) known image meta-information (BMI), both the lane (FS, GL) related primary metadata, in particular the distance from parallel running lanes (FS, GL) and the width of a lane head, in particular one

Railhead, as well as to a railway line (BST), in particular ^¬ special rail line (SST), related secondary metadata, in particular at which geographical positions or location coordinates are facilities on the railway line (BST, SST), so-called track systems (SAL), like railroad crossings, Contain diverters or signaling or leading installations.

12. Device (BAV) according to claim 11, characterized in that

the calculation / evaluation device (BAWE) is designed in such a way that in the image (BI _{FSB /} BI _GLB ) it is recognized where the track system (SAL) can be found by

a) for a currently determined geographical position of the railway vehicle (BFZ, SFZ) an image subsection (BTB) in the

Image (BI _FSB , BI _GLB ), in which a striking feature ^¬ genmerkmal, in particular a signal (SI) of the signal system can be seen,

c) is determined using a first known metadata about the Ab ^¬ stand of the track system (SAL) of the secondary metadata in the stored image meta-information (BMI) how far on the basis of a width of the lane (FS, GL) in the image section (BAS1) the track system (SAL) from the Fahr ^¬ track (FS, GS) is spaced,

d) on the Au ^¬ ßenkanten the track system (by means of a second known metadata item SAL) of the secondary metadata (in the stored image meta-BMI) and a

Standard edge detection algorithm is attempted to recognize the outer ^¬ edges of the track system (SAL) in the distance determined in c) from the lane (FS, GL) until the known and recognized outer edge substantially covers.

13. Device (BAV) according to claim 11 or 12, characterized in that

the calculation / evaluation device (BAWE) is such trained det that is recognized in the image (BI _{FSB /} BI _GLB), runs in wel ^¬ che direction, in particular left, right or straight on the rail vehicle (BFZ, SFZ), by identifying the lane (FS, GL) identified from the detected course a) a pseudo-GERADEAUS lane, in particular a pseudo-GERADEAUS track, is artificially generated by linear updating of the lane (FS, GL) used by the railway vehicle (BFZ, SFZ) according to the image area (BI);

b) the pseudo-straight-ahead lane is compared with the used lane (FS, GL) in such a way that

bl) is detected a straight-ahead direction when the lanes were comparing ^¬ NEN (FS, GL) are identical,

b2) a LEFT direction, in particular a LEFT curve, is recognized when the used lane (FS, GL) is to the left of the pseudo-STRAIGHT-EASE lane,

14. Device (BAV) according to claim 13, characterized in that

the calculation / evaluation device (BAWE) is designed such that from the distance of the used lane (FS, GL) from the pseudo-GERADEAUS lane and from the railway vehicle (BFZ, SFZ), which consists of a width of the lane (FS, GL ) in the image area (BIB) and a width of the lane (FS, GL) is calculated in each case (a relevant for the respective detection of the straight-ahead direction, left direction or right direction second image portion BAS2), a curve ^¬ degree of LINKS Direction / curve respectively the RIGHT direction / curve is calculated.

15. Device (BAV) according to one of claims 11 to 14, characterized in that

the calculation / evaluation device (BAWE) is designed such that it is recognized in the image (BI _{FSB /} BI _GLB ) whether at least one object (OBJ), preferably a person, an animal, a movable or stationary object, on which used lane (FS, GL) by

a) a personal size of the person and / or a ge ^¬ genstandbezogene size of the movable object in the Ver ^¬ ratio (to a known width of the lane FS, GL) Sex is set, which is obtained from a third known metadata about the width of the lane (FS, GL) of the primary metadata in the stored image meta-information (BMI), b) in the image (BI _FSB , BI _GLB ) the lane used (FS, GL) as well as a critical region, which preferably after an environment of the lane (FS, GL), for example, country or city, different width dimensions, left and right of the be ^¬ used lane (FS, GL) into a plurality of analysis portions be classified with increasing distance to the railway vehicle (BFZ, SFZ),

c) an environment-specific pattern recognition algorithm to detect the object (for each analysis section uses ^¬ be OBJ) is.

16. Device (BAV) according to claim 15, characterized in that

the calculation / evaluation device (BAWE) is designed such that with them a moving object (OBJ) such as a person, an animal, other oncoming or forward-propelled railway vehicles, etc., the analysis of sections of several ^¬ ren successive images (BI _FSB , BI _G LB) are set in relation to one another in order to recognize and track the movement for the precalculation of a possible collision with the object (OBJ).

17. Device (BAV) according to claim 15 or 16, characterized in that

the calculation / evaluation device (BAWE) is designed in such a way that in the image (BI FSB _/ BI GLB) it is detected how far the rail vehicle (BFZ, SFZ) is from the object (OBJ) that is located on the used lane ( FS, GL) has been recognized by

the group consisting of a pixel width of the lane (FS, GL) in Bildbe ^¬ rich (BIB) and a pixel width of the lane (FS, GL) in a third image section (BAS3), in which the object (OBJ) is set in relation to each other from this ratio and from the known third meta- Date obtained width of the lane (FS, GL) of said distance is calculated.

18. Device (BAV) according to one of claims 11 to 17, characterized in that

the calculation / evaluation device (BAWE) is designed in such a way that in the image (BI FSB _/ BI GLB) it is detected whether at least one foreign object (OBJ) is present on the used lane (FS, GL) of the railway line (BST, SST) lies by

a) the detection of the course of the lane (FS, GL) used by the railway vehicle (BFZ, SFZ) using the edge detection algorithms fails or

b) an irregularity detection is carried out on the basis of regular patterns formed by lane markers of the lane (FS) or track carriers between the parallel tracks (GL) in such a way that when an irregularity is detected, preferably when an object (OBJ ), for example is a beacon or a playing child, on the shared lane (FS) respectively (the track used GL), this irregularity used as a reference ^¬ information and traveling route initialization runs previously recorded track images is adjusted.

19. Device (BAV) according to one of claims 11 to 18, characterized in that

several image recording devices (BAZG) of the same design, eg several video cameras, or of different types, eg several video cameras, laser sensors, RADAR-based sensors based on radio-based location and distance measurement, infrared cameras and / or thermal imaging cameras, which contain the images (BI _FSB , BI _GLB ) record.

20. Device (BAV) according to one of claims 11 to 19, characterized in that

the image recording device (BAZG) is designed to be pivotable.

21. Device (BAV) according to one of claims 11 to 20, characterized in that the image recording apparatus (BAZG) a correction component (COC), which includes in the evaluation of the photographic material Wet ^¬ ter- and brightness data.

22. Device (BAV) according to one of claims 11 to 21, characterized in that

the image recorder (BAZG) has a focal length Variegated ^¬ insurance component (BVK), which depending on the From ^¬ stood for lane (FS, GL) chooses the right angle, so the multiple evaluation optimally under support alarm ^¬ zen.

23. Device (BAV) according to one of claims 11 to 22, characterized in that

the image recording device (BAZG) has a lighting component (BLK), in particular a headlight which operates inside or outside the human visible region.

24. Device (BAV) according to one of claims 11 to 23, characterized by

a virtual machine that is in the sense of a "software

Is defined signal recognition of Rail Traffic Systems out forms ^¬ "and working.

25. Device (BAV) according to one of claims 11 to 24, characterized in that

Automated (autonomous) or assisted driving of the railway vehicle (BFZ, SFZ) without additional infrastructure along a route can be assisted with the device (BAV).

26. Railway vehicle (BFZ) for the lane-based image analysis in railway traffic (BVK), in particular for track-based image analysis in rail traffic (SVK), characterized in that

A device (BAV) for lane-based image analysis or for track-based image analysis, according to one of the Proverbs 11 to 25 in the railway vehicle (BFZ, SFZ) is integrated.