EP4195176A1 - Method for identifying vehicles and vehicle identification device - Google Patents

Abstract

A method for identifying vehicles, comprising the steps of (i) acquiring magnetometer data from (a) a first magnetometer, (b) a second magnetometer, (c) a third magnetometer, (d) a fourth magnetometer, (e) a fifth magnetometer and (f) at least one sixth magnetometer, (g) at least three magnetometers being arranged along a first straight line and forming a first magnetic line sensor, (h) at least three magnetometers being arranged along a second straight line and forming a second magnetic line sensor, wherein the second straight line runs along the first straight line, so that a magnetic field data set is obtained which describes at least one spatially resolved magnetic field caused by the vehicle in at least two dimensions, and (ii) identifying a vehicle type and/or a vehicle using the magnetic field data set.

Description

The invention relates to a method for the contactless identification of vehicles. According to a second aspect, the invention relates to a vehicle identification device for identifying vehicles, having (a) a first magnetometer for acquiring first magnetometer data, (b) a second magnetometer for acquiring second magnetometer data, (c) a third magnetometer for acquiring third magnetometer data, (d) a fourth magnetometer for acquiring fourth magnetometer data, (e) a fifth magnetometer for acquiring fifth magnetometer data, (d) at least one sixth magnetometer for acquiring sixth magnetometer data and (e) an evaluation circuit for automatically acquiring the magnetometer data and identifying a vehicle or vehicle type based on the magnetometer data,

Devices for identifying motor vehicles are usually based on optical recognition of number plates or other characteristic optical features. Another possibility is the wireless reading of radio transponders attached to the vehicle. Such systems are used, for example, for toll collection, parking lots, multi-storey car parks or in logistics and fleet management.

Optical systems have the disadvantage that sufficient natural or artificial lighting must be available. Even with additional lighting equipment, incorrect detection can occur in unfavorable lighting conditions.

The identification of vehicles based on radio transponders is independent of light or weather conditions, but requires the attachment of a corresponding transponder to the vehicle. While in optical identification For example, features that are already present in motor vehicles, such as license plates, can be used for identification, corresponding transponders are usually not installed on the vehicle at the factory.

Both of the systems described can be deceived by theft or copying of license plates or transponders.

The WO 2013/044389 A1 describes a method for identifying vehicles using vector-valued magnetic field measurements with at least one or more magnetometers spaced perpendicularly to the expected direction of travel. A scale-invariant dipole model for the magnetic signature of a vehicle is used for identification. The accuracy of the identification depends on the quality of the magnetic field approximation. However, a dipole approximation, especially in the near field, is only a rough approximation for the complex magnetic field of a vehicle, which is composed of several magnetized components of different geometries. Therefore, deliberate manipulation by attaching simple permanent magnets to the vehicle, which generate the desired dipole field, is particularly possible. Furthermore, the magnetic signatures can only be viewed individually, so that information from spatially resolved magnetic field measurements is not used. In particular, the method is designed in such a way that it should also work with only one vector-valued magnetometer. Furthermore, the system is not invariant to an oblique crossing.

The WO 2019/155324 describes a system of at least two magnetometers placed along the expected direction of travel that uses cross-correlation to determine the speed of an overrun vehicle. Similar approaches are in the U.S. 6,208,268 and U.S. 2013 057264 described. However, these systems cannot identify or reliably distinguish between vehicle types and individual vehicles.

Some systems are also able to provide an estimate of the vehicle class (e.g. truck, car), e.g. B. with the purpose of excluding incorrect detections due to interference or other magnetic objects or to carry out speed measurements depending on the type of vehicle.

The WO 2013/189985 describes a method that can determine the vehicle class and the speed of the passing vehicle using a vector-valued magnetometer. For this purpose, at least part of the time-resolved magnetic field signal B(t) is compared with reference signals in a database.

In the identification of vehicles using a magnetic signature, the stretching or compression of the signature due to different crossing speeds of vehicles, as well as the distortion of the signature due to an oblique crossing, are particularly problematic. In previous systems, the additionally required information was provided by external sensors, e.g. B. radars, light barriers or pressure sensors are provided.

Therefore, previous magnetic field-based systems for vehicle identification either require mathematical models based on a modeling of the magnetic signature generated by a vehicle, or they use a reduced set of invariant characteristics, e.g. B. minima and maxima of the signature, for a comparison. Both with the model-based approach and with the reduced set of characteristic properties, there are ambiguities or inaccuracies in the identification due to the assumptions made or the insufficient information content. Due to these ambiguities and inaccuracies, targeted manipulation is potentially possible, especially with knowledge of the modeling processes, which are usually published in the context of patent specifications or approval processes.

The object of the invention is to reduce the disadvantages of the prior art.

The invention solves the problem by means of a generic method in which (a) at least three magnetometers are arranged along a first straight line and form a first magnetic line sensor, and (b) at least three magnetometers are arranged along a second straight line and one form the second magnetic line sensor, the second straight line running along the first straight line. By capturing the magnetic field data, a magnetic field data set is obtained that contains at least one in describes at least two dimensions spatially resolved magnetic field caused by the vehicle. In addition, the method includes the step of identifying a vehicle type using the magnetic field data set.

According to a second aspect, the invention solves the problem with a generic vehicle identification device in which (h) at least three magnetometers are arranged along a first straight line and form a first magnetic line sensor, (i) at least three magnetometers are arranged along a second straight line and form a second magnetic line sensor. The second straight line preferably runs along the first straight line. The magnetic field data describes a change over time in a static magnetic field generated by a vehicle driving over the magnetometers. The evaluation circuit is designed to automatically carry out a method according to the invention.

According to the invention is also a building with (a) a lane for vehicles, the lane having a lane travel direction, and (b) a vehicle identification device according to one of the preceding claims, which is arranged to identify a vehicle type of vehicles driving on the lane . The vehicle identification device can be arranged above the roadway, below the roadway or embedded in the roadway.

In the context of the present description, a magnetometer is understood to mean, in particular, a thin-film magnetometer. A thin-film magnetometer is understood to mean sensors for detecting the magnetic field, the structure of which is layered and sensitive to the magnetic field. In particular, a thin-film magnetometer is based on the GMR (giant magnetoresistance), AMR (anisotropic magnetoresistive) or TMR (magnetic tunneling resistance) effect. The advantage here is the low overall height made possible by the use of thin-film magnetometers, combined with high sampling rates and low measurement uncertainties.

Preferably, at least one magnetometer, at least a majority of the magnetometers and particularly preferably all magnetometers are designed to measure at least two, in particular three, components of the magnetic field. A sampling rate is preferably more than 50 Hz, in particular more than 75 Hz, and/or at most 100 kilohertz.

Preferably, at least one magnetometer, at least a majority of the magnetometers, and most preferably all magnetometers have a resolution of 1000 nanotesla or better, preferably 100 nanotesla or better.

The feature that the magnetometers are arranged along a straight line means in particular that it is advantageous, but not necessary, for the magnetometers to lie on a straight line in the strict mathematical sense. It is possible that the actual position of the magnetometer deviates from the ideal position, for example by a maximum of 20 cm, in particular a maximum of 15 cm, preferably a maximum of 10 centimeters.

The feature that at least six magnetometers are present means in particular that it is favorable, but not necessary, for further magnetometers to be present. A larger number of magnetometers increases the spatial resolution, but at the same time increases the cost and complexity of the setup. In the general case, n magnetometers are used, in particular with n < 300.

Each magnetometer measures magnetic field data that describes a magnetic field caused by a vehicle at the location of the magnetometer as a function of time. The magnetic field can be caused by magnetized parts of the vehicle. Alternatively or additionally, the magnetic field can be caused by a change in the earth's magnetic field by non-magnetized but ferromagnetic components.

Since the Earth's magnetic field can usually be assumed to be homogeneous, the magnetic field data describe the respective instantaneous influence of the magnetization and/or the susceptibility of the vehicle driving over the respective magnetic field on the magnetic field at the location of the magnetometer as a function of time.

Vehicles are understood to mean, in particular, land vehicles, for example passenger cars and trucks.

Although the magnetic fields detected by the magnetometer are time-dependent, they do not have a carrier frequency. In other words, the magnetometers are not receivers of electrogenetic waves actively emitted by a transmitter.

Preferably, at least three of the at least six magnetometers are located at a distance of between 2 cm and 20 cm from each other approximately along the direction of travel and the remaining magnetometers are displaced parallel to or against the direction of travel at a distance of typically between 0.01 m and 1 m. The spatially separated magnetometers thus form two mutually shifted magnetic line sensors, which together represent a measuring unit.

The line sensors of this measuring unit are preferably arranged in a common housing in order to ensure a constant and known distance from one another. This measuring unit is mounted in, on or above the ground, in particular a roadway.

A vehicle type is understood to mean, in particular, a group of vehicles with the same external geometry and/or the same permissible total mass interval. For example, a vehicle type can be formed by all those vehicles that are allowed to pass a route. Preferably, not only vehicle types but also individual vehicles are identified using the magnetic field data set.

The evaluation circuit is understood to mean, in particular, an electronic circuit which automatically, ie without human intervention, identifies the vehicle type and/or a vehicle on the basis of the magnetic field data set. The identification of the vehicle type and/or the vehicle preferably includes the calculation of a magnetic signature from the magnetic field data set. A magnetic signature is a data record that encodes the magnetic properties of a vehicle in such a way that it is possible to assign the data record to the vehicle unambiguously, in particular one-to-one.

The identification of the vehicle type and/or the vehicle preferably also includes comparing the calculated magnetic signature with stored reference signatures and determining a similarity parameter. encoded If this similarity parameter has a sufficiently high similarity, the vehicle is identified as the vehicle to which the reference signature is assigned.

A unique assignment of the signature to the vehicle means that each vehicle has exactly one magnetic signature. A one-to-one assignment means that each magnetic signature is assigned to exactly one vehicle. A one-to-one mapping is a one-to-one mapping. It is possible that there is an unambiguous or one-to-one association in the strictly mathematical sense. However, the unambiguous or one-to-one association is to be understood in the technical sense and is preferably also present when any deviation from the strictly unambiguous or one-to-one association is so rare that it can be ignored.

It is possible, but not necessary, for the evaluation circuit to be enclosed by the housing which also encloses the magnetometer. The evaluation circuit can be connected to the measuring unit either by cable or wirelessly.

The recorded magnetometer data measurement results and/or a magnetic signature calculated from them is preferably compared with reference signatures in a database in order to trigger an action directly, or forwarded to another EDP system (e.g. programmable logic controller or computer) for further processing.

The magnetic signature of vehicles is primarily generated by various components of the vehicle, mostly by remanent magnetization. Most conventional motor vehicles, including modern electric vehicles, have numerous magnetizable and partially magnetized materials, for example in the chassis, in the body, in engines or transmissions. These are magnetized in various ways during production, especially during casting, machining or forming, whereby the magnetization process of the individual components depends on a sufficient number of factors (temperature, magnetic background field, magnetic fields from machines, type of processing, etc.) , resulting in an almost unique magnetic signature.

It is theoretically possible for two vehicles to have an identical signature. However, since this case is very unlikely, it is referred to as a unique or quasi-unique signature, as is the case with human fingerprints.

This quasi-unique magnetic signature of the vehicle is used for identification. The measurement result of this magnetic signature also depends on the magnetic field at the measurement location. With a suitable choice of the measurement location, which has a homogeneous geomagnetic field and no strong external magnetic fields, the measurement of the magnetic signature of the vehicle is, on the other hand, independent of the measurement location to a sufficiently good approximation.

An advantage of the invention is that the magnetic signature used for identification can hardly be copied or forged because it is linked to the magnetization of integral components of the vehicle. Even with precise knowledge of a desired signature, the corresponding parts of the vehicle cannot be remagnetized with justifiable effort. In other words, while it is possible to alter a vehicle's magnetic signature, it is largely impossible to do so in a targeted manner to obtain a predetermined magnetic signature. This is particularly advantageous for access controls.

By using thin-film magnetometers, measuring units with a typical height of less than 2 cm can be implemented, with the length of the unit depending on the width of the vehicles to be identified and the required accuracy, and typically between 2.75 m and 3.75 m amounts to. Due to the low overall height, the invention can either be installed in the roadway, similar to the induction loops or pressure sensors already customary in traffic engineering, or mounted directly on the subsurface, in particular the roadway. It can also be mounted on trusses above the roadway.

From the magnetometer data recorded by the magnetometers, magnetic field data can be calculated which describe the magnetic field under the vehicle in a spatially resolved manner (in particular not as a function of time).

The characteristic magnetic signature for a given position on the vehicle results as a whole from the measured, time-resolved magnetic field data recorded by at least six magnetometers. The magnetic signature may be the magnetic field data or data extracted from the magnetic field data describing characteristics of the magnetic field data.

When a vehicle passes the measuring unit located across the route, the vehicle's magnetic signature is scanned. The beginning and end of the vehicle are preferably detected based on the characteristic course of the magnetic field strength.

Due to the known positions of the line sensors relative to one another, a uniform coordinate system k ₁ related to the line sensors can be used for both. In order to obtain the magnetic signature, the magnetic field data are converted into a coordinate system k ₂ related to the vehicle. This eliminates the orientation of the vehicle relative to the line sensors. Any tilting of the line sensors relative to the roadway is greatly limited during installation on or in the roadway due to the nature of the roadway. On the other hand, if the magnetometer is mounted above the roadway, there may be larger deviations in the vertical tilting. In both cases, these constant deviations can be eliminated by a calibration provided according to a preferred embodiment.

According to a preferred embodiment, the method therefore includes the step of automatically compensating for an influence of a yaw angle α between a target direction of travel and an actual direction of travel of a vehicle, whose influence on the magnetic field detected by the line sensors is detected. One possibility for compensating for the yaw angle α is described further below.

The roadway preferably has a guide marking, for example a roadway marking or lateral guide structures, which are arranged in such a way that the vehicles drive essentially at right angles to the line sensors. The feature that the vehicles do not drive essentially perpendicularly to the line sensors is understood in particular to mean that it is possible, but not necessary, that the vehicles drive perpendicularly to the line sensors in a strictly mathematical sense. In particular, deviations of, for example, plus ±5° are tolerable.

According to a preferred embodiment, the method comprises the step of acquiring magnetometer data from at least three magnetometers which are arranged along a third straight line and form a third magnetic line sensor, the third straight line running along the first straight line. In other words, three, four, five or more line sensors can be placed side by side along the roadway. Two line sensors are sufficient to carry out the identification of vehicles according to the invention.

The method preferably comprises the following steps: (i) detecting a magnetic rigid part signature of non-rotating components of the vehicle from the magnetic field data set and (ii) identifying a vehicle type, in particular only using the magnetic rigid part signature. It has been found that identifying a vehicle using the magnetic signature is possible with particularly high accuracy if only the magnetic rigid part signature is used.

According to the invention is also a method for identifying vehicles, with the steps (a) acquiring magnetic field data from at least six magnetometers, so that a magnetic field data set is obtained which describes at least one magnetic field caused by the vehicle and is spatially resolved in at least two dimensions, and (b) identifying a vehicle type and / or a vehicle based on the magnetic field data set, wherein the identification of the vehicle type and / or the vehicle comprises the following steps: (i) detecting a magnetic rigid part signature of non-rotating components of the vehicle from the magnetic field data set and (ii) identify of a vehicle type, in particular only on the basis of the magnetic rigid part signature. It is then advantageous, but not necessary, for at least three magnetometers to be arranged in each case along a straight line, as stated in claim 1 .

According to the invention is also a vehicle identification device for identifying vehicles, with at least 6 magnetometers for the respective detection of Magnetic field data, the magnetic field data describing a change in a static magnetic field over time, and an evaluation circuit that is designed to automatically carry out a method according to the invention. It is advantageous, but not necessary, for 3 magnetometers to be arranged along a straight line and each to form a line sensor.

Rigid parts are mostly the vehicle components that do not rotate when the vehicle is moving. In particular, the wheels are not rigid parts. The reason for this is that the magnetic signature of the wheels depends on the angle of rotation of the respective wheel. However, this one is random. If the magnetic signature also has parts that depend on the magnetization of the wheels, even a different angular position of the wheels can result in vehicles being identified incorrectly or not at all.

Preferably, the detection of the magnetic rigid part signature includes the step of determining those areas in the magnetic field data in which the magnetic field measured by the at least two line sensors can be described within a predetermined error tolerance exclusively by a translation of a fixed magnetic field value with a vehicle speed, so that rigid part areas be obtained.

When the vehicle crosses the first line sensor, its magnetic signature is scanned. When the vehicle crosses the second line sensor, the rigid parts of the vehicle have moved on by a distance that corresponds to the distance between the two line sensors to a good approximation. A point on the tread of a wheel, on the other hand, has traveled a distance that can deviate from this distance both positively and negatively. The magnetic signatures scanned by the two line sensors therefore only match in the rigid part areas - within the scope of measurement accuracy.

Alternatively or additionally, the detection of the magnetic rigid part signature comprises the steps (i) determining at least two areas in which the magnetic field measured by the at least two line sensors cannot be described within the specified error tolerance by a translation of a fixed magnetic field value with a vehicle speed, so that receive at least four wheel areas and (ii) thereafter computing the rigid portion regions by masking the wheel regions from the spatially resolved magnetization. In other words, it is possible to also determine those areas that are not rigid part areas and then to eliminate these areas from the overall result.

According to a preferred embodiment, the rigid-part signature can be determined under the boundary conditions that the areas that are not rigid-part areas have an axis of symmetry.

Preferably, calculating the magnetic rigid part signature comprises the steps of (i) calculating a difference image from a first line sensor image describing the spatially resolved magnetization measured by the first line sensor and a second line sensor image describing the spatially resolved magnetization measured by the second line sensor , (ii) determining a binary image from the difference image using a threshold filter, (iii) determining at least one contiguous area in the binary image and (iv) setting the at least one contiguous area either as a rigid part area or as a wheel area. Whether the at least one contiguous area is set as a rigid part area or as a wheel area depends on the selection of the threshold value filter. The areas, in particular pixels, in which the first line sensor image and the second line sensor image hardly differ form the rigid part area.

The threshold filter is a filter that assigns either a first or a second value, typically 0 or 1, to a pixel of the difference image, depending on how much the pixel deviates from a central pixel. The middle pixel can, for example, be the median or the average or another mean value from the other pixels. Such a threshold filter is well known from image analysis. For example, applying the Threshold filter causes a flood fill.

So that the vehicle can be identified with little uncertainty when the vehicle is traveling at different speeds, identifying the vehicle using the magnetic field data preferably comprises the following steps: (i) scaling a time component of the magnetometer data, with the time component being selected in such a way that the at least one magnetometer with corresponds to a specified target driving speed, and/or normalization of a signal strength of the magnetometer data, so that a scaled magnetic field data set is obtained, (ii) reading out reference magnetic field data sets from a database that contains a large number of reference magnetic field data sets from different vehicles, (iii) for the reference magnetic field datasets each determine a similarity parameter that encodes a similarity of the scaled measurement field data of the magnetic field dataset with the reference magnetic field data of the reference magnetic field dataset, and (iv) identify a vehicle based on the similarity parameter. The reference magnetic field data records can also be referred to as reference signatures, the magnetic field data records as magnetic signatures.

The similarity parameter is a number that describes how similar the vehicle's magnetic signature is to the stored signature. If the similarity parameter is within a predetermined interval, the vehicle is identified as the vehicle with which the reference magnetic field data is linked.

The scaled magnetic field dataset is a magnetic signature. If—as provided according to a preferred embodiment—the rigid part signature is calculated using the scaled magnetic field data set, the rigid part signature is the magnetic signature. The similarity of the stored reference rigid part signatures to this signature encoded with the similarity parameter is determined.

The method according to the invention and the vehicle identification device according to the invention can be used particularly advantageously when detecting whether a vehicle is authorized to pass a section of the roadway. For example, the roadway has a section that may not be driven on by vehicles above a maximum permissible total weight. It is then advantageous to detect those vehicles whose mass is above the maximum permissible total mass. Alternatively or additionally, the roadway can lead to a building, for example a multi-storey car park, to which only authorized vehicles should have access. In this case, it is desirable that only authorized vehicles are allowed through based on the magnetic signature.

According to a preferred embodiment, the method therefore comprises the steps of (i) detecting an authorization parameter that is linked to the reference magnetic field data record that identifies the vehicle and (ii) emitting a signal to an access release device so that it clears a route for the vehicle (a ) authorizes if the authorization parameter encodes authorization to do so, and (b) does not authorize if the authorization parameter does not encode authorization to do so. The signal is, for example, a control signal that controls an access release device. The access release device is a device for obstructing or preventing further travel, for example a barrier or a gate.

The magnetic field data preferably describe the magnetic field in at least two dimensions, in particular in three dimensions. The feature that the magnetic field data describe the magnetic field in at least two dimensions means in particular that at least two independent spatial components of the magnetic field, which is a vector field, are measured. The room components are each measured as a function of time. The third spatial component is the normal component, i.e. the component that is perpendicular to a compensation plane through the roadway in the area of the line sensors. In addition, the third spatial component is preferably the vertical component.

To measure the vertical component, the vehicle identification device preferably has a third line sensor, which is arranged at a different height than at least one of the other line sensors.

The method preferably comprises the steps of (i) reading a maximum permissible total mass of the vehicle identified by its magnetic signature from a database using the similarity parameter or the vehicle type, and (ii) sending a signal to the access release device that depends on the total mass and/or Sending a signal encoding total mass to a monitoring device. If, for example, reading out the maximum permissible total mass of the identified vehicle shows that it is above a specified maximum total mass, a signal is sent to the access release device, which causes the vehicle to be blocked from continuing its journey. With the method according to the invention, for example, a Bridge must be protected from being driven over by vehicles that are too heavy. The signal can also be an optical, acoustic and/or electronic warning signal. For example, the signal signals to the driver of the vehicle that further travel is prohibited or that further travel is possible with restrictions. For example, the signal can indicate a minimum distance and/or a maximum speed.

The more a vehicle is loaded, the lower it is. In other words, the greater the mass of the vehicle, the smaller the distance between the rigid parts and the roadway. The smaller the distance between the rigid parts and the road, the greater the magnetic field measured by the line sensors. The method therefore preferably comprises the steps of (a) determining a signal intensity parameter which encodes a strength of the magnetic field from the magnetic field data and (b) determining the total mass from the signal intensity parameter and the scaled magnetic field data set.

A sampling frequency of at least one, in particular a plurality, preferably all, magnetometers is preferably at least 50 Hertz.

For the majority of magnetometers, in particular all magnetometers, of a line sensor, the distance between adjacent magnetometers is preferably no more than 30 centimeters, in particular no more than 20 cm, preferably no more than 10 cm.

A vehicle identification device according to the invention preferably has a clock for determining the time. This can be the absolute time or a machine time, for example measured as a time increment from a specified point in time.

In order to obtain as high an identification probability as possible, it is advantageous to standardize the magnetic field data to a predefined target driving speed. The escalated magnetic field data obtained in this way are then independent of the speed with which the vehicle drove over the line sensors. The evaluation circuit is therefore preferably designed to carry out a method with the steps (i) determining an actual vehicle speed the magnetic field data from two of the magnetometers arranged along the straight line, and (ii) scaling the time component of the magnetic field data of the magnetic field data set, the time component being selected in such a way that it corresponds to driving over the at least one magnetometer with the specified target driving speed, based on the actual vehicle speed.

It is favorable if at least the three magnetometers arranged along a straight line are arranged at the same height. The feature that the magnetometers are arranged at the same level is understood in particular to mean that it is possible, but not necessary, for the magnetometers to be arranged at the same level, strictly mathematically. In particular, deviations of at most 30 cm, or at most 20 cm, particularly preferably at most 10 cm, from the ideal arrangement at the same height are possible. The height is determined as the distance to the compensation level through the roadway in the area of the corresponding line sensor.

Preferably, at least three magnetometers are arranged at a second level, which differs from the first level by a level difference. The height difference is preferably at most 50 cm, in particular at most 30 cm. The height difference is preferably greater than 4 cm, in particular greater than 10 cm.

A structure according to the invention preferably has an access release device for obstructing or releasing the roadway. The vehicle identification device is preferably designed to automatically control the access release device as a function of a drive-through authorization that depends on the identification of the vehicle.

Figur 1

ein einzelnes Zeilensensor-Modul auf einer gemeinsamen Platine 11 mit acht Magnetometern,

Figur 2a

die nicht maßstabsgetreue Anordnung von zwei Zeilensensoren in einer Fahrbahn,

Figur 2b

die Anordnung (nicht maßstabsgetreu) von zwei Zeilensensoren auf einer Fahrbahn,

Figur 3a

ein Erstzeilensensorbild, das das von einem Fahrzeug hervorgerufene Magnetfeld zeigt, das vom ersten Zeilensensor gemessen wurde, wobei der Gierwinkel null ist, und

Figur 3b

das Erstzeilensensorbild bei einem von null verschiedenen Gierwinkel.

The invention is explained in more detail below with reference to the accompanying drawings. while showing

figure 1

a single line sensor module on a common circuit board 11 with eight magnetometers,

Figure 2a

the arrangement of two line sensors in a lane, which is not true to scale,

Figure 2b

the arrangement (not true to scale) of two line sensors on a roadway,

Figure 3a

a first line sensor image showing the magnetic field evoked by a vehicle measured by the first line sensor with yaw angle zero, and

Figure 3b

the first line sensor image at a non-zero yaw angle.

figure 1 shows a vehicle identification device 1 with a line sensor 10, which in the present case has eight magnetometers 12.j (j=1, . . . There may be more or fewer magnetometers 12. The magnetometers 12.j are arranged along a first straight line G1. A distance d ₁₂ between two adjacent magnetometers is d=5 cm.

The vehicle identification device 1 has an evaluation circuit 13 in the form of a microcontroller and associated logic modules, which in the present case is also arranged on the circuit board, but this is optional.

Line sensor 10 has at least one plug-in connection 14 for cascading the sensors, communicating with at least one second line sensor, outputting data and for power supply.

Figure 2a shows the arrangement of at least two line sensors 20.i (i=1, 2) on a roadway 21 in plan view, which is not true to scale. A vehicle 22, im present case, a passenger car, drives at a speed v over the line sensors 20.1 and 20.2.

Figure 2a shows a coordinate system k ₁ of the line sensors 20.i, a coordinate system related to the vehicle 22 k ₂ and a yaw angle a. This yaw angle α exists between the velocity vector v and the straight line G1 and thus between the coordinate systems k ₁ and k ₂ .

Figure 2b shows the arrangement, not true to scale, of two line sensors 20.i at a distance d from one another on the roadway 21 in a side view. The vehicle 22 runs over the line sensors (20.1, 20.2) at a height h .

Each magnetometer 12.j supplies measurement data, which are called magnetometer data and are time-resolved for up to three components (x, y, z components) of a magnetic field B ( x,y,z ) describe.

The entirety of the magnetometer data form a magnetic field data set. For this is the magnetic field B ( x,y,z ) under the vehicle 22 can be calculated.

mit j ∈ {x, y, z} und ${\mathit{m}}^{\left(i,j\right)}\in {ℝ}^{N\times t_{\mathit{end}}-t_0}$

dargestellt werden. Darin ist t_end der Endzeitpunkt der Vermessung.The recording process is triggered at a point in time t ₀ . For example, the recording process starts when a threshold value for the measured magnetic field of a magnetometer is exceeded or due to an external start signal, for example due to the triggering of another transmitter such as a light barrier, which can be part of the vehicle identification device. The magnetometer data of the line sensors 20.i can as a measurement data matrix, for example the form $${\mathit{m}}^{\left(i,j\right)}\left[t_0...t_{\mathit{end}},1...N\right]=\left(\begin{array}{ccc}{B}_{j,1,t_0}^{\left(i\right)}& \cdots & B_{j,1,t_{\mathit{end}}}^{\left(i\right)}\\ ⋮& \ddots & ⋮\\ B_{j,N,t_0}^{\left(i\right)}& \cdots & B_{j,N,t_{\mathit{end}}}^{\left(i\right)}\end{array}\right)$$

with j ∈ { x, y, z } and ${\mathit{m}}^{\left(i,j\right)}\in {ℝ}^{N\times t_{\mathit{end}}-t_0}$

being represented. In this, t _end is the end time of the measurement.

The measurement is completed, for example, when the measured magnetic field falls below a predetermined threshold value for all magnetometers for a predetermined time, or after a certain time has elapsed.

A maximum measurement duration can be estimated for a given application based on the maximum expected vehicle length l _max and a minimum speed v _min as t _end =t ₀ + l _max / v _min . For the application of a random check, this would be, for example, walking speed with v _min = 4 km/h. The matrix of the magnetic field magnitude is accordingly: $$\begin{array}{l}{\mathit{m}}^{\left(i\right)}\left[t_0...t_{\mathit{end}},1...N\right]\\ =\left(\begin{array}{ccc}\sqrt{{B_{x,1,t_0}^{\left(i\right)}}^2+{B_{y,1,t_0}^{\left(i\right)}}^2+{B_{\mathrm{e.g},1,t_0}^{\left(i\right)}}^2}& \cdots & \sqrt{{B_{x,1,t_{\mathit{end}}}^{\left(i\right)}}^2+{B_{y,1,t_{\mathit{end}}}^{\left(i\right)}}^2+{B_{\mathrm{e.g},1,t_{\mathit{end}}}^{\left(i\right)}}^2}\\ ⋮& \ddots & ⋮\\ \sqrt{{B_{x,N,t_0}^{\left(i\right)}}^2+{B_{y,N,t_0}^{\left(i\right)}}^2+{B_{\mathrm{e.g},N,t_0}^{\left(i\right)}}^2}& \cdots & \sqrt{{B_{x,N,t_{\mathit{end}}}^{\left(i\right)}}^2+{B_{y,N,t_{\mathit{end}}}^{\left(i\right)}}^2+{B_{\mathrm{e.g},N,t_{\mathit{end}}}^{\left(i\right)}}^2}\end{array}\right)\end{array}$$

zwischen den bis zu dreikomponentigen (x-, y-, z-Komponente) Magnetfeldmesswerten der beiden Zeilensensoren m ^(1,j) und m ^(2,j). Der zusätzliche Summand Ψ berücksichtigt dabei technisch bedingte Verschiebungen, insbesondere durch einen Versatz im Abtastzeitpunkt der Zeilensensoren und muss zum Messzeitpunkt bekannt sein.Since the at least two line sensors are spaced apart from one another, there is a time shift $\mathit{\Delta t}=\frac{\mathrm{i.e}}{v}+\Psi$

between the up to three-component (x, y, z component) magnetic field measurement values of the two line sensors m ^{(1, j )} and m ^{(2, j )} . The additional summand Ψ takes into account technically caused shifts, in particular due to an offset in the scanning time of the line sensors, and must be known at the time of measurement.

These magnetometers 12 can acquire magnetometer data in either only one, two or three spatial dimensions at a time t. For the sake of simplicity, the general case for three-component measurements is described here. Δt is proportional to the distance between the sensors d and the speed v of the vehicle 22.

In addition to a time shift, a spatial shift Δn of the signature can occur, for example, when the vehicle 22 drives over the line sensors at a sufficiently oblique angle. With a given time base, the temporal and spatial shift can be determined, for example, directly by comparing the time-resolved measurement series m ⁽¹⁾ and m ^(2), for example using a method based on the calculation of the 2D correlation according to Pearson become. The direction in which vehicle 22 crosses line sensors 20.i is irrelevant. If a correlation method according to Pearson is used as an example, this results in a sign change of the time shift Δt .

The speed of the vehicle 22 may be time dependent. In order to take this into account, a preferred embodiment provides for the measurement series m ⁽¹⁾ and m ⁽²⁾ to be divided into subintervals for the correlation analysis and, for example, to be assigned a time-resolved shift Δ t ( t ) using a sliding window-based approach receive.

, wobei d der Abstand zwischen zwei Zeilensensoren ist und Δl der Abstand von zwei benachbarten Magnetometern in einem Zeilensensor ist. Aus der vektoriellen Geschwindigkeit lässt sich der Gierwinkel α bestimmen.The spatial displacement Δn is assumed to be constant over time, since it is assumed that the vehicle crosses the line sensors 20.j obliquely, but linearly. This creates a vectorial velocity profile v ( t ) determined. Since the distance x traveled and the addend Ψ are known or can be determined within the framework of a calibration, the speed of the vehicle v ( t )= x /( Δt ( t ) −Ψ ) can be determined directly. The path results from x = $\sqrt{{\mathrm{i.e}}^2+{\left(\mathit{\Delta n}\ast \mathit{\Delta l}\right)}^2}$

, where d is the distance between two line sensors and Δ l is the distance between two adjacent magnetometers in a line sensor. The yaw angle α can be determined from the vectorial velocity.

Since the magnetic field measurement m ^{(1, j )} is both temporally and spatially resolved, typical image processing methods can be used to identify characteristic components, such as wheels, of a vehicle (see Fig Figure 3a ). This can significantly reduce the rate of false detections caused by disturbances (animals, wheelbarrows, bicycles, etc.).

The vehicle 22 has wheels 23.k. The magnetic field caused by the wheels 23.k on the respective magnetometer varies depending on the angle of rotation of the respective wheel. Therefore, the magnetic signature of the wheels is typically different between spatially spaced line sensors. The magnetic signature of the rigid components of the vehicle, on the other hand, is identical between the line sensors. In other words, such components of the vehicle are regarded as rigid components for which it applies that their magnetic signature is identical between the line sensors. Movable components can therefore be clearly identified in the difference in the magnetic measurement data of spaced line sensors.

When calculating the difference, both the time shift Δ t ( t ) and the spatial shift Δ n must be taken into account, which was calculated in a previous step, e.g. B. can be determined with a 2D cross-correlation.

wobei dm ^(i-g) die Matrix des Magnetfeldbetrag von dm ^(i-g,j) ist.If Δ t ( t ) and Δ n are known, the difference to another line sensor 20.g, which is referred to as difference image D, results in: $$\mathit{i.e}{\mathit{m}}^{\left(i-G,j\right)}={\mathit{m}}^{\left(i,j\right)}\left[t_0...t_{\mathit{end}}-\mathit{\Delta t}\left(t\right),1...\mathit{\Delta n}\right]-{\mathit{m}}^{\left(G,j\right)}\left[t_0+\mathit{\Delta t}\left(t\right)...t_{\mathit{end}},\mathit{\Delta n}...N\right],$$

where dm ^{( i - g )} is the matrix of the magnetic field magnitude of dm ^{( i - g,j )} .

Depending on the direction of movement of the vehicle, the signs of Δ t ( t ) and Δ n can change. If the sign of Δt ( t ) or Δn changes, the corresponding arguments of the line sensors 20.i and 20.g must be exchanged in the above formula.

überführt, wobei τ ein gegebener Schwellwert ist. Die Wahl des Schwellwerts τ ist abhängig von den gemessenen Magnetfelddaten und wird für jede Messung individuell bestimmt. Beispielsweise kann dies über folgende Vorschrift erfolgen $$\mathrm{\tau }=\mathrm{MED}+3\ast 1.4826\ast \mathrm{MAD},$$

welche typischerweise für Schwellwertdetektion von Differenzbildern benutzt wird. MED entspricht dem Median aller Messwerte $$\mathrm{MED}={\mathit{median}}_{a,b}\left({\mathit{dm}}_{a,b}^{\left(i-g\right)}\right)$$

und MAD entspricht dem Median der absoluten Abweichung der Messwerte von MED $$\mathrm{MAD}={\mathit{median}}_{a,b}\left(\left|{\mathit{dm}}_{a,b}^{\left(i-g\right)}-\mathrm{MED}\right|\right).$$

For the subsequent detection of the wheels 23.k, the difference image is first converted into a binary matrix B according to a threshold value detection $$B_{a,b}=\{\begin{array}{c}1,\mathrm{for}{\mathit{dm}}_{a,b}^{\left(i-G\right)}>\tau \\ 0,\mathrm{otherwise}\end{array}$$

converted, where τ is a given threshold. The selection of the threshold value τ depends on the measured magnetic field data and is determined individually for each measurement. For example, this can be done using the following rule $$\mathrm{\tau }=\mathrm{MED}+3\ast 1.4826\ast \mathrm{MAD},$$

which is typically used for threshold detection of difference images. MED corresponds to the median of all measured values $$\mathrm{MED}={\mathit{median}}_{a,b}\left({\mathit{dm}}_{a,b}^{\left(i-G\right)}\right)$$

and MAD corresponds to the median of the absolute deviation of the measured values from MED $$\mathrm{MAD}={\mathit{median}}_{a,b}\left(\left|{\mathit{dm}}_{a,b}^{\left(i-G\right)}-\mathrm{MED}\right|\right).$$

mit dem Mittelpunkt der Ellipse (c_t , c_n ) und den Halbachsen a und b, gefittet. Das Fitten kann z. B. mit dem Levenberg-Marquardt-Algorithmus durchgeführt werden. Um die Signaturen der Räder von anderen Störeinflüssen zu unterscheiden, werden Bedingungen an die Form und Position der Ellipsen gestellt. Zum einen wird überprüft, ob die Halbachsen der Ellipsen sich in Bereichen typisch für Räder befinden. Weiterhin können anhand der Anzahl der identifizierten Räder Rückschlüsse auf den Fahrzeugtyp gezogen werden (Pkw, Lkw, Motorrad, dreirädrige Kraftfahrzeuge). Dabei müssen bei Pkw und Lkw die Mittelpunkte der Räder ungefähr auf einem Trapez liegen und bei dreirädrigen Kraftfahrzeugen zumeist auf einem Dreieck, bei dem alle Innenwinkel < 90° sind.In the binary matrix B, all connected areas are identified, for example, with a flood filling known from computer graphics. Attached to these faces are ellipses of shape $$\frac{{\left(n-c_n\right)}^2}{a^2}+\frac{{\left(t-c_t\right)}^2}{b^2}=1,$$

with the center of the ellipse ( c _t , c _n ) and the semi-axes a and b, fitted. The fitting can e.g. B. be performed with the Levenberg-Marquardt algorithm. In order to distinguish the signatures of the wheels from other disturbing influences, conditions are placed on the shape and position of the ellipses. On the one hand, it is checked whether the semi-axes of the ellipses are in areas typical for wheels. Furthermore, based on the number of wheels identified, conclusions can be drawn about the vehicle type (car, truck, motorcycle, three-wheeled motor vehicle). In the case of passenger cars and trucks, the centers of the wheels must lie approximately on a trapezoid and, in the case of three-wheeled motor vehicles, usually on a triangle in which all interior angles are < 90°.

The yaw angle α can be determined with high accuracy on the basis of the vehicle transverse axes or the vehicle longitudinal axes, which run through the center points of the identified wheels. Is e _q the direction of an identified vehicle transverse axis and e _s is the direction of the longitudinal axis of the line sensor, the yaw angle α is given by α = cos ^-1 ( e _q · e _s ), where e _q · e _s is the scalar product. The process works in the same way for the longitudinal axis of the vehicle and the transverse axis of the sensor. Independent calculation of the yaw angle α for all identified axes and subsequent averaging significantly reduces the error due to measurement uncertainties.

Since the wheel areas cannot be used for identification, they are masked in the measurement data. The masking can be realized, for example, by setting the values of the matrices for the affected areas to a constant value c: $$m_{p,q}^{\left(i,j\right)\prime }=\{\begin{array}{c}c,\mathrm{for}{B}_{q,p}=1\\ m_{p,q}^{\left(i,j\right)},\mathrm{otherwise}\end{array}.$$

In order to obtain a comparable magnetic signature from these series of measurements, regardless of the speed of the vehicle, the temporal stretching or compression caused by the speed of the vehicle is taken into account.

Areas of the magnetic field created by moving components are preferably masked. This allows a comparable magnetic signature of the vehicle to be determined. Since the spatial resolution along the car depends on the speed and is therefore variable between measurements, the measurement data matrix can be converted to a predefined resolution.

für die bis zu drei Komponenten j ∈ {x, y, z} des Magnetfelds aus den zeitabhängigen Magnetfeldmesswerten m ^(i,j)' des i-ten Zeilensensors mithilfe von Interpolationsalgorithmen bestimmt, wobei n = 1 ... N die Anzahl der Magnetometer ist und w = 1 ... W die vorgegebene Anzahl der räumlichen Messpunkte in Fahrtrichtung des Fahrzeugs. Die Fingerabdruck-Matrix wird als Binärbild bezeichnet. Das Binärbild erlaubt die Erkennung der Starrteilkomponenten.For this purpose, a magnetic fingerprint matrix is created depending on the speed of the vehicle v ( t ). ${\mathit{f}}^{\left(\mathit{i},\mathit{j}\right)}\in {ℝ}^{W\times N}$

for up to three components j ∈ { x, y, z } of the magnetic field from the time-dependent magnetic field measurement values m ^{( i,j )'} of the i-th line sensor using interpolation algorithms, where n = 1 ... N the number of magnetometers and w = 1 ... W is the specified number of spatial measuring points in the direction of travel of the vehicle. The fingerprint matrix is called a binary image. The binary image allows the identification of the rigid part components.

W results from the typical spatial resolution Δ x = v / ƒ at a vehicle speed that is typical for the application v and a typical sampling rate of the magnetometers ƒ to W = l _max /Δ x , where l _max is an upper limit for the length of the vehicles to be detected. In one possible embodiment, the fingerprint matrix F ^{( i,j )} can be combined to form a common magnetic fingerprint matrix F ^{( j )} after suitable preprocessing, for example by averaging across all line sensors i.

überführt.When vectorial magnetometers are used, the mean value of the absolute amounts of the elements of the magnetic fingerprint matrix F that belong together in each case can preferably be determined. The averaging also reduces the measurement inaccuracy. The binary matrix B also becomes analogous by interpolation into a matrix $\tilde{\mathit{B}}\in {ℝ}^{W\times N}$

transferred.

For a subsequent comparison with reference signatures in a database, it is advantageous if the magnetic signature F at a corresponding reference F _ref is aligned eg by a 2D cross-correlation. In particular, the signatures of movable components are already masked in the reference signatures F _ref . This makes it easier to compare the recorded magnetic fingerprints, even if different variants of the line sensors are used.

Similarly, the matrix B̃ is aligned with F _ref . All areas marked with 1 in the binary matrix B̃ are omitted for vehicle identification both in the measured magnetic fingerprint F and in the references F _ref .

Since these segments of the magnetic fingerprint have changed when at least two spaced line sensors are passed over, they cannot be used for identification. This is particularly the case for moving components of the vehicle, but also for interference. The latter can be magnetic objects, for example, which move relative to the vehicle while crossing the line sensors and thus falsify the magnetic fingerprint. The procedure for masking is analogous to that for wheel masking.

in einer Datenbank, vorzugsweise durch eine Korrelationsanalyse verglichen. Ein Vergleich gilt dann als positiv, wenn beispielsweise der Pearson Korrelationskoeffizient einen Schwellwert von r_T = 0.9 überschreitet. Ist dies der Fall für mehrere Referenzen ${\mathit{F}}_{\mathit{ref}}^{\prime }$

, so wird die Referenz mit dem größten Pearson Korrelationskoeffizienten ausgewählt.This approach makes the system robust against interference or attempts at deception and provides the necessary accuracy, e.g. B. to implement access controls. The matrix F̃ ' is then filled with references ${\mathit{f}}_{\mathit{ref}}^{\prime }$

in a database, preferably compared by a correlation analysis. A comparison is considered positive if, for example, the Pearson correlation coefficient exceeds a threshold value of r _T = 0.9. Is this the case for multiple references ${\mathit{f}}_{\mathit{ref}}^{\prime }$

, the reference with the largest Pearson correlation coefficient is selected.

, wobei h der vertikale Abstand zwischen den Zeilensensoren und dem Fahrzeug ist. Anhand der detektierten Achsen bzw. Räder kann mit Hilfe einer Datenbank auf den Fahrzeugtyp geschlossen werden. Abhängig davon lässt sich aus einer Datenbank auch die zulässige Gesamtmasse bestimmen und damit am Messort das zulässige Gesamtgewicht. Aus der Höhe kann dann auf das tatsächliche Gesamtgewicht des Fahrzeugs geschlossen werden, da die Federkonstante des Fahrzeugs bekannt ist und vorzugsweise in der Datenbank ebenfalls gespeichert ist. Dazu kann, selbst wenn für das spezifische Fahrzeug keine bekannte Signatur aus einer Datenbank oder Ähnlichem vorhanden ist, zum Vergleich eine mittlere Feldstärke für den jeweiligen Fahrzeugtyp verwendet werden. Ist für das Fahrzeug bereits ein Fingerabdruck bekannt, kann durch einen Vergleich der Amplituden eine deutlich genauere Bestimmung des Gesamtgewichts erfolgen. Insbesondere ist es dadurch möglich eine Überladung von typischen Fahrzeugtypen zu detektieren um insbesondere eine Verkehrsüberwachung zu realisieren.The amplitude of the captured magnetic fingerprint is also scaled in a dipole approximation $\frac{1}{H^3}$

, where h is the vertical distance between the line sensors and the vehicle. Based on the detected axles or wheels, the vehicle type can be deduced with the help of a database. Depending on this, the permissible total mass can also be determined from a database and thus the permissible total weight at the measuring location. The actual total weight of the vehicle can then be deduced from the height, since the Spring constant of the vehicle is known and is preferably also stored in the database. For this purpose, an average field strength for the respective vehicle type can be used for comparison, even if there is no known signature from a database or the like for the specific vehicle. If a fingerprint is already known for the vehicle, the total weight can be determined much more precisely by comparing the amplitudes. In particular, this makes it possible to detect overloading of typical vehicle types, in particular in order to implement traffic monitoring.

An advantage of the invention is that identification is also possible when the measuring unit or a line sensor has not been completely passed by the vehicle. This can be the case, for example, with parking spaces, ramps or driveways due to the nature of the vehicles or the structural conditions. The partial signature can then still be compared with an already known, fully captured fingerprint, taking into account the incomplete capture. A combination with other sensor systems, e.g. B. optical systems possible.

1. 1. Start der Aufnahme einer magnetischen Signatur durch einen Auslöser, z. B. das Überschreiten eines Schwellwerts im Magnetfeld.
2. 2. Vermessen einer magnetischen Signatur m ^(1,j) und m ^(2,j) durch mindestens zwei in der erwarteten Fahrtrichtung zueinander beabstandete Zeilensensoren. Die Vermessung ist abgeschlossen, wenn das gemessene Magnetfeld wieder auf einen zuvor bestimmten und ungestörten Grundzustand zurückfällt, oder nach Ablauf einer bestimmten Zeit.
3. 3. Bestimmen der räumlichen und zeitlichen Verschiebung der magnetischen Signaturen zwischen den Zeilensensoren z. B. mithilfe einer 2D-Kreuzkorrelation und anschließende Berechnung eines vektoriellen Geschwindigkeitsprofils v (t). Zusätzlich kann eine Konsistenzprüfung stattfinden, indem die Signaturen m ^(1,j)und m ^(2,j) auf ihre Ähnlichkeit, ausgedrückt z. B. durch den normalisierten Pearson Korrelationskoeffizienten, überprüft werden. Liegt der Koeffizient unterhalb eines Schwellwertes (< 0.7), wird eine Fehlermeldung ausgegeben.
4. 4. Identifikation veränderlicher Bereiche im Differenzsignal dm ^(1-2) durch eine Schwellwertdetektion und Berechnung einer binären Matrix B.
5. 5. Identifikation der Fahrzeugräder in der binären Matrix B.
6. 6. Bestimmung des Gierwinkels α anhand der Fahrzeugquerachsen bzw. der Fahrzeuglängsachsen, die durch die Mittelpunkte der identifizierten Räder verlaufen (siehe Figuren 2 und 5).
7. 7. Korrektur der Magnetfeldmessungen und der binären Differenzmatrix B um den Gierwinkel a, beispielsweise durch Anwendung einer geeigneten Drehmatrix.
8. 8. Umrechnung der Magnetfeldmessungen Signatur m ^(1,j) und m ^(2,j) in magnetische Fingerabdrücke F ^(1,j) und F ^(2,j) mithilfe des Geschwindigkeitsprofils v (t) unund Interpolationsalgorithmen. Anschließende Mittelwertbildung F = ( F ⁽¹⁾ + F ⁽²⁾)/2. Überführung der binären Matrix B in B̃ durch Interpolation.
9. 9. Übermittlung des finalen magnetischen Fingerabdrucks F und der binären Matrix B̃ an einen Identifikationsserver. Falls keine Kommunikationsverbindung besteht, wird der Fingerabdruck, inklusiver noch benötigter Vorprozessierung, mit einer internen Datenbank abgeglichen.
10. 10. Maskieren ungeeigneter Segmente, markiert in der binären Differenzmatrix B̃ .
11. 11. Vergleich der Matrizen F ' und ${\mathit{F}}_{\mathit{ref}}^{\prime }$
    
    . Der Abgleich kann z. B. durch eine 2D Korrelation erfolgen. Ein Fingerabdruck gilt dann als positiv zugeordnet, wenn der normalisierte Pearson Korrelationskoeffizient über einen vordefinierten Schwellwert liegt.
12. 12. Übermittlung der Ergebnisse an entsprechende Aktoren, z. B. Schranken oder Ampeln.

Bezugszeichenliste 1 Fahrzeug-Identifikationsvorrichtung Δl Abstand zwischen zwei Magnetometern in einem Zeilensensor 10 Zeilensensor-Modul 11 Platine x Zurückgelegter Weg bei der Überfahrt zweier Zeilensen-soren 12 Magnetometer 13 Auswertschaltung 14 Steckverbindung v Geschwindigkeit des Fahrzeugs 20 Zeilensensoren d Abstand zwischen den Zeilensensoren 21 Fahrbahn 22 Fahrzeug ψ technisch bedingte zeitliche Verschiebungen zwischen den Zeilensensoren i Laufindex der Zeilendetekto- ren t Zeit j Laufindex der Magnetometer t ₀ Startzeitpunkt der Aufnahme einer Magnetfeldmessung k Laufindex der Räder t_end Endzeitpunkt der Aufnahme einer Magnetfeldmessung D Differenzbild m ^(ij)[t ₀ ... t_end, 1 ... N] zeitlich aufgelöste Magnetfeldmesswerte der j-ten Komponente des i-ten Zeilensensors k ₁ Zeilensensor bezogenes Koordinatensystem m ⁽ⁱ⁾[t₀ ... t_end, 1 ... N] zeitlich aufgelöste Magnitude der Magnetfeldmesswerte des i-ten Zeilen-sensors k ₂ Fahrzeug bezogenes Koordinatensystem α Gierwinkel zwischen Fahrzeug und Zeilensensor dm ^(i-g,j) Matrix der Magnetfelddiffe-renz zwischen dem i-ten und g-ten Zeilensensor der j-ten Komponente Δt zeitliche Verschiebung zwischen erfassten magnetischen Signaturen Δn räumliche Verschiebung zwischen erfassten magnetischen Signaturen dm ^(i-g) Matrix der Magnetfeldmagnitude von dm ^(i-g,j) B binäre Differenzmatrix q, p Matrixindizes druck-Matrix der j-ten Komponente i Schwellwert zur Bestimmung der Differenzmatrix B F Mittelwert über alle Zeilensensoren der betrags-wertigen magnetischen Fingerabdruck-Matrix a große Halbachse einer Ellipse b große Halbachse einer Ellipse F _ref Referenz Fingerabdruck-Matrix aus einer Datenbank c_n , c_t Koordinaten des Mittelpunkts einer Ellipse B̃ Binäre Differenzmatrix interpoliert auf eine gemeinsame Auflösung e _q Einheitsvektor in Richtung einer Fahrzeugquerachse e _s Einheitsvektor in Richtung der Längsachse eines Zeilensensors ${\mathit{F}}_{\mathit{ref}}^{\prime }$

F' maskierte FingerabdruckMatrix maskierte Referenz Fingerabdruck-Matrix N Gesamtanzahl der Magnetometer eines Zeilensensors c Konstante n Index eines Magnetometers r_T Schwellwert des PearsonKorrelationskoeffizienten w Gesamtanzahl der räumlichen Messpunkte Δx Räumliche Auflösung w Index der räumlichen Messpunkte ƒ Abtastrate der Magnetometer F ^{( i,j )} Fingerabdruck-Matrix der jten Komponente des i-ten Zeilensensors l_max obere Grenze für die Länge der zu erfassenden Fahrzeuge F ^{( j )} Mittelwert über alle Zeilensensoren der Fingerab- l Länge des Fahrzeugs h vertikaler Abstand zwischen Zeilensensor und Fahrzeug The invention can be used particularly advantageously to identify vehicles during admission controls. For this purpose, the following procedural steps are carried out as an example:

1. 1. Start recording a magnetic signature by a trigger, e.g. B. exceeding a threshold value in the magnetic field.
2. 2. Measurement of a magnetic signature m ^{(1, j )} and m ^{(2, j )} by at least two line sensors spaced apart from one another in the expected direction of travel. The measurement is complete when the measured magnetic field falls back to a previously determined and undisturbed basic state, or after a certain time has elapsed.
3. 3. Determination of the spatial and temporal displacement of the magnetic signatures between the line sensors z. B. using a 2D cross-correlation and subsequent calculation of a vectorial velocity profile v ( t ). In addition, a consistency check can take place by checking the similarity of the signatures m ^{(1, j )} and m ^{(2 ,j )} , expressed e.g. e.g. by the normalized Pearson correlation coefficient. If the coefficient is below a threshold value (< 0.7), an error message is issued.
4. 4. Identification of variable areas in the difference signal dm ^(1-2) by threshold detection and calculation of a binary matrix B.
5. 5. Identification of the vehicle wheels in the binary matrix B.
6. 6. Determination of the yaw angle α based on the vehicle transverse axes or the vehicle longitudinal axes, which run through the centers of the identified wheels (see figures 2 and 5).
7. 7. Correction of the magnetic field measurements and the binary difference matrix B for the yaw angle a, for example by using a suitable rotation matrix.
8. 8. Conversion of the magnetic field measurements signature m ^{(1, j )} and m ^{(2, j )} into magnetic fingerprints F ^{(1, j )} and F ^{(2, j )} using the velocity profile v ( t ) unand interpolation algorithms. Subsequent averaging F = ( F ⁽¹⁾ + F ⁽²⁾ )/2. Conversion of the binary matrix B into B̃ by interpolation.
9. 9. Transmission of the final magnetic fingerprint F and the binary matrix B̃ to an identification server. If there is no communication connection, the fingerprint, including the pre-processing that is still required, is compared with an internal database.
10. 10. Masking inappropriate segments marked in the binary difference matrix B̃ .
11. 11. Comparison of the matrices F ' and ${\mathit{f}}_{\mathit{ref}}^{\prime }$
    
    . The adjustment can e.g. B. done by a 2D correlation. A fingerprint is considered positive if the normalized Pearson correlation coefficient is above a predefined threshold.
12. 12. Transmission of the results to the relevant actors, e.g. B. barriers or traffic lights.

<b>Reference List</b> 1 Vehicle Identification Device Δl Distance between two magnetometers in a line sensor 10 Line sensor module 11 circuit board x Distance covered when crossing two line sensors 12 magnetometers 13 evaluation circuit 14 connector v speed of the vehicle 20 line sensors i.e Distance between the line sensors 21 roadway 22 vehicle ψ technically caused time shifts between the line sensors i running index of the line detector ren t Time j Running index of the magnetometer t ₀ Start time of recording a magnetic field measurement k running index of the wheels t _end End time of recording a magnetic field measurement D difference image m ^{( ij )} [ t ₀ ... t _end , 1 ... N] time-resolved magnetic field measurement values of the j-th component of the i-th line sensor k ₁ line sensor related coordinate system m ^{( i )} [ t ₀ ... t _end , 1 ... N] Time-resolved magnitude of the magnetic field measurement values of the i-th line sensor k ₂ Vehicle related coordinate system a Yaw angle between vehicle and line sensor dm ^{( ig,j )} Matrix of the magnetic field difference between the i-th and g-th line sensor of the j-th component Δt temporal shift between captured magnetic signatures Δn spatial shift between captured magnetic signatures dm ^{( ig )} Matrix of magnetic field magnitude of dm ^{( ig,j )} B binary difference matrix q, p matrix indices pressure matrix of the jth component i Threshold for determining the difference matrix B f Mean value over all line sensors of the absolute value magnetic fingerprint matrix a semi-major axis of an ellipse b semi-major axis of an ellipse F _ref Reference fingerprint matrix from a database _cn , _ct Coordinates of the center of an ellipse B̃ Binary difference matrix interpolated to a common resolution e _q Unit vector in the direction of a vehicle transverse axis e _s Unit vector in the direction of the longitudinal axis of a line sensor ${\mathit{f}}_{\mathit{ref}}^{\prime }$

F ' masked fingerprint matrix masked reference fingerprint matrix N Total number of magnetometers of a line sensor c constant n Index of a magnetometer r _T Threshold of the Pearson correlation coefficient w Total number of spatial measurement points Δx Spatial resolution w Index of spatial measurement points ƒ Sampling rate of the magnetometers F ^{( i , j )} Fingerprint matrix of the jth component of the i-th line sensor l _max upper limit for the length of the vehicles to be recorded F ^{( j )} Average value over all line sensors of the fingerprint l length of the vehicle H vertical distance between line sensor and vehicle

Claims (15)

Method for identifying vehicles, with the steps (i) Acquiring magnetometer data from (a) a first magnetometer, (b) a second magnetometer, (c) a third magnetometer, (d) a fourth magnetometer, (e) a fifth magnetometer and (f) at least a sixth magnetometer, (g) wherein at least three magnetometers are arranged along a first straight line and form a first magnetic line sensor, (h) wherein at least three magnetometers are arranged along a second straight line and form a second magnetic line sensor, the second straight line running along the first straight line, so that a magnetic field data record is obtained which describes at least one magnetic field generated by the vehicle and which is spatially resolved in at least two dimensions, and (ii) Identifying a vehicle type and/or a vehicle based on the magnetic field data set. Method according to Claim 1, characterized in that identifying the vehicle type using the magnetic field data set has the following steps: (i) detecting a magnetic rigid part signature of non-rotating components of the vehicle from the magnetic field data set and (ii) Identifying a vehicle type from the magnetic rigid part signature. Method according to claim 2, characterized in that the detection of a magnetic rigid part signature comprises the following steps: (i) determining those areas in which the magnetic field measured by the at least two line sensors can be described within a specified error tolerance exclusively by a translation of a fixed magnetic field value at a vehicle speed, so that rigid part areas are obtained and/or (ii) Determining at least two areas in which the magnetic field measured by the at least two line sensors cannot be described within the specified error tolerance by a translation of a fixed magnetic field value with a vehicle speed, so that at least four wheel areas are obtained, and then calculating the rigid part areas by masking the wheel areas from the spatially resolved magnetization. Method according to one of the preceding claims, characterized by the step: automatically compensating for an influence of a yaw angle between the target running direction and an actual running direction of a vehicle passing over the line sensors. Method according to one of Claims 2 to 4, characterized by the steps: (i) Calculating a difference image (D) from • a first line sensor image describing the spatially resolved magnetization measured by the first line sensor, and • a second line sensor image that describes the spatially resolved magnetization measured by the second line sensor, (ii) determining a binary image from the difference image using a threshold filter, (iii) determining at least one contiguous area in the binary image and (iv) Setting the at least one contiguous area either as a rigid part area or as a wheel area. Method according to one of the preceding claims, characterized in that the identification of the vehicle using the magnetic field data set has the following steps: (i) scaling a time component of the magnetometer data of the magnetic field data set, the time component being selected such that it corresponds to driving over the at least one magnetometer at a predetermined target driving speed, and/or
normalizing a signal strength of the magnetometer data,
so that a scaled magnetic field data set is obtained, (ii) Reading out reference magnetic field data sets from a database that contains a large number of reference magnetic field data sets from different vehicles, (iii) determining a similarity parameter for each of the reference magnetic field data sets, which encodes a similarity of the scaled measurement field data of the magnetic field data set with the reference magnetic field data of the reference magnetic field data set, and (iv) identifying a vehicle based on the similarity parameter. Method according to one of the preceding claims, characterized by the steps: (i) acquiring an authorization parameter associated with the reference magnetic field data set identifying the vehicle and (ii) emitting a signal to an access release device so that it provides a route for the vehicle - releases if the permission parameter encodes permission to do so, and - do not release if the authorization parameter does not encode the authorization to do so. Method according to one of the preceding claims, characterized in that the magnetic field data describe the magnetic field in at least two dimensions, in particular in three dimensions. Method according to one of the preceding claims, characterized by the steps: (a) retrieving a total mass of the vehicle from a database using the similarity parameter or the vehicle type, and (b) sending a signal to the access release device which depends on the total mass and/or
Sending a signal encoding total mass to a monitoring device. Method according to one of the preceding claims, characterized by the steps: (a) determining a signal intensity parameter encoding a strength of the magnetic field from the magnetic field data and (b) determining the total mass from the signal intensity parameter and the scaled magnetic field data set. Method according to one of the preceding claims, characterized in that (a) a sampling frequency of at least one [in particular a plurality, preferably all] magnetometers is at least 50 Hertz and/or (b) for at least three magnetometers arranged along the straight line, the distance between adjacent magnetometers is at most 30 centimeters. Vehicle identification device for identifying vehicles, with (a) a first magnetometer for acquiring first magnetometer data, (b) a second magnetometer for acquiring second magnetometer data, (c) a third magnetometer for collecting third party magnetometer data, (d) a fourth magnetometer for acquiring fourth magnetometer data, (e) a fifth magnetometer for collecting fifth magnetometer data, (f) at least one sixth magnetometer for acquiring sixth magnetometer data and (g) an evaluation circuit for automatic
Acquiring the magnetometer data and
Identifying a vehicle or vehicle type based on the magnetometer data,
characterized in that (h) at least three magnetometers are arranged along a first straight line and form a first magnetic line sensor, (i) at least three magnetometers are arranged along a second straight line and form a second magnetic line sensor, the second straight line running along the first straight line, and (j) the magnetic field data describes a change in a static magnetic field over time, and (k) the evaluation circuit is designed to automatically carry out a method according to one of the preceding claims. Vehicle identification device according to claim 12 , characterized in that (a) the evaluation unit is designed to automatically carry out a method with the steps: (i) determining an actual vehicle speed from the magnetic field data from two of the magnetometers arranged along the straight line and (ii) scaling the time component of the magnetic field data of the magnetic field data set, the time component being selected such that it corresponds to driving over the at least one magnetometer at the specified target driving speed, based on the actual vehicle speed. Vehicle identification device according to one of claims 12 to 13,
characterized in that (a) at least the three magnetometers arranged along a straight line are arranged at the same height and (b) at least three magnetometers are located at a second elevation differing from the first elevation by an elevation difference. building with (a) a lane for vehicles, the lane having a lane direction of travel, and (b) a vehicle identification device according to any one of the preceding claims, arranged to identify a vehicle type of vehicles running on the roadway.

Images