ES2401324T3 - Procedure for the recognition of registration and categorization of vehicle tolls - Google Patents

Abstract

Procedure for automatic vehicle registration recognition, in particular for toll collection in road tolling systems, with the steps: a) make available a database (DB) to assign (5) possible registration reading results (LPOCR, mem) to vehicle license plates (LP) and reference fingerprints (FPref) assigned to them, b) perform the OCR reading of the vehicle license plate (LP) (2) as a result of current license plate reading (LPOCR ), with previously or subsequently c) perform the detection by vehicle characteristics sensor (2) as a current fingerprint (FP) of the vehicle, d) determine a license plate reading result (LPOCR, mem) equal to the reading result of current enrollment (LPOCR) in the database (DB) and all reference fingerprints (FP) assigned to it as candidate digital fingerprints (FPref, cand), e) compare the similarity of the current fingerprint (FP) with fingerprints s of candidate (FPref, cand) to determine the most similar candidate fingerprint, and f) output the vehicle registration (LP) assigned to the most similar candidate fingerprint (FPref, cand) as the result of automatic registration recognition, manually determining the vehicle registration (LP) for the current license plate reading result (LPOCR) if in step d) an equal license plate reading result (LPOCR, mem) cannot be determined, or in step e) no It is possible to determine a more similar candidate fingerprint (FPref, cand) or no similar candidate fingerprint (FPref, cand), and in these cases being assigned to the vehicle registration (LP) in the database (DB) of the result of Current license plate reading (LPOCR) as a result of new or additional license plate reading (LPOCR, mem), characterized in that the new or additional assignment is carried out only if the current license plate reading result (LPOCR) reaches a frequency minimum and / or minimum quality.

Description

Procedure for the recognition of registration and categorization of vehicle tolls

The present invention relates to a method for automatic vehicle registration recognition, in particular for toll collection in road toll systems, according to the preamble of claim 1.

Such a procedure is known from US2007 / 008179A1.

Road toll systems, which are based on automatic license plate recognition through images taken from the vehicle, are also identified as video toll systems. In this regard, vehicles are recorded with one or more cameras when passing through a toll collection station. The cameras record an image of the front and / or the rear of the vehicle, in which the license plate can be read better. From the image, the license plate of the vehicle is read by means of automatic license recognition methods (optical character recognition, OCR, optical character recognition). In a database of the toll exchange, a vehicle account with the vehicle registration is managed for each vehicle. The passage of the vehicle is assigned to that vehicle account, whose registration coincides with the result of the OCR registration reading.

If vehicles from different countries or different federated states pass through a toll collection station, as in the case of the US, it is often necessary to recognize, in addition to the registration, the country that issued the license plate. registration, the state or country shield or the type of license plate in order to enable a unique assignment of the license plate read to a vehicle account.

A problem with known procedures is the high error rate of OCR registration recognition. License plates with an erroneous reading must be manually corrected and assigned to a vehicle account. In the worst case, it can happen that an unread license plate is confused with a license plate of another vehicle and the pass is charged to the account of this vehicle.

Therefore, improvements have already been proposed to assign the current license plate image to a vehicle account and these improvements also take into account, in addition to the license plate itself, the appearance of the license plate. For this purpose, essential characteristics of a current license plate image are extracted. The essential features must uniquely represent the appearance of the license plate and are called "fingerprint". In addition to the registration, a reference fingerprint is also stored in each vehicle account in the database. By comparing the current fingerprint with the reference fingerprint of the database, the vehicle is assigned to a vehicle account.

Since the fingerprints are calculated from the images taken in different light and atmospheric conditions and the relative position between the license plate and the camera also varies during the passage through a toll station, the registered fingerprints of the same vehicle are slightly different in most cases, making comparison difficult. In addition, fingerprints comprise many features, that is, they resemble highly dimensional vectors, thus causing the search for similar fingerprints to require a large computing capacity and not be unlimitedly possible, even through database techniques, especially if the databases contain many records, as in video toll systems.

The invention aims to create procedures for automatic license plate recognition and toll categorization that have a lower error rate and greater reliability than known procedures and can be implemented quickly and easily.

This objective is achieved by means of a procedure for automatic registration recognition with the characteristics of claim 1.

The process of the invention is based on the use of OCR registration and fingerprint reading results, as well as on the knowledge that the errors of the OCR readings are not randomly distributed, but that the same registration repeatedly occurs. same erroneous readings. Thus, for example, it can happen that an ABC123 license plate is almost always read correctly and sometimes, however, is read as ABCI23 or A8C123. If automatic license plate recognition then reads ABC123, it is convenient to consider only those vehicle accounts, whose license plate is sometimes read as ABC123, to assign the toll step with the help of fingerprint comparison. In this regard, the method of the invention proposes to store in the database a list of possible reading results of the automatic license plate recognition containing only those known variants of license plate reading results for vehicle identification that meet the minimum requirements. of quality or quantity.

Therefore, the method of the invention takes into account the reading errors, selected so far, of the automatic recognition of OCR registration for each individual registration and thus provides better results:

-

 There are fewer assignable candidates in the vehicle database.

-

 The probability that the correct vehicle account appears in the list of assignable candidates is greater.

In an advantageous embodiment, the n: n assignment list of possible vehicle and license plate reading results is successively elaborated during the operation of the toll system. The new vehicle accounts created have registered only the correct registration in the list of possible registration results. Whenever an erroneous reading occurs, the reading result is checked manually, assigned to a vehicle account and recorded in the list of possible license plate reading results. It will only be necessary to process once each variant of the license plate reading and register it in the vehicle account.

Accordingly, the invention according to claim 1 is characterized in that if in step d) an equal enrollment reading result cannot be determined or in step e) a more similar candidate fingerprint or no fingerprint cannot be determined As a similar candidate, vehicle registration is determined manually for the current license plate reading result.

In these cases, the vehicle registration result is assigned in the database the result of the current license plate reading as a result of the reading of new or additional registration and the current fingerprint is also preferably assigned as a new or additional reference fingerprint If the result of the current license plate reading reaches a minimum frequency and / or a minimum quality.

Preferably also in the cases, in which

-

 the comparison of the current fingerprint with the fingerprint or the reference fingerprints, assigned to the vehicle registration, failure and the current fingerprint has a better quality or more information than the stored fingerprints, the current fingerprint is assigned to vehicle registration as a new or additional reference fingerprint; or in which - no reference fingerprint is stored for vehicle registration and the current fingerprint meets the minimum quality requirements, it is assigned to the vehicle registration as a new reference fingerprint.

An assignment is also preferably removed from the database if it is not used in a predefined period of time.

If the OCR license plate reading determines different variants of the license plate reading, it is also useful to use the reading variants in step d) during the search of the vehicle accounts or assigned fingerprints and / or store them as possible results of reading a license plate during the preparation of the assignment list.

The fingerprint can be generated in any way known in the art. It is especially advantageous for the sensor detection to be carried out by means of a camera that takes a picture of the license plate and / or the vehicle, taking away the characteristics of this image and calculating the fingerprint from this. Alternatively or additionally, sensor detection can be carried out by light barriers, laser scanners, induction loops, weight sensors or radio communication means that detect particular characteristics of the vehicle, from which it is calculated The fingerprint.

The investigations carried out by the applicant have shown that a large part of the records in the list of results of the license plate reading refers only to a very small amount of vehicle license plates. This may be due to the fact that there are particularly difficult individual license plates. Thus, for example, license plates ABC123 and ABCI23 could be available with a very similar embodiment. Therefore, another aspect of the invention consists in addition to each record in the list of possible license plate reading results information is stored to differentiate the fingerprints of the assigned vehicle license plates. If, for example, both the license plate ABC123 and ABCI23 are read as ABCI23, it is convenient to evaluate or weigh with a higher range especially the region, in which both license plates ("I" and "1") differ. This "information on the relevant differences" of two or more license plates, which could be confused due to automatic license plate recognition, is stored in the list of possible registration reading results as a weighting for the assigned fingerprints.

This is particularly convenient, because the license plates often differ only slightly, but, on the other hand, large variations of the fingerprint can be caused for the same vehicle due to dirt, light reflections or the like. These influences, significant in certain circumstances, may cause small local differences to be sufficiently taken into account when comparing fingerprints.

Accordingly, another preferred embodiment of the invention is characterized in that for each registration reading result, preferably for each individual assignment thereof to a vehicle registration, weightings for individual regions of the reference fingerprints are stored in the database assigned, which are taken into account in the comparison of similarity. These weights constitute the aforementioned "information on the relevant differences".

In toll systems there is often the problem that vehicles, not subject to tolls, are classified as vehicles subject to tolls. An example is the classification of a large tourism not subject to tolls (for example, a van) as a small van subject to tolls. This misclassification is due to inaccurate measurements in the classification sensor system or is conditioned simply by the fact that some vehicle classes cannot be delimited by means of defined external characteristics. In these cases, the classification error must be corrected manually. Often, the correction in the same vehicle must be performed during each step of the vehicle. The application of the method according to the invention may also be useful in these cases if a vehicle record is stored in the vehicle list, in which it has been indicated that the vehicle is not subject to tolls. The problem here is that in this way a record is created for a vehicle not subject to tolls. Such a procedure may violate the regulations on data protection depending on the legal situation. To solve this problem, the procedure presented can also be used in an advantageous embodiment so that the results of the license plate reading are coded before being stored in the list.

Accordingly, a preferred embodiment of the invention is that the license plate reading results are stored coded in the database and the current license plate reading result for the search for equality in step d) is coded in the same way.

Additionally, the method of the invention can preferably be used for the toll categorization of the vehicle, since in the database a toll category is also assigned to each reference fingerprint and, if this reference fingerprint is defined as the most similar candidate fingerprint, it is output at the same time to its assigned toll category.

Other objects, features and advantages of the invention derive from the following description of its preferred embodiments with reference to the accompanying drawings. They show:

Fig. 1 a flow chart of the method according to the invention for automatic license plate recognition; and Fig. 2 a flow chart of the method according to the invention for toll categorization.

Figure 1 shows the procedure for automatic license plate recognition by means of an image 1 of a vehicle 2 with an LP vehicle registration within the framework of a road toll system. The term "vehicle" used here encompasses any land, air and marine vehicle, both self-propelled and driven by outside force.

In a first step (b) an OCR reading of the LP license plate of the vehicle 2 is carried out from the image 1 to obtain a result of the current LPOCR license plate reading, for example, "ABC123".

In a previous, parallel or subsequent step (c), one or more particular characteristics of the vehicle 2 are detected by sensor to obtain a so-called "current fingerprint" FP of the vehicle 2. From these particular characteristics are, for example, the appearance of the license plate, the color of the vehicle, the appearance of the front of the vehicle that can be used to verify the identity of the vehicle, and / or the dimensions of the vehicle, the shape of the vehicle, the presence of a trailer, etc., which can also be taken into account, for example, for toll collection.

The FP fingerprint can be extracted, for example, directly from photographic image 1, for example, by determining significant light-dark areas, areas of image parameters, such as contrast, horizontal sums, etc., namely , both of the entire vehicle 2 and parts of the vehicle 2. In particular it is possible that the FP fingerprint is formed in this way from the image of the LP vehicle registration itself.

Alternatively or additionally, the FP fingerprint is determined from other images or more images 1 of the vehicle 2 and from sensor data, such as laser scanners, light barriers, induction loops, camcorders, Weight sensors, radio installations, for example, inquiries to an RFID transponder, etc. In this regard, different sensor data can also be combined in the same FP fingerprint.

In a step (d) the result of reading LPOCR registration is consulted in a DB database of the road toll system. In the embodiment shown, the DB database contains a first list 3 with LPOCR stored license plate reading results, a second list 4 with all LP vehicle registrations registered in the road toll system and all fingerprints of FPref reference assigned to these. Among lists 3, 4 there is a n: n (n EN) relational assignment of the LPOCR license plate reading results, mem to LP vehicle license plates, on the one hand, or to their FPref reference fingerprints, on the other hand , as represented by lines 5. An LPOCR license plate reading result, mem can refer to one or more LP vehicle license plates or their FRref reference fingerprints; and for one LP vehicle license plate (or an FPref reference fingerprint) one or more possible LPOCR license plate reading results, mem.

Therefore, in the query step (d) the current LPOCR license plate reading result can be consulted directly in list 3 of the LPOCR stored license plate reading results, mem, because for each possible LPOCR license plate reading result only there is a single record equal LPOCR, mem in list 3. This facilitates and significantly accelerates the implementation.

In a next step (e), all FPref reference fingerprints, assigned to the current LPOCR license plate reading result by lists 3, 4, of the LP vehicle license plates in question are determined as "candidate fingerprints" FPref , cand and are compared respectively with the current FP fingerprint. This comparison is a comparison of similarity, in which the "most similar" candidate fingerprint is determined FPref, cand; LP vehicle registration assigned to this fingerprint is output as a result of the registration recognition procedure (step f).

To determine the "most similar" candidate fingerprint, the most diverse similarity criteria can be predefined, for example, the maximum value of a numerically calculated similarity measure and / or the overcoming of a minimum similarity measure, etc. When determining the "most similar" candidate fingerprint, it is also possible to verify whether the similarity criteria do not apply to any other candidate fingerprint and, therefore, if the "most similar" candidate fingerprint is uniquely determined. , that is, if only one unequivocally more similar candidate fingerprint has been used for further processing.

In list 4 of the DB database, several FPref reference fingerprints can also be stored for an LP vehicle registration. Thus, for example, for the same LP license plate, reference fingerprints FPref can be calculated from different images showing the license plate of the vehicle 2 under different lighting conditions or viewing angles. In the fingerprint similarity comparison of step (e), similarity comparisons of the current FP fingerprint can be combined with all FPref candidate fingerprints, cand of the same LP license plate to obtain a common similarity value, for example, by sum, selection of the maximum value, etc.

If in step (d) an equal LPOCR license plate reading result cannot be determined, mem in list 3 or in step (e) a more similar candidate fingerprint or similar candidate fingerprint cannot be determined FPref, cand, a step (g) is executed, in which the LP license plate is determined by an operator manually, in particular visually, from the image 1 or the result of the current LPOCR license plate reading and from this it is generated A new record for the DB database. The manually determined LP vehicle registration result is then assigned the LPOCR current license plate reading result as a result of the new or additional LPOCR license plate reading, mem and / or the current FP fingerprint is assigned as a new or additional reference fingerprint FPref. In this sense, it is also checked whether the current LPOCR license plate reading result or the current FP fingerprint reaches a minimum frequency and / or a minimum quality that qualifies it to be registered in the DB database.

Preference is also given to criteria for eliminating assignments of LPOCR license plate reading results, mem to LP license plates in list 3 of the DB database, for example, if they are not accessed for a prolonged period of time, for example, that the license plate was renewed and can be read better now, so that certain registration reading results no longer appear.

In a step (h), WT weights can also be optionally established for FPref reference fingerprints and stored in the DB database, for example, for each LPOCR assignment, mem to an LP license plate, taking into account these WT weights in the fingerprint comparison of step (e) in order to place them in especially distinctive or relevant areas of the FP and FPref fingerprints, which are to be compared.

It is also possible that for certain LP vehicle license plates or certain LPOCR license plate reading results, if there are increasingly frequent allocation errors, high requirements have to be met with respect to comparing fingerprint similarity in the step (e) or an automatic assignment of them is completely dispensed with.

Some procedures for the recognition of OCR registration provide several alternative enrollment reading results, one of these being rated as the most likely outcome in most cases, while the other results represent possible, but less likely, variants. The alternative enrollment reading results can be used effectively in step (d) and step (g). In step (d), the alternative reading results can also be consulted in the list 3 of stored license plate reading results LPOCR, mem and the assigned reference fingerprints FPref can be used for fingerprint comparison. In step (g) all possible variants of the license plate reading can be registered in the DB database.

As just explained, the LPOCR license plate reading results, mem stored in list 3 may be encoded. The current LPOCR license plate reading result is then coded in the same way for the equality comparison in step (d), thus it is unnecessary to use, transmit or save in a legible text the license plate reading results in the DB database. or within the framework of this procedure.

The coding can be carried out by any known procedure that guarantees authorized access to the results of license plate reading LPOCR, LPOCR, mem. In fact, in the case of the procedure presented here it is not necessary to know the result of reading the license plate in readable text or re-decoding it, that is, irreversible coding can be used. It is sufficient that the coding is an injective function in the mathematical sense and that, therefore, each coding result is generated only exactly from a registration reading result, since the procedure is based on comparing a registration reading result current with a list of stored tuition reading results and looking for exact matches and not just similar enrollments. However, this search for an exact match can also be performed if both the current LPOCR license plate reading result and the LPOCR stored license plate reading result, mem are coded in the same way.

The coding is advantageously a so-called "unidirectional function", that is, a difficult function to reverse, for example, an asymmetric coding with public key and private key. The private key, which would be required for decoding, can be destroyed, so no one can deduct the actual license plate reading result from the coded license plate reading result.

The coding of the results of the license plate reading also allows the procedure to be applied in section control systems, parking systems or other closed systems that apply especially strict requirements regarding data protection. In this case, a distinction must be made between closed systems, in which a vehicle passage when leaving the closed system is assigned to a vehicle passage when entering the closed system, and systems with registered users, in which the vehicle passes are assigned to a vehicle account. In closed systems (without registration) it makes no sense to store a list of possible enrollment results. However, in these systems, a fingerprint preselection can also be done by means of the registration reading result, and especially in these application cases it may be necessary to encode the registration reading results for reasons related to data protection.

A very advantageous variant of the described procedure consists in also using the place and time of a vehicle crossing when identifying the vehicles in order to plausibilize the assignment of the vehicle passes to vehicle accounts. In this case it is checked, for example, from several vehicle detections, if a vehicle can actually be in a certain place at the indicated time.

Figure 2 shows another advantageous embodiment of the invention, in which, in addition to the automatic license plate recognition, an automatic toll categorization of a vehicle is also executed, and specifically based on the comparison of fingerprints that have been explained, the fingerprint may contain characteristics that allow to verify the properties of the vehicle. To this end, one or more MK toll categories are also assigned to the fingerprint or FPref reference fingerprints of a vehicle identified with an ID identification, for example, its LP vehicle registration. If the current FP fingerprint of the vehicle passage is as close as possible to one of the FPref candidate fingerprints, as determined in the manner mentioned above, the corresponding toll category MK of the DB database is also applied and Outputs it in step (f) (see also figure 1).

The following example illustrates a possible variant of this procedure.

In a first step of the vehicle, a van passes through a toll station. The sensor system, for example, a laser scanner installed at the toll station, detects the vehicle and classifies it (erroneously) as a small van. Since the van pays only for the toll category (lower) "van", a contradiction is created. In the first step of the vehicle, which generates a contradiction, a manual processing is carried out, during which the contradiction is corrected and the van is correctly classified with the "van" toll category. When the contradiction is corrected, the fingerprint is stored in the vehicle account together with the toll category set by the operator.

If a contradiction occurs once again in the same vehicle, the fingerprint of the current step is compared with the reference fingerprint stored in the vehicle account. If both fingerprints match (that is, the similarity measure exceeds a predefined threshold value), the toll category set by the operator in the first step is applied and the contradiction is automatically corrected.

The procedure is based on the knowledge that the same vehicle, which has not changed its exterior shape, can be reassigned to the same toll category. In this regard, in a first step the identification of the vehicle or the assignment to a vehicle account is carried out, for example, with the aid of the automatic registration recognition procedure described above. In a second step, it is guaranteed by comparing the fingerprint that changes are recognized in the vehicle and that the toll category stored in the vehicle account is only applied if the vehicle has not changed externally. Examples of these changes in the vehicle can be mentioned the renewal of the vehicle (for example, the same registration number in a new vehicle) or, for example, the presence of a trailer or a caravan.

Therefore, the procedures, which check whether the correct toll category was paid, use only the data measured by the sensors to determine the toll category based on an identical classification procedure for all vehicles. Therefore, the categorization procedure presented here is capable of storing individual vehicle characteristics and reusing them in a subsequent categorization decision.

As can be seen in the database of Figure 2, a vehicle with a specific ID or LP identification can also be assigned different FPref reference fingerprints for its appearance in different K configurations with corresponding MK toll categories, so It is possible to categorize and charge the toll automatically to the same vehicle in different K configurations, for example, with and without a trailer.

In a simplified embodiment of the procedure, only the external dimensions of the vehicle can be stored as an FPref reference fingerprint; The presence of a trailer can also be easily recognized, for example, from a large change in length.

Another very advantageous embodiment consists in storing trailers as their own vehicles in the DB database with their own reference fingerprint. During the passage, the tractor vehicle and the trailer are identified separately and the properties of the tractor vehicle and the trailer are optionally verified. The determined toll category MK is derived from a combination of the categories of the towing vehicle and the trailer.

The determination of the MK toll category with the help of FP fingerprints can be carried out on the basis of the registration recognition procedure presented here by means of the OCR LPOCR license plate reading results, in which case it can be used directly Your LP license plate as vehicle identification ID. Alternatively, any other known method for vehicle identification could also be used, for example, vehicle identification by means of an electronic tag attached to the vehicle, such as an RFID transponder chip or an onboard unit (OBU) for electronic toll collection.

Figure 2 schematically shows the components of the procedure for automatic toll categorization with the help of fingerprints and other system components involved in it, as follows:

-

a "vehicle identification" component, which executes step (b), identifies the vehicle and assigns it to a vehicle account;

-

 a "retrieve reference sensor fingerprint" component, which executes step (d), downloads the fingerprint or reference fingerprints assigned as candidate fingerprints from the vehicle account of the identified vehicle and from the assigned database records DB;

-

 a "generate sensor fingerprint" component, which executes step (c), calculates the current fingerprint of the vehicle passage;

-

 a "fingerprint comparison" component, which executes step (e), compares the current fingerprint with one or more candidate fingerprints and results in whether and with which candidate fingerprint there is a match; and - another component "enforcement decision" uses the result of the fingerprint comparison together with the result of a separate classification ("automatic classification") and a toll category ("claimed class") indicated by the vehicle user and Check if there is a contradiction.

The invention is not limited to the embodiments shown, but includes all variants and modifications that fall within the scope of the appended claims.

Claims (9)

1. Procedure for the automatic recognition of registration of a vehicle, in particular for toll collection in road toll systems, with the steps: a) make available a database (DB) to assign (5) possible license plate reading results (LPOCR, mem) to vehicle license plates (LP) and reference fingerprints (FPref) assigned to them, b) perform OCR reading of the license plate (LP) of the vehicle (2) as a result of the current license plate reading (LPOCR), with previously or subsequently c) perform the detection by vehicle characteristics sensor (2) as a current fingerprint (FP ) of the vehicle, d) determine a license plate reading result (LPOCR, mem) equal to the current license plate reading result (LPOCR) in the database (DB) and all reference fingerprints (FP) assigned to it as candidate fingerprints (FPref, cand), e) compare the similarity of the current fingerprint (FP) with candidate fingerprints (FPref, cand) to determine the most similar candidate fingerprint, and f) output the vehicle registration (LP) assigned to the hu her digital candidate (FPref, cand) more similar as a result of automatic enrollment recognition, manually determining the vehicle registration (LP) for the current license plate reading result (LPOCR) if in step d) an equal license plate reading result (LPOCR, mem) cannot be determined, or in step e) no a more similar candidate fingerprint (FPref, cand) or no similar candidate fingerprint (FPref, cand) can be determined, and in these cases the vehicle registration (LP) in the database (DB) is assigned the Current license plate reading result (LPOCR) as a result of new or additional license plate reading (LPOCR, mem), characterized in that the new or additional assignment is carried out only if the current license plate reading result (LPOCR) reaches a frequency minimum and / or minimum quality.

2.

Method according to claim 1, characterized in that if the comparison of the current fingerprint (FP) with the fingerprint or the reference fingerprints (FPref) assigned to the vehicle registration (LP) fails and the current fingerprint (FP) presents better quality or more information than stored reference fingerprints (FPref), the current fingerprint (FP) is assigned to the vehicle registration (LP) as a new or additional reference fingerprint (FPref).

3.

Method according to claim 1 or 2, characterized in that if no reference fingerprint is stored for the vehicle registration (LP) and the current fingerprint (FP) meets the minimum quality requirements, this fingerprint is assigned to the license plate vehicle (LP) as a new reference fingerprint (FPref).

Four.

Method according to one of claims 1 to 3, characterized in that an assignment (5) is removed from the database (DB) if it is not used in a predefined period of time.

5.

Method according to one of the claims 1 to 4, characterized in that the sensor detection is carried out by means of a camera that takes a picture (1) of the license plate (LP) and / or the vehicle (2), extracting the characteristics of this image and calculating from this the fingerprint (FP).

6.

Method according to one of claims 1 to 5, characterized in that the sensor detection is carried out by means of light barriers, laser scanners, induction loops, weight sensors or radio communication means that detect particular characteristics of the vehicle (2) , from which the fingerprint (FP) is calculated.

7.

Method according to one of claims 1 to 6, characterized in that for each registration reading result (LPOCR, mem), preferably for each individual assignment thereof to a vehicle registration (LP), they are stored in the database (DB ) weights (WT) for individual regions of the assigned reference fingerprints (FPref), which are taken into account in the comparison of similarity.

8.

Method according to one of claims 1 to 7, characterized in that the registration reading results (LPOCR, mem) are stored encoded in the database (DB) and the current registration reading result (LPOCR) for the search for equality in step d) it is coded in the same way.

9.

Method according to one of claims 1 to 8 additionally for the categorization of vehicle tolls, characterized in that a toll category (MK) is also assigned to each reference fingerprint (DB) and, if This reference fingerprint (FPref) is determined as the most similar candidate fingerprint (FPref, cand), along with it, the assigned toll category (MK) is output.

 REFERENCES CITED IN THE DESCRIPTION This list of references cited by the applicant is solely for the convenience of the reader. It is not part of the European patent document. Despite the care taken in the collection of references, errors or omissions cannot be excluded and the EPO denies all responsibility in this regard. Patent documents cited in description10 US2007008179A1 [0001]