GB2569173A - Anti-fraud measures in relation to cheques - Google Patents

Abstract

A cheque 10 carries visible markings 12-30 representing cheque data in natural language comprising details of a payment to be made to a payee, and a computer readable marking 34 representing validation data comprising selected cheque data. The validation data is encrypted by a key constructed from key generation data comprising cheque data represented in the visible markings. The key may be a hashed version of the key generation data. The validation data may comprise a sum payable 30, a payee name 24, and a cheque serial number 18. The computer-readable marking may be in the form of a response code or a QR code (RTM). The cheque may be analysed by digitally imaging the cheque, machine reading the visible markings from the image to obtain cheque data from which the key generation data is selected, and determining the key from the key generation data. The computer-readable marking may then be read and its content decrypted by use of the key to obtain the validation data, which is then compared with the cheque data to detect mismatches between them. Except for the payee name, the validation data may be truncated if it exceeds a predetermined length.

Description

ANTI-FRAUD MEASURES IN RELATION TO CHEQUES

The present invention is concerned with measures to be taken to enable detection of attempted fraud in relation to cheques.

The word cheque, or in US English check, refers to a document authorising a bank to make a payment from a specified account to a specified person or entity. Despite the increasing use of purely electronic means of payment, paper cheques are still used for a large number of transactions. Typically, to obtain payment the recipient of the cheque (the payee) deposits it at a first bank (the collecting bank). Through the clearing system, in typical paper clearing environments, the cheque is transferred from the collecting bank to a second bank on which the cheque was drawn (the drawer's bank). The second bank then makes the specified transfer of funds to the collecting bank, (or 'beneficiary bank', if different) and the funds are credited to the payee's account.

Personal and business cheques are often issued to an account holder in pre-printed books with marked but empty fields on them for the account holder to write in details such as date, payee, amount and so on. But an alternative much used by businesses issuing multiple cheques is to print much of the cheque content themselves. Commercially available software packages can print the entire visible cheque content, including the payee name, amount, and so on, in a form acceptable to the cheque clearing system.

Cheque fraud is, and has long been, a major issue for the cheque clearing system. Fraudsters are known to tamper with cheques in a variety of ways in order to try to obtain improper payments from banks. This may for example involve theft of bona fide cheques which are then fraudulently infilled and signed prior to presentation to a bank. Tampering with a cheque covers both forgery of the cheque itself and illicit alteration of a genuine cheque's information content. There are various ways in which cheques can be altered for fraudulent purposes. For example, by altering the payee on a completed cheque, the fraudster may divert payment into its own account. This type of tampering often involves adding characters to the end of the payee name. By altering the amount, the fraudster may obtain a larger payment than it is entitled to.

At the time of writing, a change is underway in the cheque clearing systems of various countries. Traditionally the clearing system has been based on transfer of the paper cheque itself from the collecting bank via the clearing system to the drawer's bank. Given the large daily volume of cheques handled by the clearing system, and despite the use of automatic sorting methods, this exchange of paper documents is a substantial burden for banks. An alternative to reliance on exchange of the paper cheques themselves is to electronically scan the cheques, so that banks exchange digital cheque images and data rather than the paper documents. Making use of modern computer networks, this process can be considerably less onerous than sorting and transport of the paper cheques. Cheque clearing based on exchange of electronic images has at the time of writing been adopted in some countries but has yet to be taken up in others. The UK clearing system is currently reliant on exchange of paper cheques, (although some images are currently exchanged, but not to the exclusion of the exchange of paper) and it is planned that it will move to use of scanned cheque images in the near future, the paper cheque being discarded once it has been scanned.

In the context of an image-based cheque clearing system, measures are needed to combat fraud which are not reliant on inspection of the paper cheque itself. Such measures need to enable cheque validation to be implemented in software by analysis of the cheque's digital image, and to be reliable to a high degree.

US6,073,121, Ramzy, discloses a cheque fraud prevention system in which each issued cheque has a line of machine-only readable symbols, which may take the form of a bar code, that contains all the information printed on the cheque encrypted according to a key-selectable encryption algorithm. When the cheque is presented to a bank teller, he/she inserts it in a reader which:-

- reads the printed, human-readable, information on the cheque;

- decodes the bar code; and

- decides whether the printed data and the decoded data match, rejecting the cheque if they do not.

Certain practical problems arise which are not addressed in the Ramzy patent.

There is the problem of how the data should be encrypted in a manner which is adequately secure against imitation by a malfeasor. US6,073121 refers to a key selectable encryption algorithm said to be made selective by the operator, who can periodically change the key, after which all banks and companies using the system would have to be notified of the key code changes. This manner of encryption with periodic key code changes is not thought to be practical in a cheque clearing system, at least without a major change in operating processes.

The system needs to be robust against errors caused by misreading of human-readable data from the cheque, especially since resolution of the scanned cheque images is limited. False positives, in which a valid cheque is wrongly rejected, are a problem in this context as well as false negatives, in which a fraudulent cheque is passed for payment. Given the huge number of cheques dealt with daily in the clearing system, even a very small proportion of false positives has the potential to create a great deal of labour in further checks, or in dealing with issues following erroneous rejection of a cheque.

The limited image resolution also limits the amount of data that can be encoded in machine-readable form, which can be problematic if the payee name is a long one, especially as fraud often involves adding characters at the end of the payee name, so that it is desirable to avoid truncating the name if possible.

The solution of one or more of these problems is an objective of the present invention.

According to a first aspect of the present invention there are (a) a cheque, (b) a method of analysis of a cheque and (c) a computer-implemented method of generating a cheque according to the appended claims.

Specific embodiments of the present invention will now be described, by way of example only, with reference to the accompanying Figure 1, which shows a cheque embodying the present invention.

The cheque 10 represented in Figure 1 comprises a thin rectangular substrate which is paper in this example although it could in principle comprise some other suitable material, subject to applicable regulations in the relevant country. Printed upon the cheque in human-readable form (natural language) are certain items of cheque data which will be common to multiple cheques issued by a given drawer:-

- a bank name/logo 12,

- a bank branch sort code 14,

- a number line comprising the bank branch sort code 16 and the drawer's bank account number 18. In this example these are printed in magnetic ink, in a font which is suited to reading by MICR (magnetic ink character recognition) (the number line also includes a cheque serial number 22, which of course varies from one cheque to the next);

- the identity 20 of the drawer. This is an example of a corporate cheque where the identity is often included in this or a similar position. A personal cheque will not show this, only by reference to an account name adjacent to the signature position.

Additionally the cheque 10 includes variable human-readable printed cheque data which will vary from one cheque to another issued by a given drawer:-

- the aforementioned cheque number 22

- the payee name 24

- the legal amount 26 - the sum payable, expressed in words,

- the date 28,

- the courtesy amount 30 - the sum payable, expressed in numerical digits, and

- a signature space 32, since in this example the cheque requires a manual (wet ink) signature.

In accordance with the present invention, the cheque 10 additionally carries a machine-readable marking 34 which, in the present exemplary embodiment, is:-

- printed on the cheque 10;

- optically readable. In the present embodiment it is visible to the naked eye, although in principle it could be detectable only by use of detectors operating at different electromagnetic frequencies. The term optical in this context refers to techniques involving light, but not necessarily light in the visible part of the electromagnetic spectrum. For instance cheque clearing can involve imaging the cheque using a UV sensitive detector, and the marking 34 could use UV visible ink;

- not in a natural language form. The marking 34 may be essentially unreadable by humans. It could take the form of a bar code, for instance, but in the present embodiment it comprises a 2D response code of the type commonly referred to as a QR Code, which is an abbreviation for Quick Response code and is a registered trade mark of Denso Wave Incorporated. Such codes are of course well known in themselves to the skilled person. Software for their generation is open source and is widely available. The present invention may be implemented using other 2D response codes or using other printable codes which are optically readable and able to be interpreted by computer system;

- image survivable, meaning that when the cheque has been imaged for entry into an imagebased clearing system, the marking 34 is present in the resultant image and can be read in it.

The marking 34 represents in encrypted form a subset of the cheque data which is also printed in human-readable form on the cheque. In the present embodiment the data encrypted in the marking 34 includes both fixed cheque data common to multiple cheques from the same drawer and variable cheque data which is expected to vary from one cheque to another. Specifically it includes the drawer's account number and sort code, the cheque number, date, amount payable and payee name. A different selection of data may be used in other embodiments of the present invention. The marking 34 also includes a key, whose creation and function will be explained below.

The data represented in the marking 34 will be referred to below as the validation data, regardless of the form in which it is represented. The validation data is, in the present embodiment, represented in a machine-readable, encrypted form in the marking 34. It is also printed in human-readable form on the cheque.

When the cheque is imaged and read in the clearing system, the important items of information can thus be read from the image in two different ways - by machine reading of the human-readable data 12 - 30 and by reading and decryption of the marking 34. A discrepancy between the two may be indicative of attempted fraud and may lead to rejection of the cheque.

Consider for example the type of cheque tampering in which the malfeasor alters the legal and courtesy amounts 26, 30, or the payee name 24 on the paper version of an otherwise valid cheque prior to its presentation. Such tampering may, if well done, not necessarily be evident from a study of the cheque image used in the clearing system, but it will - unless corresponding adjustment is made to the marking 34 - result in a discrepancy between that and the human-readable data on the cheque, which can be detected in the clearing system and used as a basis for rejection of the cheque.

Other attempts at fraud might involve counterfeiting of a complete cheque by the malfeasor. But unless the malfeasor is in possession of the algorithm used to encrypt the cheque data in the marking 34, it will not be able to generate a marking to match the human-readable cheque data. If an incorrect marking 34 is used, then the counterfeit cheque can be detected as such in the clearing system.

It is desirable that the marking 34 should be encrypted in a manner which is difficult for potential malfeasors to break. That is, it should be difficult or impossible for malfeasors to infer from a study of valid cheques how the encryption is carried out, in order to carry out the same process themselves.

In accordance with the present invention, part of the cheque data which is printed in human-readable form on the cheque is used to generate a key used in encrypting the validation data prior to its representation in the marking 34. The key is needed in order to decrypt the validation data. The cheque data used to generate the key (the key generation data) can in principle be taken from any of the fixed cheque data 12-20 and/or the variable cheque data 22 - 30 written upon the cheque in human-readable form. It is preferred however that it includes variable cheque data, so that different cheques will have different keys. Of course it is possible that two or more cheques may be issued which have identical variable data in some fields, as for example where two cheques are made out to the same payee for the same amount. To avoid the key itself being duplicated from one cheque to another, the cheque number is in the present embodiment included in the key generation data. It is not expected that two cheques from the same drawer will be issued with the same cheque number, so in this way it can be ensured that the key generation data is unique for each cheque. An alternative would be to include at least the closing digit or digits of the cheque number, making any duplication of the key generation data improbable although not necessarily impossible.

In the present embodiment the key generation data also includes digits of the legal amount 26 and characters taken from the payee name 24.

In other embodiments the key generation data may omit the legal amount and the payee name. The key generation data is in some embodiments drawn only from data represented in the number line, which is easily machine readable.

The key generation data is hashed to generate the key itself. Suitable hashing functions are well known to the skilled person. The key will typically be a character string which is shorter than the key generation data from which it is derived.

The aforementioned validation data can be assembled in a character string to be supplied to the encryption algorithm. In the present embodiment the string is of a fixed length of the order of 100 to 200 characters. Some fields of the validation data, such as the account number and sort code, are of fixed length. But the amount payable and the payee name are strings of variable length. If they are especially long, e.g. because the payee has a long name, then the entire validation data set may be too long to fit in the available string length. In this case the validation data is truncated to fit in the string length. In the present embodiment truncation is not applied first to the payee name, since it is desirable to include it in full if at all possible given that fraud often involves inserting extra characters at the end of the payee name, making it important to be able to detect tampering of that sort. Instead other parts of the validation data are truncated, according to a predetermined order of preference, and the payee name itself is truncated only if necessary after those other parts of the data have been truncated as far as possible.

The character string includes field delimiter characters to enable it to be parsed despite some parts of the string being of variable length. For example, delimiter characters may mark the beginning and end of the amount payable field, and the beginning and end of the payee field. Data fields of normally fixed length may be grouped together, with a field delimiter character at their end. In this case if truncation of these fields is carried out then when the character string is parsed, the number of characters removed in the truncation can be determined by determining how far from its usual position the relevant string delimiter character has moved. Which specific characters have been removed in the truncation process can be determined by reference to the known order of preference.

Once generated, the character string is encrypted using the aforementioned key and converted to form the marking 34.

As mentioned above, the marking 34 takes the form of a Q.R code in the present embodiment. Testing has confirmed that following imaging of the cheque for the purposes of an image based cheque clearing system, a Q.R code containing the validation data remains readable to an acceptably low degree of error.

Q.R codes have what is referred to as an error correction level. A code with a high error correction level is more reliably readable than one with a lower level. But for a given number of pixels in the Q.R code, the data carrying capacity decreases as the error correction level increases. The length of the character string made up of the validation data varies from cheque to cheque according to the length of the payee name and of the amount payable. To minimise misreads or failed reads of the character 34, the present embodiment provides for the error correction level to be set with reference to the length of the string of validation data, a higher correction level being set if the string is short and a lower correction level being set if the string is long.

The marking 34 can be generated in a software package used to generate and print cheques. Such packages (albeit lacking the ability to generate and apply the marking 34 in accordance with the present invention) are widely available and used by a range of organisations that have a need to issue cheques in quantities. The marking may in other embodiments be generated elsewhere, e.g. in a remote server which could be cloud based.

Consider now what happens when the payee presents the cheque at a bank. That bank scans the paper cheque and the resultant image is put into the clearing system. The MICR number line 16, 18, is also read in known fashion, its content being associated with the cheque image. The cheque image must also be read to obtain from it the relevant data about the intended transfer of funds payee identity, account details, sum payable and so on. This can be done in known manner using techniques referred to as OCR (optical character recognition). In the clearing system the cheque image and associated data/metadata are transferred through a computer network to the bank on which the cheque is drawn.

Before funds are exchanged, the cheque image will also be subject to tests intended to determine whether it has been forged or tampered with. Such tests are, in the present embodiment of the invention, carried out in the clearing system itself, as a service to the banks, and include an authentication process carried out using the marking 34.

The authentication process comprises several steps. In order to decrypt the marking 34, the key must be reconstructed from the human-readable data on the cheque image. This data is obtained by reading the image using OCR, and also by MICR reading of the number line. The key generation data can then be selected from the data read from the cheque, and hashed using the known algorithm to reconstruct the key.

The marking 34 is read from the cheque image in known manner. Using the reconstructed key, its data content can then be decrypted to reconstruct the character string containing the validation data. The character string is then parsed and its data content compared with the data read from the cheque by OCR/MICR.

As noted above, it is important that the validation process should be resistant to errors, be they false negatives or false positives. Known OCR techniques are not in themselves sufficiently error free for this purpose. But there is a statistical pattern to the errors made in OCR. For example some characters are difficult to distinguish and hence prone to be mistaken one for another. Because the present process involves comparing one known string against another, it is can be configured to be tolerant of certain OCR errors and so to avoid many false positives that might otherwise occur. For example, it may be that o is commonly misread as c in OCR. If the actual validation data, encrypted in the marking 34, contains an o but the string reconstructed by OCR contains instead a c, the system may be configured to ignore the discrepancy, or for example to ignore a certain proportion of such discrepancies before indicating a problem with the cheque.

Although fraud detection has been discussed above as a major function of the present invention, it should be noted that the marking 34 also provides resistance to simple errors in reading of the cheque which might otherwise result in transactions being wrongly processed.

The aforegoing embodiments are presented by way of example and not limitation. Any number of variants and alternative embodiments are possible without departing from the scope of the invention as expressed in the appended claims. For instance the marking 34 is a Q.R code in the specific embodiment, but various other computer readable codes exist and the marking may take any suitable 5 alternative form.

Claims (18)

1. A cheque carrying

- visible markings representing in natural language form cheque data comprising details of a payment to be made to a payee, and

- a computer readable marking which represents in encrypted from validation data which comprises selected cheque data the validation data being encrypted by use of a key constructed from key generation data which comprises cheque data represented in the said visible markings.

2. A cheque as claimed in claim 1 in which the key is a hashed version of the key generation data.

3. A cheque as claimed in claim 1 or claim 2 in which the validation data comprises a sum payable and a payee name.

4. A cheque as claimed in claim 3 in which the validation data further comprises a cheque serial number.

5. A cheque as claimed in any preceding claim in which the computer readable marking is in the form of a response code.

6. A cheque as claimed in any preceding claim in which the computer readable marking is in the form of a Q.R code.

7. A method of analysing a cheque of the type claimed in any preceding claim, the method comprising:

digitally imaging the cheque to obtain an image;

machine reading the visible markings from the image to obtain the cheque data;

selecting the key generation data from the cheque data;

determining the key from the key generation data;

reading the computer readable marking from the image and decrypting its content by use of the key to obtain the validation data;

comparing the validation data with the cheque data machine read from the visible markings to detect mismatches between them.

8. A method as claimed in claim 7 in which the determination of the key comprises hashing the key generation data.

9. A computer-implemented method of generating a cheque, the comprising:

receiving cheque data comprising details of a payment to be made to a payee, constructing a key from key generation data, the key generation data comprising selected cheque data, using the key to encrypt validation data, the validation data comprising selected cheque data, generating a computer readable marking representing the encrypted validation data, and generating a printer ready file for printing of a cheque which carries the computer readable marking and which also carries the cheque data in natural language.

10. A computer-implemented method as claimed in claim 9 further comprising truncating the validation data if it is in excess of a predetermined length.

11. A computer-implemented method as claimed in claim 10 in which the validation data comprises a payee name whose length is able to vary from one cheque to another, and in which validation data other than the payee name is truncated in preference to it if the validation data is in excess of the predetermined length.

12. A computer-implemented method as claimed in any of claims 9 to 11 in which the key generation data comprises a cheque serial number.

13. A computer-implemented method as claimed in any of claims 9 to 12 in which construction of the key comprises hashing the key generation data.

14. A computer-implemented method as claimed in any of claims 9 to 13 in which the computer readable marking is in the form of a response code.

15. A computer-implemented method as claimed in any of claims 9 to 14 in which the computer readable marking is in the form of a QR code.

16. A computer-implemented method according to claim 15 further comprising selecting error correction level of the QR code with reference to length of the validation data.

17. A computer-implemented method according to any of claims 9 to 16 which is carried out within an automated cheque clearing system.

18. A computer system for analysis of cheques of the type claimed in any of claims 1 to 6, the computer system being configured to receive digital images of the cheques and to apply to them a method of authentication comprising:

digitally imaging the cheque to obtain an image;

machine reading the visible markings from the image to obtain the cheque data;

selecting the key generation data from the cheque data;

determining the key from the key generation data;

5 reading the computer readable marking from the image and decrypting its content by use of the key to obtain the validation data;

comparing the validation data with the cheque data machine read from the visible markings to detect mismatches between them.