WO2003063098A1

WO2003063098A1 - Counterfeit detector for bank notes

Info

Publication number: WO2003063098A1
Application number: PCT/GB2002/003398
Authority: WO
Inventors: Joseph Vijay Kumar
Original assignee: Eurosystems Limited
Priority date: 2002-01-26
Filing date: 2002-07-25
Publication date: 2003-07-31
Also published as: NO20034310D0

Abstract

A device for checking the validity of a bank note or other valuable instrument having a driver for moving the note through a sensor unit. The sensor unit (2) has a plurality of infrared (IR) sensors (12) arranged in a row, a plurality of ultraviolet (UV) sensors (13) arranged in a row, and a plurality of magnetic sensors (14) arranged in a row. These are all mounted on one surface of the sensor unit. In addition, there is a control unit (3) having a plurality of IR sources (22) and a plurality of UV sources (23), all mounted on an opposing surface of the control unit. The plurality of IR sensors face the plurality of IR sources, the plurality of UV sensors face the plurality of UV sources. As the bank note is driven through the device between the control unit and the sensor unit, the resulting output signals from the sensors are used to determine the validity of the bank note.

Description

Counterfeit Detector

The present invention relates to a device for detecting counterfeit banknotes and method therefor.

The history of the problem of counterfeit currency is as old as currency itself. The importance of identifying and eliminating counterfeit banknotes does not need any special emphasis as the presence of large volumes of them can topple the economy of a country. Present day technologies involved in the production of counterfeit banknotes make it almost impossible to identify them with the naked eye or by any other physical method. Hence electronic equipment capable of reliably identifying them is needed more than ever before.

To address the above problem, banknotes have been designed which incorporate various security features or markings, such as magnetic threads, numbering, holograms, visible infrared markings, invisible infrared markings and ultraviolet markings and so on. The incorporation of these security features makes the production of counterfeit currency difficult and uneconomical. In addition, these features provide uniqueness to a currency by implicitly giving its country of origin.

Countries have chosen various zones on the banknotes of their currency to create these markings and any banknotes that deviate from these markings can be considered as counterfeit. These markings may not be identifiable by the naked eye but may be detected by various sensors that are sensitive in different energy ranges. The markings that can be seen on a banknote when a sensor, sensitive for a particular type of energy is employed, is known as its image for that type of energy and this information when stored in a computer, is known as its digital image for that energy.

Thus for electronic identification purposes, a banknote can be characterized by its physical size, IR image, UV image and magnetic image. These characteristic signatures in IR, UN and magnetic energies are given at various predefined physical locations on the note area.

Although vario.ύs devices are known for assisting in the determination of whether a banknote is genuine, these tend to be rather limited in their effectiveness. For example, UV lights that allow a user to see fluorescent features on notes are used in shops, etc. Also, there are vending machines and machines that dispense railway tickets or the like which accept bank notes as well as coins. These contain simple devices for checking that a bank note of the correct type is being used. However, such machines often do little more than check the magnetic parts of a note, and/or its size.

According to one aspect of the invention there is provided a device for checking the validity of a banknote or other valuable instrument comprising: a driver; a sensor unit having a plurality of infrared (IR) sensors arranged in a row, a plurality of ultraviolet (UV) sensors arranged in a row, and a plurality of magnetic sensors arranged in a row, all mounted on one surface of the sensor unit; and a control unit having a plurality of IR sources and a plurality of UV sources, all mounted on an opposing surface of the control unit, whereby, the plurality of IR sensors face the plurality of IR sources, the plurality of UV sensors face the plurality of UN sources, the driver is arranged to drive the banknote through the device between the control unit and the sensor unit, and the resulting output signals from the sensors are used to determine the validity of the banknote.

The relative number of sensors may be adjusted as desired, however it is preferred that the number of IR sensors is greater than either the number of UV sensors or the number of magnetic sensors. This is because IR features tend to be more precisely defined. For example, there may be 10 IR sensors, 3 UV sensors and 6 magnetic sensors.

Preferably the control unit contains a controller which, in use, controls the IR, UV and magnetic sensor and source pairs, and the driver, wherein the controller further comprises a memory and a computer processor. Although any suitable design may be used, the output signals from the sensors are preferably fed to the controller via an amplifier, multiplexer and analogue to digital converter.

In a preferred implementation, a first optical source and associated optical sensor are mounted adjacent to the front of the device and function as an initiator for the driver. A second optical source and an associated second optical sensor may be employed to function as a measurement initiator. In addition, a third optical source and an associated third optical sensor may be used to measure the length of the banknote .

Where the device is configured as a free-standing unit, for use is shops, banks, etc, it preferably further comprises an indictor for indicating that a banknote is not genuine. This may take the form of a display readout, and audible alarm, etc. The invention may be applied in other circumstances where large amounts of currency is sorted in which case it may be more useful for suspect notes to be diverted into a store without an indicator being triggered.

The invention also extends to a corresponding method and so viewed from another aspect the invention provides a method for checking the validity of a banknote or other valuable instrument, by reading the IR, UV and magnetic data, comprising the step of providing: a plurality of IR sensors arranged in a row, a plurality of UV sensors arranged in a row, and a plurality of magnetic sensors arranged in a row, all sensors being mounted on one surface of a sensor unit; and a plurality of IR sources and a plurality of UV sources, all mounted on an opposing surface of a control unit, wherein, the plurality of IR sensors face the plurality of IR sources and the plurality of UV sensors face the plurality of UV sources, and wherein said banknote is moved between one said surface of the sensor unit and said opposing surface of the control unit in a direction perpendicular to the rows of said sensors, by means of a driver, whereby the resulting output signals from the sensors and the measured length of the banknote are used to determine its validity.

The method preferably employs the previously described apparatus.

In order to increase accuracy, the IR and UV data readings are preferably taken whilst the banknote is stationary. However, the magnetic data readings are most easily taken whilst the banknote is being driven by the driver. Preferably the result is a grid or map of readings .

In addition, data related to the length of the banknote may be fed to the memory of a controller, together with the output signals from the IR, UV and magnetic sensors. In this way, said output signals may be fed to the controller via an amplifier, multiplexer and analogue to digital converter.

Although various means may be used to determine whether the measured data corresponds to a genuine note, preferably a processor of the controller compares the IR, UV and magnetic data readings with reference data templates or maps stored in the controller.

This arrangement is in itself believed to be inventive and so viewed from an further aspect, the invention provides a method of checking the validity of a banknote or other valuable instrument comprising the step of generating an IR and/or UV and/or magnetic map of the banknote; and comparing the map(s) with pre-determined maps corresponding to genuine banknotes .

The invention also extends to a corresponding apparatus and so, from another aspect, the invention provides an apparatus for checking the validity of a banknote or other valuable instrument comprising a map generator for generating an IR and/or UV and/or magnetic map of the banknote; and a comparator for comparing the map(s) with pre-determined maps corresponding to genuine banknotes. The apparatus preferably operates in accordance with the preferred methods described herein.

Preferably the map(s) comprise a plurality of cells and the method comprises comparing values in each cell of the measured map with the corresponding cell of the predetermined map. The comparison step may be designed to results in an error factor and the error factors of all cells may then be used to determine if the banknote is genuine. Preferably, the error factor for a cell is deemed to be zero if the difference between the measured and pre-determined values for each cell is less than the pre-set amount. The apparatus may determine an error value per cell per type of reading by calculating the product of each error factor value and a preset weightage. The weightage may be stored on the controller means. Preferably it is dependent on the type of reading and the location of the cell on the banknote. It will be appreciated that different areas of banknotes tend to become worn at different rates. In this way account may be taken of this. For example, the edges of notes quickly become damaged which fray result in an "incorrect" reading. On the other hand, an inboard region of the note is much less likely to be damaged and so an "incorrect" reading here is more likely to indicate a forged note.

Preferably, the determination of a total error value for each type of reading is calculated on the basis of the sum of the individual error value readings calculated for each cell, whereby if the total error values for the IR, UN and magnetic readings are all less than a preset tolerance, the banknote is deemed to be genuine.

The inventors have found that in many genuine notes, the security features are shifted from their theoretical position due to tolerances in the printing and cutting process during manufacture . In order to accommodate this, the individual security feature or features corresponding to each type of reading may be allowed to shift in the longitudinal direction, within a preset positional tolerance, with respect to the reference data template under consideration. Preferably, the individual security features corresponding to a particular type of reading can shift independently of the remaining security features corresponding to that particular type of reading.

As previously mentioned, on detecting a suspect note, an indicator may be activated. Additionally or alternatively, if the note is genuine, the driver may eject the note from the sensor and control units in the feed direction, and if not genuine, it may ejects it in the reverse direction.

Where the system uses data measured from genuine notes to compare with the measured values (rather than purely theoretical data) the reference data templates are preferably derived from tests on new, worn out, soiled and used banknotes. In this way, due account may be taken of the actual distribution of notes that will be tested and old/soiled notes will not be falsely rejected. Furthermore, the weightages may be modified to take account of the data derived from the tests on worn out, soiled and used banknotes. In other words, areas of genuine notes that are found to have a wide variation may be given a lower weightage .

In order to accommodate changes in currency design the reference data and weightages may preferably be updated. This may most conveniently be done via a link to an external source of data, such as an internet link or a dedicated remote server.

As an additional security measure, a check that the distance between a watermark and a magnetic thread, determined from the sensor readings, is a preset distance, may be made before the banknote is considered genuine.

A still further step is for the rates of wear of different security features, determined from the sensor readings, to be compared to check expected rates of wear between said readings fall within expected levels before the banknote is considered genuine. With regard to UV readings, these may (as well as being compared to a map) be checked to determine the type of paper of the banknote, and only passes the banknote as genuine if the readings fall within expected levels. Thus, viewed from another aspect, the invention provides a method for checking the validity of a banknote or other valuable instrument, wherein the UV readings of the banknote are checked to determine the opacity of the banknote, and the banknote is only passed as genuine if the readings fall within expected levels.

Viewed from another aspect, the invention provides a device for checking the validity of a banknote or other valuable instrument, which comprises at least one ultraviolet sensor, at least one infrared sensor, at least one magnetic sensor and at least one optical sensor, which in combination are able to detect the following security features on the banknote: holograms, magnetic threads, UV markings, IR markings, invisible IR markings, watermarks, magnetic numbering, its opacity and its dimensions.

Preferred embodiments of the invention will now be described by way of example only, and with reference to the accompanying drawings, in which:

Figure 1 is a front perspective view of the counterfeit currency de ector;

Figure 2 is a rear perspective view of the currency detector device of Figure 1;

Figure 3 shows a front perspective view of the sensor unit and control unit of the detector device in the open position;

Figure 4 shows an exploded perspective view of the control unit of the detector device;

Figure 5 shows a perspective view of the driver mechanism of the detector device;

Figure 6 shows an exploded perspective view of the sensor unit;

Figure 7 shows the positions of the sources and sensors on the source and sensor cards;

Figure 8 is a block diagram of the principle components of the detector device;

Figure 9 shows the basic steps involved in the verification process;

Figure 10 shows the grid of measurement cells on a typical banknote; and

Figure 11 shows a typical IR reference grid on a banknote .

Figure 1 shows an embodiment of the counterfeit currency detector 1 from a front perspective view and Figure 2 shows the detector 1 from a rear perspective view. The detector comprises two main parts, a sensor unit 2 and a control unit 3 which are connected together at their rear ends by hinges . As can be seen in Figure 1 , the control unit has a front panel with an LCD display 31 and LEDs 32. The control unit has two raised side portions 35 which run along the two longitudinal edges of the control unit 3, having inner sides 34 defining a central longitudinal recess 36. The sensor unit 2 is located in recess 36, and projects therefrom in use. As can be seen from Figure 2, a RS232 port 5 is provided so that the detector unit 1 may be connected to a PC. In addition, a power port 6 and on/off switch 4 are shown.

Figure 3 shows the sensor unit 2 and the control unit 3 in the open position. In use, the sensor unit 2 and control unit 3 are closed and thus the ambient light interference on the measurements is reduced drastically. The detector is fitted with a sensor switch, which alerts the micro-controller if these two units are not closed properly. The arrangement of sources and sensors in the detector can be clearly seen in Figure 3. A first optical jfeensor 11 is located at one side of the base of the sensor unit 2 adjacent to the front of the detector, which in use, faces a corresponding first optical source 21 on the control unit 3. A second optical sensor 44 is located between a row of IR sensors 12 and a first set of rollers 7. A corresponding second optical source 45 on the control unit faces the second optical sensor 44, in use. A third optical sensor 15 is also located at one side of the base of the sensor unit 2 adjacent to the rear of the detector, which also faces in use, a corresponding third optical source 25 on the control unit 3. This third optical sensor 15 detects the presence of the note as it moves towards the exit of the detector 1 and allows the length of the note to be measured. In addition, these sensors serve to warn of errors, such as jamming or misfeeding of a note in the detector. Optionally, a plurality of sensors may be placed adjacent to one side of the recess 36 of the control unit 3 running along a lateral direction (perpendicular to the direction of note feed) , if the width of the note is also required to be measured.

Behind the first optical sensor 11, is a row of IR sensors 12. In Figure 3, ten sensors are shown on the base of the sensor unit 2, which in use face a corresponding number of IR sources 22 on the top face of the control unit 3. The IR sensors 12 check for the presence of visible and invisible IR inks and can detect security features such as watermarks. A row of UV sensors 13 are located behind this row. Three UV sensors 13 are shown, which in use face a corresponding number of UV sources 23 located on the top face of the control unit 3. The UV sensors check for the presence of fluorescent and phosphorescent inks. Behind the row of UV sensors, are located a plurality of magnetic sensors 14, six of which are shown. These sensors check for the presence of security features such as magnetic thread and numbering, and measure by contact whilst the note under test moves underneath them, unlike the IR and UV sensors which take measurements whilst the note is stationary.

In the embodiment shown, semiconductor UV sources are utilised rather than fluorescent UN sources, which have commonly been used in the past . The semiconductor sources provide controllable UV emission intensity which is not possible with fluorescent lamps. In addition, the UV sources are smaller than the detectors, hence the interference of one source on the next detector is negligible. Also the UV energy emitted from one source is negligibly small compared to a fluorescent lamp, thus the unwanted effects of stray UV light coming into contact with the human body are minimized. Finally, the power consumption of the system is less than a system utilizing fluorescent lamps.

A plurality of rollers 7 arranged in rows, protrude from the base of the sensor unit 2 and corresponding rollers 7 protrude from the top face of the control unit 3. These serve to guide and drive the note through the device in the forward or reverse direction.

Figure 4 shows the control unit in greater detail. This can be seen to comprise three optical source cards 8, corresponding to first (entry) , second and third (exit) optical sensors 11, 44 and 15, respectively, and an IR and UV source card 9. The driver mechanism 16 is shown and this will be described in more detail later. On the bottom cover 37 of the control unit 3 is mounted a digital card 10. An RS 232 port 5, power port 6 and on/off switch 4 are shown attached to the bottom cover 37.

Figure 5 shows the driver mechanism 16 employed in the detector device. This comprises a plurality of rubber rollers 7 driven by a motor 20. A U-shaped base plate 26 supports a plurality of spindles 17, three of which are shown in the figure, which are mounted laterally across the device on the upwardly protruding arms 27 of the base plate via bushes 18. A pair of rollers 7 are mounted on each of these spindles. Each roller 7 is spaced at approximately one third of the distance from the sides of the mechanism, along each spindle. Each spindle is connected to its adjacent spindle or spindles via a pulley and belt arrangement 28. The motor 20 is bi-directional and controlled by a driver panel 24. This drives the adjacent spindle 17 via gearing 19. Thus all rollers 7 are driven simultaneously.

Figure 6 shows the sensor unit 2 in greater detail . The top casing 29 is shown together with a sensor card 30 comprising a UN/IR card and a magnetic card. Below this the bottom casing 39 shows the holes 38 for the various sensors employed.

Figure 7 shows the relative positions of the sources and sensors on the source card 8,9 and sensor card 30, respectively. The direction of the movement of a banknote 40 under test is shown.

As is shown in Figure 8 , at the heart of the equipment is a micro-controller on the digital card 10, which is located inside the control unit 3, which controls all the activities of the detector such as: i) movement of the banknote by activating the motor; ii) energizing the sources on the source card, activating the sensors on the sensor card, and coordinating their timing; iii) storing the measurements; iv) comparing the test data obtained from the sensors with the reference data and determining whether the note is genuine or not; v) controlling and displaying the messages on the

Liquid Crystal Display (LCD) ; and vi) conducting a power-on self-test thus checking the condition of the system.

The RS 232 port is used to update the EEPROM with reference data for a series of banknotes. In addition, during the "assembly" process, the RS 232 port is also used to upload the software to the micro-controller, wherein the "software" is the main software used by the micro-controller to compare banknotes, display on the LCD, control the drive mechanism and control the sensors, etc.

The detector 1 can be connected to a PC in order to download new parameters of detection, i.e. reference data, onto the EEPROM chip. Downloading can be local (CD-ROM) or on-line (secured Internet link) .

As described above, the detector device comprises a bidirectional motor to facilitate the movement of the banknote in both the forward and reverse directions in a predetermined manner. In addition, a set of IR and UV sensors and magnetic sensors are arranged in rows. The equipment also has optical sensors for determining the physical size of the banknote. Thus the equipment can determine all four characteristics mentioned above i.e. size, UV, IR and magnetic images. When the note is moved by the motor, the note is moved between these sources and sensors .

When a banknote is passed through the chamber over the row of IR sources 22, the information obtained from the IR sensors 12 above the banknote can be stored in the memory of the controller. This information is the basic data in forming the IR map of the banknote of that particular physical area to which IR radiation is exposed. The infrared sources 22 arranged in a row are switched on sequentially. Sensors placed opposite each source each produce an electrical signal proportional to the transmitted light that reaches it. The signals are amplified then passed to a 16-bit ADC via a multiplexer, one after the other. The resulting digitized test data is stored in the RAM of the micro-controller. The micro-controller also controls the timing of the multiplexer and ADC. Gradually moving the entire banknote under the row of IR sensors 12 gives the basic data of the entire IR image of the banknote.

In a similar manner the UV sources 23 and sensors 13 arranged in the row adjacent to the row of IR sources and sensors, are amplified and multiplexed sequentially and the data obtained from the UV sensors 13 is digitized using the same ADC, and then stored in the RAM. Thus the UV image of the banknote is produced as the banknote travels under them. Fewer UV sensors are required as compared to IR sensors, since UV printing is often in coarse blocks and not in fine detail as is the case with IR images.

The third row consisting of the magnetic coils act as magnetic sensors 14. When a magnetic material moves under these coils, this movement causes an electrical signal to be induced in these coils. The amplitude and frequency of this electrical signal is proportional to the quantity of the magnetic material under the sensor, its magnetism and the speed of the movement. As for the IR and UV readings, the magnetic signals are amplified, passed to a multiplexer and converted to digital signals via an ADC, then stored in the RAM.

Thus the present equipment works on the principle of preparing the IR, UV and digital images of a banknote by measuring the values of UV and IR light that get transmitted through the note under test . A predetermined intensity of these energies is transmitted from one side of the note and sensors are placed on the other side and transmitted energies are measured. Thus these measurements define an image for that energy for the note under test. Thus, the device is able to detect all security features on a banknote, i.e. holograms, magnetic threads, UV and IR markings, invisible IR markings, watermarks, magnetic numbering, opacity and dimensions. It is then able to compare these readings with reference data which is stored for a range of different banknotes. Furthermore, it is a portable desktop device which is fully automatic in its checking procedure and does not require any human intervention.

Figure 9 shows the basic steps of the above process. The banknote is manually inserted into the front of the device. As the banknote reaches the first optical sensor 11, the motor is activated which in turn causes the rollers 7 to rotate within the device, and the note is pulled into the machine by the first set of rollers. When the note reaches the second optical sensor 44, a delay is introduced, whereby the motor is stopped then the rollers driven in the reverse direction so that the note is driven for a short time in the backwards direction (with respect to the feed direction) . The motor is then stopped and the rollers rotated again in the forward direction, so that the note moves in the feed direction once again. When the note reaches the second optical sensor 44 for the second time, the source and sensor pairs are activated in the appropriate order. This sequence of events allows accurate positioning of the note in the feed direction, before readings commence. The IR source 22 and sensor 12 pairs are activated sequentially, and the resulting signals from the sensors are allowed sufficient time to stabilize before being fed via an amplifier, into a multiplexer and then converted from analogue to digital signals within the ADC comprising a chip having 256 discrete levels. Once the ADC conversion process is complete the digital test data is stored in the RAM. The UV source 23 and UV sensor 13 pairs and magnetic sensors 14 are similarly activated and resulting signals amplified and fed via a multiplexer into the ADC. The note is fed through the machine in steps and this process is continued with the IR and UV values being taken whilst the note is stationary and magnetic readings being taken whilst the note is moving under the magnetic sensors. Once the end of the note has been detected by the third optical sensor 15, the test data is complete and the motor is stopped. In addition, the third optical sensor 15 enables the length of the banknote to be determined. Hence a grid of readings of UV/IR/Magnetic values are obtained, i.e. the note can be considered to consist of an array of measurement cells. The approximate time between introducing the banknote into the device and the banknote being ejected at the opposite end of the device is two seconds. This includes the 0.5 second delay at the point when the banknote reaches the second optical sensor. This delay may later be adjusted, in which case, the scanning time may increase/decrease further. Reference data for various denominations of a currency is stored in EEPROM memory. This data is initially obtained by conducting a series of tests on genuine banknotes. The derivation of the reference data will be described later. The test data is compared with the reference data and if the result is less than a predetermined tolerance, the note can be considered genuine and an LED and/or message is displayed on the front of the device, for example giving the value of the banknote. The motor is activated to drive the note all the way through the machine i.e. in the feed direction. If the result of the comparison process is that it is greater than a predetermined tolerance, then the note is rejected as a suspected counterfeit, and an LED and/or an error message is displayed on the device. The motor is driven in the reverse direction so that the note is fed back out of the front of the machine .

In general terms, the digital image data for each type of security feature reading i.e. UN/IR/magnetic , is compared on a cell-by-cell basis with the reference data, corresponding to that security feature, stored in the equipment for example, in a EEPROM. A comparison is performed between the test data and reference data for a particular location (cell) on the note, for each type of reading (i.e. UV, IR or magnetic) . From this, an array of error factor values are obtained which are each multiplied in turn by a weightage. The weightage is dependent on the security feature being measured and the particular location (cell) on the note. This gives an error value per cell, for each type of reading. The values obtained are summed up to form the total error figure for each type of reading/security feature. The magnitude of this total error figure is a measure of its degree of counterfeit nature. Only if the total error figure for each type of reading is less than that of a pre-determined tolerance value, is the note considered to be genuine. Assigning a specific weightage to each cell is important in reducing spurious errors, since the weightages allow for dust accumulation on various locations on the banknote and soiled banknotes. For example, after a note has been in circulation for some time, the note may have been folded, thus the incorporation of a weightage into the error factor when determining the overall error value of the note, allows normal wear and tear to be taken into consideration. Details of the error analysis and derivation of weightages wil,l be given later.

Derivation of the reference data:

As shown in Figure 10, the banknote area is divided up into an array of cells for measurement purposes. In the example shown, the banknote has dimensions of 74 mm x 120 mm. The edges of the note are shown by short dashed lines 43. The inner edge 34, of the longitudinal ridge of the control unit 3 is shown to the right of the

Figure. The note 40 is fed into the device so that it abuts against this ridge 34. The sensor (or source) locations are shown as small crosses 41 and these define the centre of each cell in the lateral direction of the note. In the example shown, the centre-to-centre separation of the sensors (or sources) is 7 mm and the rightmost sensor is located 7 mm from the ridge 34. Thus as can be seen, a margin, a (= 3.5 mm) , exists at the right side of the note, and a margin b (= 0.5 mm) , at the left side of the note. This leaves a readable area of c x L (where c = W - a - b) . In this example, c = 70 mm, hence there are 10 cells in the lateral direction. Since the length of the note, L = 120 mm and the note is stepped through the device in 4 mm intervals, the number of cells in the longitudinal direction is 30. Number of cells of Grid = No. of sensors across the width (fixed) x Length of the note

4

Hence the total number of cells for which IR readings are obtained for a note of the above dimensions would be 300, with each cell having an area of 7 mm x 4 mm.

For each cell, IR data, UN data and Magnetic data is measured and stored in RAM. Thus for a note of the above dimensions 300 memory locations for IR data, 90 memory locations for UV data and 180 memory locations for Magnetic Data are required.

In the case of the IR data, no portion of the banknote is disregarded for measurement. However, in the case of the UV and magnetic data, only the areas that come in the region of the respective sensors get measured.

The number of cells along the width is dependent on the width of the banknote (i.e. the number of sensors that come within the width) . The number of cells in the longitudinal direction is variable and it is dependent on the length of the banknote. For example, a banknote having a length of 96 mm will have 24 cells in the longitudinal direction.

The maximum width of banknote that the device can handle is 84 mm and this is shown by the long dashed line 42. There is no limit to the maximum length of the note. However, the software does detect that if the banknote is not out within a fixed period, for example, five seconds, that it has jammed within the device and an appropriate error message is displayed. The smallest note allowed is one having a length of 65 mm (as per the spacing between two rollers which move the note in the chamber) . There is no limitation for the width of smallest note.

Creation of REFERENCE DATA for a banknote of type X is carried out by the following steps:

(a) Several samples (for example 10) of genuine good quality (i.e. new) banknotes of type X are measured. Each banknote is passed through a device similar to the detector that has been modified (by software) to act as a scanner. The IR, UV and magnetic values are obtained for each cell on each sample note and the average of these values {DATA AVERAGE (new) ) is determined for each cell and each type of reading (IR/UV/magnetic) .

(b) The apparatus then computes the DATA AVERAGE (worn out) similarly using the procedure (a) above on a number of samples of worn out banknotes of type X.

(c) The apparatus then computes the DATA AVERAGE

(soiled) similarly using the procedure (a) above on a number of samples of soiled banknotes of type X.

(d) The apparatus then computes the DATE AVERAGE (used) similarly using the procedure (a) above on a number of samples of used banknotes of type X.

(e) The apparatus then computes the REFERENCE DATA as the average of the four values computed above (i.e. for each cell and each type of reading) .

REFERENCE DATA = [DATA AVERAGE (new) + DATA AVERAGE (worn out) + DATA AVERAGE (soiled) + DATA AVERAGE (used) ] ÷ 4

The size of the REFERENCE DATA look up table for the banknote of type X (if of dimensions 74 mm x 120 mm) consists of 570 locations in the memory (300 for IR ₊ 90 for UV + 180 for Magnetic) .

Figure 11 shows a typical IR reference grid superimposed on a banknote to which it corresponds.

Creation of REFERENCE WEIGHTAGE:

Different weightages are given for the cells of the banknote of type X, depending on the type of image (IR/UV/Magnetic) and the zone it is located. For example, Phosphorus ink markings (UV) wear off relatively rapidly, hence this security feature should be allotted a low weightage.

The initial weightage for each cell (REFERENCE WEIGHTAGE (temp. ) ) is dependent on the location of that cell. For example, a lower weightage is given to cells near the edges of the note and locations along the centrelines of the notes where folds are likely. In addition, for IR/UV/magnetic images each covering several cells, the periphery of those images may only cover part of a cell . Therefore a lower weightage can be allotted to the partially covered cell as compared to a cell which is has its area completely filled with the image .

Thus the REFERENCE WEIGHTAGE (temp . ) for IR/UV/magnetic data for each cell of the banknote of type X is determined.

The weightages in REFERENCE WEIGHTAGE (temp . ) are then modified taking into consideration the data obtained from old (worn out or used) or soiled banknotes of type X. This ensures that the low intensity images appearing on old /soiled genuine banknotes are also considered and not automatically rejected by the detector .

The procedure for modification of Weightages is as follows :

For each cell and each type of reading, the minimum of the three values corresponding to DATA AVERAGE (worn ou t) , DATA AVERAGE (used) and DATA AVERAGE (soiled) , calculated when obtaining the reference data, is taken as DATA AVERAGE (other) and this value is compared with the REFERENCE DATA. If the difference is less than the square root of the REFERENCE DATA, there is no need to modify the REFERENCE WEIGHTAGE of that cell. Otherwise, a deviation factor (D. F. ) is calculated as per the following formula:

REFERENCE DATA - DATA AVERAGE ( Other)

D. F.

J REFERENCE DATA

If D. F. = 1 or more, the value of REFERENCE WEIGHTAGE (temp. ) is reduced accordingly, and thus the REFERENCE WEIGHTAGE look up table is created for the banknote of type X.

The size of the REFERENCE WEIGHTAGE look up table will be same as that of the REFERENCE DATA look up table.

Determination of Error:

Each banknote is checked using a statistical error computation procedure. When the banknote under test is moved through the equipment, the IR, UV and magnetic data of the banknote is read, i.e. the TEST DATA. The data thus obtained forms the basic data for computation of the image for that particular energy. The data obtained from the ADC and stored in the RAM ( TEST DATA) is compared to the REFERENCE DATA stored in a EEPROM and the difference is again stored in the RAM. If the difference is less than the square root of the REFERENCE DATA, then the data can be considered to match the reference data, otherwise this difference contributes to the error value and an Error Factor (E. F. ) is computed as the integer value (rounded down) given by the following formula:

REFERENCE DATA - TEST DATA

E. F.

J REFERENCE DATA

For example:

Case A: At a physical location/cell, if the value for IR REFERENCE DATA is 196 and the measured value of IR ( TEST DATA) is 187.

σ = J REFERENCE DATA = 14

196 - 187

E. F. = 0.64= 0

14 14

Case B: REFERENCE DATA is 196, as above and TEST DATA is 134.

196 - 134 62

E. F. = 4.43 = 4 14 14 Thus if the deviation is within one sigma, the Error Factor is 0. A deviation of more than σ and less than 2σ gives an Error Factor of 1. Deviation by more than 4σ and less than 5σ, gives an Error Factor of 4.

If the reference value is 255 and the measured value is 0, this would constitute a maximum Error Factor of 15 as the deviation is more than 15σ, since:

255 - 0

E. F. 15.97 = 15

J255

The actual error at any cell/location (for each type of reading) is the product of the Error Factor (E. F. ) for that cell and REFERENCE WEIGHTAGE for that cell.

Using the above steps, the error in the IR readings is computed for each and every location/cell of the banknote area and the sum of these values becomes the TOTAL ERROR VALUE (IR) for the Infrared Image. Similarly a TOTAL ERROR VALUE is computed for the UV and also for the magnetic images.

Depending on these three error figures and the data obtained from the errors in the size of the banknote, it can be determined whether the banknote is a possible counterfeit or whether it is genuine, i.e. if the TOTAL ERROR VALUE for each type of reading calculated above is less than a pre-set TOLERANCE LIMIT for each type of reading, then the banknote may be considered genuine. A banknote passed through the device and rejected by the device (under normal conditions) means that either i) the banknote is counterfeit or ii) the banknote is genuine, but that its reference data is not available in the device. Thus in use, the banknote is classified on the basis of comparison with a series of templates. The templates are stored in the memory and each contain the reference data (for each type of data) , as well as the length data for each type of banknote.

After the banknote has passed through the chamber of the equipment, the test data length parameter is compared with reference data length parameter for each of the templates described above. Where a match is found, only those template data sets will be considered for further comparison of the IR, UN and magnetic data.

In general terms, IR test data is compared with the first matched template IR (reference) data. If the IR test data is found to be within a pre-set tolerance, then the UV test data is compared, if the UV test data is also within a pre-set tolerance, comparison is then performed on the magnetic test data. If at any stage there is no match (beyond the pre-set tolerance levels) , then the comparison process shifts to the second matched template and so on.

The first part of the banknote paper manufacturing process involves the incorporation of the magnetic thread and the watermark into the cotton banknote paper. This is then cut/manufactured into large sheets. Multiple banknotes are printed onto these large sheets i.e. the remaining security features (numbering, hologram (s) , and remaining IR and UV markings) are printed onto the paper. These large sheets are then cut to the eventual banknote size. Due to the nature of the paper and the cutting process involved, the watermark and magnetic thread can become misaligned (in the longitudinal direction) , with the cut banknote. As a further step, a shift of + 4 mm in the longitudinal direction of the watermark and magnetic thread is allowed for. In general terms, the system checks the measured test data in order to locate the position of these two security features, by comparing the test data with the reference data, i.e. it checks firstly for a match in value of the security feature (e.g. the watermark), and secondly for a match in the shape of the security feature. If the above two matches are made, and the shift from the standard position is within the above tolerance, the reference values "corresponding" to those cells (i.e. corresponding to the watermark) are allowed to shift accordingly by the same amount, for the purposes of the error calculation in respect of those cells. This process is then repeated for the magnetic thread readings .

In practice all features detectable by either IR/UV/magnetic sensors are allowed to shift within the above tolerance, and there are two steps involved in reading the IR/UV/magnetic data. The first step is the reading of the actual data, the second is the determination of the "edges" of the security feature being measured, from which the "length" of that feature can be ascertained. These "lengths" are allowed to shift by 4 mm in each direction (i.e. one pixel/cell in each direction) , however, the "lengths" must be exact. The steps involved are as follows:

i) Length of note is measured and only templates of that length are considered;

ii) IR test values are considered. The values are compared with the IR reference data of a first IR template and error analysis is performed, if result is less than an acceptable tolerance then step vi can be performed, if result is greater than an acceptable tolerance then it is possible that shifting of the or each IR security feature has occurred and step iii is carried out;

iii) The values of a particular (e.g. first) IR security feature are considered (as there may be more than one IR feature and shifting can occur independently) . In other words, the "length" of the particular (e.g. first) IR feature is determined and this "length" is shifted by 4 mm (one pixel) to the right in the longitudinal direction. A comparison is once again performed between the shifted IR values and the IR reference data for the first template (for this particular (e.g. first) security feature only). Thus localised comparison rather than complete comparison with the reference data is performed as other security features may not have shifted. The error analysis is performed as before for this localised region. If the result is greater than an acceptable tolerance then step iv is carried out. If the result is within an acceptable tolerance, then another IR security feature (i.e. a second security feature) can be considered and step v carried out;

iv) The values of the first IR security feature are considered once more and allowed to shift by 4 mm to the left in the longitudinal direction and a localised comparison performed as in step iii. If the result is greater than an acceptable tolerance then as a final check, no shift is allowed for (since separate IR security features can shift independently and this particular one may not have shifted at all) . If the result of both these steps is that the values are still above an acceptable tolerance, then comparison should be performed with the second IR reference template and step ii repeated; v) If the result of steps iii - iv is that the particular (e.g. first) IR values are within an acceptable tolerance, the second IR security feature is considered and step iii is performed;

vi) If all the values of the separate IR security features are found to be within an acceptable tolerance, the values of the hologram feature (s) and "length" of the or each hologram security feature is determined and compared with the reference data of the particular "matching" IR template in the manner described with reference to the IR security features above. As before, if all comparisons including that with the shifted data yield values that are greater than an acceptable tolerance, then comparison shifts to the next template and step ii is instigated;

vii) The UV values are then compared with the particular UV template of reference values "in use" and as in steps ii to vi above, the edges of the UV security features are determined and allowed to shift by one pixel in either direction;

viii)The magnetic values and magnetic edges are compared in a similar manner as the IR and UV values .

Since the watermark and magnetic thread are applied simultaneously, the distance between these two security features is constant, irrespective of any shift due to the cutting process. Thus as a further embodiment, which can be implemented in addition to the above, incorporating a check on this distance during the testing process adds further accuracy to the verification process.

A further embodiment, which may be used in conjunction with those listed above, incorporates a comparison between the values of security features measured during the testing process. Since the rates of wear of particular security features should tally, for example, IR images tend to be the most hardwearing, but, magnetic numbering and phosphorous inks tend to wear off relatively rapidly. Thus any discrepancies between the intensity of images which should ordinarily wear at the same rate, can also be checked during the verification process.

A further embodiment, which may be used in conjunction with any of those listed above, utilises the readings given by the UV sensors to determine the type of paper used to produce the banknote . UV measurements performed on banknotes printed on cotton paper give readings of for example, between 0.5 V to 2.8 V (this includes reading from UV and non-UV image areas) . However, banknotes printed on ordinary white paper give UV sensor readings of for example, above 2.8 V. Incorporating threshold levels in this manner adds a further security check in the validation process.

According to a further embodiment, which may be used in conjunction with any of those listed above, the width of the note can also be estimated by the row of IR sensors 12, if they are programmed to do so.

According to a further embodiment, which may be used in conjunction with any of those listed above, the RS232 port allows a user to maintain records of notes tested, by sending the results to a PC.

According to a further embodiment , which may be used in conjunction with any of those listed above, the micro-controller can control an audio alarm according to the result of the note tested. According to a further embodiment , rather than the IR sources 22 being switched on sequentially, they are switched on synchronously.

Claims

1. A device for checking the validity of a banknote or other valuable instrument comprising: a driver; a sensor unit having a plurality of infrared (IR) sensors arranged in a row, a plurality of ultraviolet (UV) sensors arranged in a row, and a plurality of magnetic sensors arranged in a row, all mounted on one surface of the sensor unit; and a control unit having a plurality of IR sources and a plurality of UV sources, all mounted on an opposing surface of the control unit, whereby, the plurality of IR sensors face the plurality of IR sources, the plurality of UV sensors face the plurality of UV sources, the driver is arranged to drive the banknote through the device between the control unit and the sensor unit, and the resulting output signals from the sensors are used to determine the validity of the banknote.

2. The device as claimed in claim 1 wherein the number of IR sensors is greater than either the number of UV sensors or the number of magnetic sensors.

3. The device as claimed in claim 1 wherein there are 10 IR sensors, 3 UN sensors and 6 magnetic sensors .

4. The device as claimed in any preceding claim wherein the control unit contains a controller which, in use, controls the IR, UV and magnetic sensor and source pairs, and the driver, wherein the controller further comprises a memory and a computer processor.

5. The device as claimed in claim 4 wherein the output signals from the sensors are fed to the controller via an amplifier, multiplexer and analogue to digital converter.

6. The device as claimed in any preceding claim wherein a first optical source and associated optical sensor are mounted adjacent to the front of the device and function as an initiator for the driver.

7. The device as claimed in claim 4 wherein a second optical source and an associated second optical sensor function as a measurement initiator.

8. The device as claimed in any preceding claim wherein a third optical source and an associated third optical sensor are used to measure the length of the banknote .

9. The device as claimed in any preceding claim wherein it further comprises an indictor for indicating that a banknote is not genuine.

10. A method for checking the validity of a banknote or other valuable instrument, by reading the IR, UV and magnetic data, comprising the step of providing: a plurality of IR sensors arranged in a row, a plurality of UV sensors arranged in a row, and a plurality of magnetic sensors arranged in a row, all sensors being mounted on one surface of a sensor unit; and a plurality of IR sources and a plurality of UV sources, all mounted on an opposing surface of a control unit, wherein, the plurality of IR sensors face the plurality of

IR sources and the plurality of UV sensors face the plurality of UV sources, and wherein said banknote is moved between one said surface of the sensor unit and said opposing surface of the control unit in a direction perpendicular to the rows of said sensors, by means of a driver, whereby the resulting output signals from the sensors and the measured length of the banknote are used to determine its validity.

11. A method as claimed in claim 10, comprising the use of the apparatus of claims 1 to 10.

12. The method as claimed in claim 10 or 11, wherein IR and UV data readings are taken whilst the banknote is stationary, and magnetic data readings taken whilst the banknote is being driven by the driver, so that a grid of readings is obtained in each case.

13. The method as claimed in claim 12 wherein data related to the length of the banknote is fed to the memory of a controller, together with the output signals from the IR, UV and magnetic sensors, and wherein said output signals are fed to the controller via an amplifier, multiplexer and analogue to digital converter.

14. The method as claimed in claim 13 wherein a processor of the controller compares the IR, UV and magnetic data readings with reference data templates stored in the controller

15. The method as claimed in claim 14 wherein the processor of the controller calculates an error factor value for each type of reading, on a cell-by-cell basis .

16. The method as claimed in claim 15 wherein the processor determines an error value per cell per type of reading by calculating the product of each error factor value and a preset weightage.

17. The method as claimed in claim 16, wherein the weightage is stored on the controller means, and is dependent on the type of reading and the location of the cell on the banknote.

18. The method as claimed in claim 17, wherein the determination of a total error value for each type of reading is calculated on the basis of the sum of the individual error value readings calculated for each cell, whereby if the total error values for the IR, UV and magnetic readings are all less than a preset tolerance, the banknote is deemed to be genuine.

19. The method as claimed in claim 18 wherein the individual security feature or features corresponding to each type of reading are allowed to shift in the longitudinal direction, within a preset positional tolerance, with respect to the reference data template under consideration.

20. The method as claimed in claim 19, wherein the individual security features corresponding to a particular type of reading can shift independently of the remaining security features corresponding to that particular type of reading.

21. The method as claimed in any of claims 10 to 20 wherein if the note is genuine, the driver ejects the note from the sensor and control units in the feed direction, and if not genuine, ejects it in the reverse direction.

22. The method as claimed in any of claims 10 to 21 wherein the reference data templates are derived from tests on new, worn out, soiled and used banknotes.

23. The method as claimed in claim 22 wherein the weightages are further modified to take account of the data derived from the tests on worn out, soiled and used banknotes .

24. The method as claimed in claims 16 to 23 wherein the reference data and weightages are updated via a link to an external source of data.

25. The method as claimed in any of claims 10 to 24 wherein a check that the distance between a watermark and a magnetic thread, determined from the sensor readings, is a preset distance, is made before the banknote is considered genuine.

26. The method as claimed in any of claims 10 to 25 wherein the rates of wear of different security features, determined from the sensor readings, are compared to check expected rates of wear between said readings fall within expected levels before the banknote is considered genuine.

27. The method as claimed in any of claims 10 to 26 wherein the UV readings are checked to determine the type of paper of the banknote, and only passes the banknote as genuine if the readings fall within expected levels.

28. The method as claimed in any of claims 10 to 27 wherein the sensor and source pairs are activated synchronously.

29. A device for checking the validity of a banknote or other valuable instrument, which comprises at least one ultraviolet sensor, at least one infrared sensor, at least one magnetic sensor and at least one optical sensor, which in combination are able to detect the following security features on the banknote: holograms, magnetic threads, UV markings, IR markings, invisible IR markings, watermarks, magnetic numbering, its opacity and its dimensions.

30. A method for checking the validity of a banknote or other valuable instrument, wherein the UN readings of the banknote are checked to determine the opacity of the banknote, and the banknote is only passed as genuine if the readings fall within expected levels.

31. A method of checking the validity of a banknote or other valuable instrument comprising the step of a) generating an IR and/or UV and/or magnetic map of the banknote; and b) comparing the map(s) with pre-determined maps corresponding to genuine banknotes .

32. A method as claimed in claim 31, wherein the map(s) comprise a plurality of cells and the method comprises comparing values in each cell of the measured map with the corresponding cell of the pre- determined map.

33. A method as claimed in claim 32, wherein the comparison step results in an error factor, the error factors of all cells being used to determine if the banknote is genuine.

34. A method as claimed in claims 33, wherein the error factor for a cell is deemed to be zero if the difference between the measured and pre-determined values for each cell is less than the pre-set amount.

35. A method as claimed in any of claims 32 to 34, wherein a weightage is applied to each cell.

36. A method as claimed in claim 35, wherein the error factor for each cell is multiplied by the weightage for that cell.

37. A method as claimed in claim 36, wherein the results of each such multiplication are summed and the total used to determine if the banknote is genuine.

38. An apparatus for checking the validity of a banknote or other valuable instrument comprising a) a map generator for generating an IR and/or UV and/or magnetic map of the banknote; and b) a comparator for comparing the map(s) with pre-determined maps corresponding to genuine banknotes .

39. An apparatus as claimed in claim 38 arranged to operate in accordance with the methods of any of claims 10 to 28 or 30 to 37.