WO2008075069A1

WO2008075069A1 - Counterfeit document detector

Info

Publication number: WO2008075069A1
Application number: PCT/GB2007/004919
Authority: WO
Inventors: John Alan Skinner; Ernest Alfred Munn
Original assignee: Talaris Holdings Limited
Priority date: 2006-12-21
Filing date: 2007-12-20
Publication date: 2008-06-26

Abstract

A detector assembly for use in sheet document processing is described. The assembly comprises: an emitter for irradiating sheet documents with a beam of radiation as they pass the detector assembly along a sheet path; at least one sensor for sensing resultant radiation received from the sheet document; and a window, disposed between each of the emitter and the at least one sensor and the sheet path, through which the beam and resultant radiation pass, the window being provided with a conductive layer which may be electrically connected to an electrical ground in order to dissipate any static charge that builds up in the window.

Description

COUNTERFEIT DOCUMENT DETECTOR

This invention relates to a detector assembly for use in sheet document processing (such as banknotes) and to methods for detecting counterfeit documents and for calibrating the detector assembly.

Sheet document processing apparatus often comprises a detector assembly for use in testing the authenticity of documents being processed. These detector assemblies typically irradiate the passing documents with ultraviolet radiation and detect the blue fluorescent and/or reflected ultraviolet radiation from the passing documents. By suitable signal processing, it is possible to detect counterfeit documents and then to sort these to a different destination from the genuine documents, for example.

The majority of these detectors are fairly bulky and cannot be fitted in smaller, desktop sheet document processing machines. There is a general need for a device that is suitable for this purpose, but it has been found that detectors that are made more compact are fairly sensitive to the distance between the sheet document and the detector as the sheet passes overhead. This can lead to false designation of a document, i.e. designating a genuine document as counterfeit, or even worse designating a counterfeit document as genuine.

Other problems with existing detector assemblies include the build up of static charge in sheet documents and in the assembly itself. This can interfere with the operation of the assembly or even cause damage to it.

In accordance with one aspect of the present invention, there is provided a detector assembly for use in sheet document processing comprising:

an emitter for irradiating sheet documents with a beam of radiation as they pass the detector assembly along a sheet path;

at least one sensor for sensing resultant radiation received from the sheet document; a window, disposed between each of the emitter and the at least one sensor and the sheet path, through which the beam and resultant radiation pass; and

an opaque barrier interposed between the emitter and the at least one sensor, the barrier and part of the window defining a gap through which at least part of the beam of radiation passes.

By providing a gap between the barrier and window, it has been found that it is possible to ensure that all the radiation impinges on the passing documents and that therefore the same amount of radiation will be detected from a given document, irrespective of its height above the window. In this way, the detector's immunity to the height of passage of the sheet document above the window has been improved.

Typically, the window is separated from the opaque barrier by the gap.

Preferably, the beam of radiation passes through the gap as it travels from the emitter to the window.

Normally, the emitter emits ultraviolet radiation.

The emitter may comprise a light emitting diode.

In one embodiment, the at least one sensor is sensitive to fluorescence of the sheet documents. In this embodiment, the detector assembly may further comprise a filter that passes only blue light, thereby adapting the at least one sensor to detect only fluorescence that is blue in colour.

In another embodiment, the at least one sensor is sensitive to radiation reflected from the sheet documents. In this embodiment, the detector assembly further comprises a filter that passes only ultraviolet light, thereby adapted the at least one sensor to detect reflected radiation that is of ultraviolet wavelength. In this embodiment, the detector assembly may further comprise a hemispherical lens for focussing the at least one sensor on the document. Typically, of course the at least one sensor is sensitive to both fluorescence of the sheet documents and to radiation reflected from the sheet documents and it may then comprise both types of filter mentioned above.

The at least one sensor may comprise a photodiode.

Typically, the window is made from glass. Preferably, the glass is non-fluorescent.

The window may be provided with a conductive layer which may be electrically connected to an electrical ground in order to dissipate any static charge that builds up in the window.

In one embodiment, the detector assembly further comprises a housing within which the emitter and at least one sensor are mounted.

The emitter and the at least one sensor may lie in a cavity within the housing over which the window is mounted. The housing may be mounted on a printed circuit board which lies beneath the cavity, thereby ensuring that radiation may only enter the cavity through the window.

The opaque barrier typically forms part of the housing.

Normally, the detector assembly comprises two sensors and the housing comprises a barrier interposed between the two sensors. The two sensors are typically responsive to reflected ultraviolet and fluorescent radiation from the sheet documents respectively. The fluorescent radiation is typically of a blue colour.

The housing may be made from polycarbonate.

In accordance with another aspect of the invention, there is provided a detector assembly for use in sheet document processing comprising:

an emitter for irradiating sheet documents with a beam of radiation as they pass the detector assembly along a sheet path; at least one sensor for sensing resultant radiation received from the sheet document; and

a window, disposed between each of the emitter and the at least one sensor and the sheet path, through which the beam and resultant radiation pass, the window being provided with a conductive layer which may be electrically connected to an electrical ground in order to dissipate any static charge that builds up in the window.

By providing a conductive path to ground through the window, any charge built up in the window or in passing documents that come into contact is discharged to ground.

The window is normally made from glass, which is preferably non-fluorescent.

In accordance with a third aspect of the present invention, there is provided a method of detecting a counterfeit sheet document comprising:

a) irradiating the sheet document;

b) sensing radiation reflected from a plurality of regions on the sheet document and generating corresponding reflected radiation signal values;

c) calculating the average of the reflected radiation signal values;

d) calculating the average value of the reflected radiation signal values that are equal to or exceed the average of the reflected radiation signal values;

e) comparing the average value calculated in step (d) with predetermined upper and lower thresholds; and

f) designating the document counterfeit if the average value calculated in step (d) does not fall between the upper and lower thresholds.

This aspect of the invention addresses a problem that is encountered when processing certain types of documents to look for counterfeits. For example, with US dollar bills, the UV reflectance level measured from the unprinted paper is significantly higher on genuine US dollar bills than on most counterfeits. However, genuine US dollar bills produce a lower UV reflectance level measured from the printed paper than from the unprinted paper. The UV reflectance level measured from the printed paper depends on the colour and density of the printed ink. There is also a reduction in the UV reflectance level measured from the unprinted paper on soiled genuine US dollar bills than from good condition notes. Furthermore, the number of samples measured from printed paper can vary with each note, depending on the face, orientation, and note pattern. Detecting a counterfeit note in these conditions has proven to be difficult.

Calculating the total average of the samples measured across the note does not provide sufficient separation between the UV reflectance levels measured from good condition counterfeits and soiled genuine notes, because an increase in the number of samples measured from printed paper on genuine notes lowers the average.

Using the peak reflectance levels obtained from the samples across the note does not provide sufficient separation between good condition counterfeits and soiled genuine notes, because the peak UV reflectance levels measured from good condition counterfeits can be greater than from poor condition genuine soiled notes, although most of the samples across the counterfeits are lower.

The most effective method we have found is to use the method of the third aspect to take an "upper average" of the samples across the note that are greater than or equal to the average. The "upper average" has a weighting towards the levels measured from unprinted and less densely printed paper on the genuine notes.

The average, peak, and "upper average" were calculated using sample data captured from genuine US dollar bills and counterfeit notes. The "upper average" gave the greatest separation between genuine and counterfeit notes. From these an absolute acceptance limit was determined.

The sheet document is typically irradiated with ultraviolet radiation.

The reflected radiation is also typically of ultraviolet wavelength. In one embodiment, the reflected radiation signal values are generated for each of the plurality of regions by:

i) sensing the reflected radiation with a photodiode;

ii) amplifying the signal from the photodiode;

iii) integrating the amplified signal; and

iv) converting the integrated signal to a digital representation of the reflected radiation.

In this embodiment, the radiation reflected from the plurality of regions may be detected by causing the sheet document to move relative to an emitter for irradiating the document and a sensor for receiving the reflected radiation, and periodically performing steps (iii) and (iv).

In accordance with a fourth aspect of the present invention, there is provided a method of calibrating a detector assembly according to either of the first or second aspects of the invention, the detector assembly further comprising a signal processing amplifier connected to the at least one sensor, the method comprising:

a) irradiating a reference document having known values of fluorescence;

b) receiving fluorescent radiation from the reference document at the at least one sensor; and

c) adjusting the gain of the signal processing amplifier such that the output is equal to a predetermined value or lies within a predetermined range of values corresponding to the fluorescence value of the reference document.

Also in accordance with the fourth aspect of the invention, there is provided a method of calibrating a detector assembly according to either of the first or second aspects of the invention, the detector assembly further comprising a signal processing amplifier connected to the at least one sensor, the method comprising: a) irradiating a reference document having known values of reflectance;

b) receiving reflected radiation from the reference document at the at least one sensor; and

c) adjusting the gain of the signal processing amplifier such that the output is equal to a predetermined value or lies within a predetermined range of values corresponding to the reflectance value of the reference document.

Calibration of the detector assembly is necessary to ensure that a desired nominal measurement of fluorescence or reflectance is achieved for a given type of document (for example, a measurement of 100 may be used as the nominal value for the white unprinted paper used for US one-dollar bills). The use of reference documents to calibrate the detector assembly rather than using mint condition genuine documents ensures an accurate and stable calibration because the reference documents are made from a stable UV reflective and/or fluorescent material and are measured and graded using a calibrated UV reflectance or fluorescence detector.

The reference document is typically irradiated by ultraviolet radiation.

In one embodiment, the gain of the signal processing amplifier is defined by the ratio of a first resistor to a second resistor, step (c) comprising setting the value of the second resistor depending on the reflectance or fluorescence value of the reference document, and then setting the value of the first resistor to adjust the gain of the signal processing amplifier such that the output is equal to a predetermined value or lies within a predetermined range of values corresponding to the reflectance value of the reference document.

In the case of receiving fluorescent radiation, the method may further comprise adjusting the drive current to an emitter for irradiating the reference document in step (c). Typically, the calibration using reflected radiation is carried out after the calibration using fluorescent radiation. After the calibration using fluorescent radiation has been done, the LED drive current is not changed and remains fixed otherwise it would be necessary to recalibrate using fluorescent radiation. One embodiment of this aspect of the invention further comprises subtracting a value representative of any extraneous reflected radiation or fluorescence that does not emanate from the reference document from the output value in step (c) before adjusting the gain.

An embodiment of the invention in the form of a detector assembly for installation within banknote processing equipment will now be described with reference to the accompanying drawings, in which:

Figure 1 shows a guide plate forming part of the banknote processing equipment, the detector assembly being fitted to the guide plate;

Figure 2 shows the guide plate and detector assembly from underneath;

Figure 3 shows an exploded view of the detector assembly;

Figure 3A shows the detector assembly;

Figure 4 shows a housing forming part of the detector assembly;

Figure 5 shows an insert which fits into the housing;

Figure 6 shows the circuitry mounted on a printed circuit board forming part of the detector assembly;

Figure 7 shows further signal processing circuitry;

Figure 8 is a flowchart illustrating the data capture process;

Figures 9 and 10 are flowcharts illustrating the data processing algorithms; and

Figure 11 is a flowchart showing the calibration process.

Figures 1 and 2 show a guide plate 1 which forms part of banknote processing (e.g. counting or sorting) equipment. The guide plate 1 has various apertures provided through which transport rollers can protrude to advance notes along the guide plate 1. When mounted in the banknote processing equipment, the guide plate 1 lies opposite another guide plate (not shown) between which banknotes are fed by the processing equipment. The two guide plates typically lie 4mm apart.

As can be seen in Figure 2, a pair of threaded studs 2 and 3 protrude from the guide plate 1. A detector assembly 6 is mounted to studs 2 and 3 by respective bolts 4 and 5. The detector assembly 6 provides two sensing regions 7 and 8 in Figure 1. The sensing regions 7 and 8 are glass windows integral to the detector assembly 6 and allow sensors within the detector assembly 6 to detect the passage of banknotes past the detector assembly 6 and various properties of the banknotes for the purposes of counterfeit detection.

The detector assembly 6 is shown in a more detailed, exploded view Figure 3, and in its assembled form in Figure 3A. The detector assembly 6 comprises a printed circuit board (PCB) 10 to which a housing 11 , typically made of black polycarbonate, is mounted. The housing 11 is opaque to visible, infrared and ultraviolet light. It is held fast to the PCB 10 by locating the housing 11 such that posts 12 (two of which are shown in Figure 3) pass through corresponding holes 13 in the PCB 10. The posts 12 are then melted so that the housing 11 cannot be separated from the PCB 10. The posts 12 thereby provide not only the fastening means but also a means of registering the housing 11 with respect to PCB 10. A bead of glue is also provided along the bottom edge of housing 11 to help hold the housing 11 fast to the PCB 10. This bead of glue also acts as a means for preventing any light entering the housing 11 between the housing 11 and the PCB 10.

Two lugs 14a and 14b are provided on the housing 11 through which the studs 2 and 3 may pass to fasten the detector assembly 6 to the banknote processing equipment.

The detector assembly 6 is connected to processing circuitry within the banknote processing equipment via a flying lead 9. The flying lead 9 comprises nine individual PVC sheathed 28AWG stranded wires, each of which is soldered to a respective pad in an array of solder pads 15 on the PCB 10 at one end and to a 9-way connector (e.g. the Molex housing with part number 51021 -0900 and crimp terminals with part number 50079-8000 or 50079-8100) at the other end. The flying lead 9 passes through an aperture 16 in the PCB 10 which is then filed with a non-flammable adhesive to provide strain relief. Each of the bundle of wires making up flying lead 9 is typically 90mm in length between the PCB 10 and the 9-way connector.

The housing 11 is shown in isolation and in more detail in Figure 4. In simplistic terms, housing 11 provides a set of apertures through which light may pass. The first two apertures 17 and 18 are for an infrared track sensor comprising an infrared light emitting diode (LED) and an infrared phototransistor (both not shown). A suitable infrared LED is made by Agilent and has part number HSDL44. Agilent also make a suitable phototransistor with part number HSDL54. The infrared LED and phototransistor are soldered to the PCB 10 and sit within the apertures 17 and 18 in housing 11 when assembled. By detection of reflected infrared light, they can detect the presence of banknotes as they pass the detector assembly 6. The LED and phototransistor making up the track sensor are connected to the controller electronics of the banknote processing equipment via a 4-way connector 28 soldered to the PCB 10.

Aperture 19 is for a red LED 23 (see Figure 3) which is intended to provide a warning that ultraviolet radiation is currently being emitted by the detector assembly 6. The red LED 23 is soldered to the PCB 10 and sits within aperture 19.

Aperture 20 houses an ultraviolet LED 24 (see Figure 2) which is the source of ultraviolet light for the counterfeit detection. Aperture 20 is angled such that the LED 24 lies at an angle of 45° to the top face of the housing 11 and to window support ledge 29 (described in detail later). A suitable ultraviolet LED is made by Nichia and has part number NSHU550E. This LED has an output power of 1000μW at a forward current of 10mA and produces ultraviolet light at a wavelength of 370nm with a half-power beam angle of ±60°.

Aperture 20 also houses an ultraviolet pass band optical filter (not shown). This filter prevents unwanted spectral components of visible light produced by the ultraviolet LED 24 from illuminating the passing banknotes. A suitable filter is made by Hoya with part number U360. This filter has a thickness of 2.5mm. Apertures 21 and 22 each provide a path through which light may travel to impinge on respective photodiodes 25 and 26 (see Figure 3). Each of the two apertures 21 and 22 houses a respective optical filter.

In aperture 21 , a stack of two filters and a 5mm hemispherical lens are fitted. The filters are a Hoya U360 (2.5mm thick) and a Schott GG420 (1 mm thick) which filter infrared and visible light whilst allowing the passage of ultraviolet light. The 5mm hemispherical lens focuses on the target area through the glass window 7, effectively reducing the area of internal reflection from the inside of the glass.

In aperture 22, a stack of three filters is fitted. These are a Schott GG420 (1 mm thick), a Schott BG39 (2mm thick) and a Schott BG 4 (2mm thick). Together, these filter infrared, visible red and ultraviolet light but allow the passage of visible blue light. Thus, the photodiode 25 underneath aperture 21 can detect ultraviolet radiation reflected from passing banknotes via the 5mm lens whilst photodiode 26 underneath aperture 22 can detect any blue light emitted from passing notes as the result of fluorescence.

Suitable devices for photodiodes 25 and 26 are made by Silonex with part number

SLWD-61N2. These have a device area of 5.1mm², a nominal sensitivity of 0.55A/W (±5%), a specific sensitivity of 0.12A/W (±5%), and a half-acceptance angle of ±60°.

The housing 11 also has an aperture 27 which allows a pin to pass through it between the window 7 and the PCB 10. The purpose of this will be described later.

Housing 11 also comprises a barrier 28 disposed between the aperture 20 for the ultraviolet LED 24 and the apertures 21 and 22 for the photodiodes 25 and 26. The barrier 28 also separates the apertures 21 and 22 from each other. The purpose of the barrier 28 is to prevent light from the LED 24 being received directly by either of the photodiodes 25 and 26. The top edge of the barrier 28 lies beneath the window support ledge 29 such that a gap exists between the barrier 28 and the plane defined by the window support ledge 29. This gap is necessary due to the compact nature of the detector assembly 6 and ensures that all ultraviolet light emitted by LED 24 illuminates passing notes rather than impinging on the barrier 28, which would be the case if the gap were not present. The size of the gap is nominally 1.7mm between the top of barrier 28 and the underside of glass window 7. This gap size results in the barrier 28 lying at the same height as the top of the filter stack in aperture 22 which minimises the interruption of the beam from ultraviolet LED 24 by barrier 28 without letting light directly into the filter.

This feature helps improve the immunity of the detector assembly 6 to "note flap", i.e. the fact that a passing note could pass over the detector window 7 at any height between 0 and 4mm (the separation between the guide plate 1 and the opposing guide plate). The reasons for this are explained below.

The LED 24 is intended to provide a cone of ultraviolet light for illuminating the passing notes. The photodiodes 25 and 26 would then receive light from respective spots on the notes. As the distance between the note and the guide plate 1 varies, the size of the illuminated spot also varies as do the sizes of the spots from which photodiodes 25 and 26 receive light. However, the spots from which photodiodes 25 and 26 receive light are always totally encompassed within the illuminated spot. Therefore, although the size of the spots will vary with the distance between the note and the guide plate 1 , the amount of energy received by photodiodes 25 and 26 remains approximately the same, irrespective of the distance between the passing notes and the guide plate 1.

However, it is necessary to provide the barrier 28 to prevent direct coupling of light from LED 24 to photodiodes 25 and 26. Ideally, there would be no gap between the barrier 28 and the glass window 7 to prevent reflections from the glass window 7 being received by photodiodes 25 and 26. However, if there were no gap then barrier 28 would cast a shadow on passing notes and prevent the illuminated spot and the spots from which photodiodes 25 and 26 receive light being coincident. This would render the detector very sensitive to distance between the passing notes and the guide plate 1.

By reducing the height of barrier 28 slightly such that a gap exists between its top edge and the window support ledge 29, the detector assembly 6 is rendered relatively immune to "note flap" because the spots are then coincident, albeit with a slight decrease in the isolation of the photodiodes 25 and 26 from directly-emitted ultraviolet light from LED 24. This lack of isolation causes a slight output offset to be produced by photodiodes 25 and 26 due to continuous direct illumination of photodiodes 25 and 26 by internal reflection. It has been found that this compromise provides good immunity to "note flap" at an offset level that is within the capability of the detection system. This design of detector also allows a more compact detector assembly to be produced.

The window support ledge 29 provides a surface on which the window 7 may be borne. It lies flush with the top face of an insert 30 which fits into the housing 11 and holds the filters in place in the apertures 20, 21 and 22.

The glass window 7 is held in place on the window support ledge 29 by a gasket 31 made of 0.53mm thick black urethane foam. The gasket has a layer of 3M's 467 adhesive to adhere to the glass window 7, and a layer of 3M's 9460 adhesive on the reverse side to adhere to the window support ledge 29. The gasket has an aperture 32 allowing any light emitted by the red LED 23 to be seen. The glass window 7 is non- fluorescent.

The glass window 8 covering the track detector LED and photodiode is affixed to the housing in a similar way.

A layer of metal 33 is deposited on the underside of glass window 7. The layer of metal 33 is deposited such that apertures are provided over the ultraviolet LED 24 and the red LED 23. The layer of metal 33 is also formed in a cross-hatch pattern over the photodiodes 25 and 26. The purpose of layer of metal 33 is to provide an electrical path to ground for any electrostatic charge that builds up in the glass window 7 as a result of friction between this and the passing notes. This is necessary to prevent damage or interference to the electronics on PCB 10 being caused by electrostatic discharge. A pin 34 is electrically connected to the layer of metal 33. The pin 34 passes through aperture 27 in housing 11 and is soldered to a ground plane on PCB 10, thereby providing a conductive path to ground for any electrostatic charge that has built up in glass window 7.

The layer of metal 33 is whitish in colour. However, it is hoped to be able to replace this with a layer of metal that is black in colour, because the white layer reflects ultraviolet light to some extent which would be diminished significantly by a black layer.

As can be seen best in Figure 3A, glass windows 7 and 8 both slope along the path of the notes. The leading edge is lower than the trailing edge. As can be seen in Figure 1 , this results in the leading edge of the glass windows 7 and 8 being lower than the surface of the guide plate 1 , whilst the trailing edge is proud. This prevents notes from catching on the glass windows 7 and 8 as they pass.

The electronic circuitry will now be described with reference to Figures 6 and 7. Figure 6 shows the electronic circuit that is located on the PCB 10 whilst Figure 7 shows further circuitry for signal processing located elsewhere in the banknote processing equipment.

As shown in Figure 6, the flying lead 9 is terminated in a 9-way connector 35. The first and ninth pins of this connector provide a 5-volt power supply (labelled +5V in Figure 6) and O-volt ground reference to the PCB 10 respectively. The O-volt ground reference is tied directly to a ground plane on one of the layers of PCB 10, which is of a multilayer construction. The 5-volt supply is decoupled by electrolytic capacitor 37 (typical value 10 μF) and parallel capacitor 38 (typically ceramic of value 0.1 μF). These capacitors are supplied from the 5-volt supply by a low-value resistor (typically 10 Ω), which limits the inrush current to capacitor 37 when it initially charges.

A negative 5-volt supply is also provided via pin 6 of connector 35. This is connected to decoupling capacitor 39 (typically 1 μF) and the input of voltage regulator 40, which is a MAX1735EUK25-T device from Maxim Integrated Products. This is a linear regulator which regulates the negative 5-volt supply to a negative 2.5-volt supply

(labelled V- in Figure 6). The negative 2.5-volt supply is used to provide a negative supply rail to the operational amplifiers. The "SHDN" (shutdown) pin of voltage regulator 40 is tied to the input so that the device is permanently enabled, whilst the

"SET" pin is tied to the O-volt ground plane so that the regulator provides its intended preset voltage (-2.5V in this case). The output of regulator 40 is decoupled by capacitor

41 (typically 1 μF). The output is loaded permanently by resistor 42 (typically 470 Ω).

This is necessary because the negative supply rail of the operational amplifiers that are used take a very small bias current. The resistor 42 therefore ensures that the operatio of regulator 40 is stable at all times.

Pins 4 and 8 of the connector 35 are joined together. This provides a simple way for the banknote processing equipment to detect the presence of the detector assembly 6 since pins 4 and 8 will appear to be a short circuit then the detector assembly 6 is installed.

Pins 2 and 3 of the connector 35 provide power to the ultraviolet LED 24, which as can be seen is in series with red LED 23. This ensures that when the ultraviolet LED 24 is turned on, red LED 23 will also be emitting red light, which is easily visible by an operator. Therefore, the operator is warned of the potential danger of injury to their eyes by the ultraviolet light and can avoid exposure to it. However, in normal operation the ultraviolet LED 24 should be turned off by an interlock if the processing equipment is opened such that the LED 24 could become visible to the operator. This feature is provided in case the interlock fails.

The current supplied to the ultraviolet LED 24 is adjustable under microprocessor control, although the pulse width modulation circuitry that carries out this function is not shown as it is not mounted on PCB 10. The maximum current that may be supplied to ultraviolet LED 24 is 15mA.

The photodiodes 25 and 26 are each connected to the input of a respective channel of signal processing electronics built around operational amplifiers 43a and 44a, and 43b and 44b. Each channel is identical, and so only that built around operational amplifiers 43a and 44a will be described. The corresponding components in the second channel are referred to by the same reference numeral with the "a" suffix replaced with a "b".

All of the operational amplifiers are supplied from the 5-volt supply (+5V) and the negative 2.5-volt supply (V-). This allows the amplifier inputs and outputs to swing negative should they receive a negative-going signal from the photodiodes.

Photodiode 26 is connected directly to the inverting input of operational amplifier 43a. This is connected as a transimpedance amplifier (i.e. a resistor 45a is connected from the output of operational amplifier 43a to the inverting input), and the non-inverting input of operational amplifier 43a is connected to the 0-volt ground plane. The output of operational amplifier 43a will attempt to keep the inverting input also at this potential (i.e. a virtual ground) by injecting an identical current into the inverting input as is injected by the photodiode 26. A typical value for resistor 45a is 10 MΩ, in which case the output voltage will be 10 million times larger in magnitude (but inverted in polarity) than the current injected by the photodiode 26 (i.e. 1 volt output for every 0.1 μA of current injected by photodiode 26). A capacitor 46a (typically 8.2pF) is connected in parallel with resistor 45a and provides high-frequency filtering of any noise that is coupled into the circuit.

A guard ring is formed as a continuous copper track on the PCB around the cathode of photodiode 25, the inverting input of operational amplifier 43a and the terminals of resistor 45a and capacitor 46a that are connected to that input. The guard ring is connected to the ground plane. Its purpose is to prevent any stray leakage currents from coupling (by any mechanism - resistively, capacitively or inductively) across the surface of the PCB 10 or any of its layers into the sensitive input circuitry around the photodiode 25.

The output from the transimpedance amplifier formed around operational amplifier 43a is connected, via resistor 47a (typically 1 kΩ), to the non-inverting input of operational amplifier 44a. This is connected in a standard non-inverting amplifier configuration to act as a buffer amplifier for operational amplifier 43a.

A network of resistors 48a (typically 430 kΩ) and 49a (typically 1 kΩ) and electrolytic capacitor 50a (typically 10 μF) is also connected to the non-inverting input. Resistors 47a and 48a act as a potential divider between the filtered 5-volt supply and the transimpedance amplifier output from operational amplifier 43a. The effect of this network is therefore to provide a slight positive bias (12m V) to the signal at the non- inverting input, thereby ensuring that all negative going parts of the signals are brought above OV. This is necessary because subsequent circuitry in the signal processing path does not have the capability of handling a negative-going signal. The capacitor 50a ensures that the positive bias applied by this network remains stable. The positive bias compensates for a negative offset on the output of the transimpedance amplifier so that the final output from the signal processing circuitry has a positive value which can be measured by an analogue to digital converter.

A resistor 51a is provided in series with the output of operational amplifier 44a to limit the current that might be drawn from it in the event of a short circuit. The dc gain of the non-inverting amplifier is set by resistors 52a (typically 2.2 MΩ) and 53a (typically 160 kΩ), and thermistor 54a (typically 330 kΩ). With the values given the gain will be around 25.6. Thermistor 54a is provided to compensate for any drift in the brightness of ultraviolet LED 24 as the temperature varies by causing the gain to vary with temperature by a corresponding amount. A capacitor 55a (typically 33pF) is provided to filter out any high-frequency noise that has become coupled into the circuit.

The output from the first channel just described is connected to pin 5 of connector 25, whilst the output from the second channel is connected to pin 7.

The track sensor components (connector 28, infrared photodiode 56 and infrared LED 57) are also mounted on the PCB 10.

The remainder of the signal processing circuitry is shown in Figure 7. Again there are two channel, each of which is identical with the other. Therefore, only the first channel will be described.

Pin 5 of connector 25 is connected via digital potentiometer 56a to the inverting input of operational amplifier 57a. Another digital potentiometer 58a is connected between the output and inverting input of operational amplifier 57a. The non-inverting input of operational amplifier 57a is connected directly to the OV ground plane. Operational amplifier is thus configured as an inverting amplifier with its gain determined by the ratio of the resistance of digital potentiometer 58a to that of digital potentiometer 57a. These resistance values are adjustable by microprocessor 63. Suitable devices for the digital potentiometers 56a and 58a are made by Dallas under part number DS1267S- 100. A suitable device for operational amplifier 57a is made by Texas Instruments under part number TLC2274ACD.

The output of operational amplifier 57a is connected to an integrator formed from resistor 59a, operational amplifier 60a and capacitor 61a. An integrator is used to filter out any instantaneous changes in the level of light received by photodiode 26. The resistor 59a is connected between the output of operational amplifier 57a and the input of operational amplifier 60a whilst the capacitor 61a is connected from the output of operational amplifier 60a to its non-inverting input. A switch 62a, which may be closed under the control of microprocessor 63, is connected in parallel with capacitor 61a to discharge it at the end of each period of integration. Suitable values for resistor 59a and capacitor 61 a are 1 MΩ and 1nF respectively. This sets the integrator time constant to 1ms. A suitable device for operational amplifier 50a is made by Texas Instruments under part number TLC2274ACD. A suitable device for analogue switch 62a is made by National Semiconductors under part number DG411.

The output from the integrator is connected directly to an analogue to digital converter input of microprocessor 63. Further signal processing is performed under software control by microprocessor 63 according to the method shown in Figure 8 and described below.

The microprocessor 63 receives pulses from a timing wheel (not shown). The timing wheel produces the pulses in response to movement of the transport. It typically comprises a slotted disc that passes through between an optical emitter and receiver. The receiver produces a pulse whenever it receives light emitted by the emitter (i.e. when the light can shine through one of the slots in the disc). The slots in the disc are spaced apart around the circumference of the disc such that pulses are typically produced for every 4.42mm of transport travel.

When the rising edge of each pulse is received from the timing wheel, the microprocessor 63 starts an internal timer and opens the switches 62a and 62b, thereby allowing the integrators to commence integration of the signals at their inputs. At the end of the timer period the analogue to digital converter inputs are sampled and converted and the switches 62a and 62b closed, thereby resetting the integrators. The timer period is set to be shorter than 1 ms to ensure that the integrator does not saturate before the sample is taken when a maximum voltage step change is applied to its input. A typical value of timer period for the reflectance sensor is 600μs and 500μs on the fluorescence sensor. This difference in period exists because the sample from the fluorescence sensor is taken first with the reflectance sensor being sampled 100μs later. The sample period is set to be less than the minimum transport clock period.

The sampled data are firstly restored to a dc level of 0 volts by subtracting a constant value (which corresponds to the positive bias added to the non-inverting input of operational amplifier 44a). These dc-restored data values are then stored in a first-in first-out (FIFO) buffer capable of storing up to two notes' worth of data values.

In this way, data values corresponding to the reflected ultraviolet light detected by photodiode 25 and the fluorescence detected by photodiode 26 are continuously sampled and stored. It is possible for the microprocessor 63 to determine which data values correspond to the leading and trailing edges of each note using the track sensor because this lies a defined distance downstream of the photodiodes 25 and 26.

Up to this point the signal processing has been the same for each of the two photodiodes 25 and 26. The data that has been sampled for each of the two photodiodes 25 and 26 is now processed by different algorithms to attempt to detect the presence of a counterfeit banknote.

With respect to the data corresponding to the reflected ultraviolet light detected by photodiode 25, there are a variety of techniques that were tried. These included a simple peak detection algorithm (i.e. a note was considered genuine if it exhibited a peak above a threshold), and comparison of an average with a threshold. However, it was found that a good soiled note could exhibit a low peak whilst a counterfeit note, if well made, could exhibit a high peak. Furthermore, large patterns on genuine notes tended to pull down the average value of the samples leading to them falling below the threshold. These two algorithms were therefore discounted.

Instead, a so-called "upper average" algorithm was developed. In this algorithm, shown in Figure 9, the average value of the sampled data values for each note is calculated.

Data values falling below this average value are discarded, leaving only those data values equal to or greater than the average value remaining. An average of the remaining data values is then calculated and compared with upper and lower thresholds. The note will be considered genuine if the average of the remaining data values falls between the two thresholds, and it will be considered counterfeit if it falls outside the two thresholds. The two thresholds are empirically derived from banknotes that are known to be genuine.

The values of the thresholds will depend on the nature of the documents being processed. In one case, the threshold levels were calculated from the distributions of the "upper average" measurements captured from genuine US dollars and dollar counterfeits. The lower threshold was set above the levels obtained from the captured counterfeits so as to ensure they are rejected, with sufficient margin that some badly soiled genuine notes will also be rejected. An upper threshold was set above the maximum levels measured on the captured good condition genuine US dollars, in case there are counterfeits in the future with higher levels of UV reflectance than genuine US dollars.

With respect to the data values corresponding to the fluorescence detected by photodiode 26, the algorithm is shown in Figure 10. The sampled data values for each note are firstly scanned to see whether any have saturated (i.e. the data value is the full scale value that can be produced by the analogue to digital converter. Any data values which have saturated are discarded. If more than a predetermined number of data values have saturated then the note will be considered counterfeit. A certain number of saturated samples is expected because many documents and notes have features that fluoresce. However, they are usually quite narrow and vary from document to document. Therefore, the number of saturated data values that are accepted before a note is considered counterfeit will be variable, but would typically be less than 50%.

The remaining data values are simply averaged. This average value is then compared with a single threshold. If the average value falls below the threshold then the note is considered to be genuine, otherwise it is considered to counterfeit. The value of this threshold varies depending on the sensitivity set by an operator (mentioned below).

Before it can be used for processing banknotes, the detector assembly 6 must first be calibrated. The calibration process is described below with reference to Figure 11. The calibration process for both the reflected ultraviolet and fluorescence sensors is carried out using reference documents, each of which is unique. Each document has a specific value of normalised reflectance or normalised fluorescence (e.g. 1 , 0.5, 0.75, 2 etc.) indicating how well it reflects ultraviolet light or fluoresces relative to a standard value. Each document will be marked with a "scale" value corresponding to its normalised fluorescence and/or reflectance. It is necessary to calibrate the fluorescence sensor first. This is done by inserting a reference document into the machine so that the ultraviolet light emitted by LED 24 causes the document to fluoresce, which is detected by photodiode 26. The "scale" value marked on the reference document being used is entered into the banknote processing equipment (via a remote PC or via a keypad on the equipment) and logged by the microprocessor 63.

The microprocessor 63 adjusts the value of digital potentiometer 56a to be the same as the "scale" value. Thus, the value of digital potentiometer 56a depends on the normalised fluorescence value of the reference document. For example, if the normalised fluorescence value is 1 then the "scale" value might be 128 (the full scale value being 255), and if the normalised fluorescence value is 0.5 then the "scale" value might be 64.

The switch 62a is then opened to allow integration to start, and after a predetermined period has elapsed, an analogue to digital conversion is performed of the output of the integrator. From the resultant data value, an offset value is subtracted to compensate for the positive bias added to the non-inverting input of operational amplifier 44a. The microprocessor 63 then adjusts the value of digital potentiometer 58a, known as the "gain" value, so that the compensated data value lies within a desired range, which is determined by the value of normalised fluorescence of the reference document.

If the microprocessor 63 cannot cause the compensated data value to lie within the desired range by adjusting the "gain" value, it is possible for it to change the drive current to LED 24 using the pulse width modulation circuitry mentioned earlier. The drive current will not then be altered again because it would then be necessary to recalibrate the fluorescence sensor.

With the reflected ultraviolet sensor, the calibration process is slightly different. In this process, the ultraviolet light reflected by glass window 7 is first measured before a reference document is inserted. This must be done at each "gain" value of digital potentiometer 58b (i.e. all values from 0 to 255) and the resultant values logged by the microprocessor 63 because when the reference document is inserted there will be reflection from both this and the glass window 7 and the relative contribution of each will vary depending on the "gain" value. However, the first step is still to enter the "scale" value marked on the reference document that will be used into the banknote processing equipment (via a remote PC or via a keypad on the equipment) and logged by the microprocessor 63. The microprocessor 63 adjusts the value of digital potentiometer 56b to be the same as the "scale" value. Thus, the value of digital potentiometer 56b depends on the normalised reflectance value of the reference document. For example, if the normalised reflectance value is 1 then the "scale" value might be 128 (the full scale value being 255), and if the normalised reflectance value is 0.5 then the "scale" value might be 64.

The process of logging the values of reflection from the glass window 7 is then commenced. In this process, the LED 24 is caused to illuminate the glass window 7 and the reflected ultraviolet light is detected by photodiode 25. The switch 62b is then opened to allow integration to start, and after a predetermined period has elapsed, an analogue to digital conversion is performed of the output of the integrator. From the resultant data value, an offset value is subtracted to compensate for the positive bias added to the non-inverting input of operational amplifier 44b. This is repeated for each "gain" value and the results stored by microprocessor 63. This process is not required for calibration of the fluorescence sensor because the glass window 7 does not fluoresce.

The reference document is then inserted into the machine so that the ultraviolet light emitted by LED 24 is reflected from the document and detected by photodiode 25.

The switch 62b is then opened to allow integration to start, and after a predetermined period has elapsed, an analogue to digital conversion is performed of the output of the integrator. From the resultant data value, an offset value is subtracted to compensate for the positive bias added to the non-inverting input of operational amplifier 44b. The value of reflection from the glass window 7 for the current "gain" value (128 at the start of the calibration process) is then subtracted from the compensated data value. If the resultant value does not lie within a desired range, the microprocessor 63 adjusts the value of digital potentiometer 58b accordingly. The step of converting the reflectance signal to a digital representation, compensating for the positive bias and subtracting the value of reflection from the glass window 7 (for the new "gain" value) is repeated until the resultant value does lie within the desired range, which is determined by the value of normalised reflectance of the reference document. When the calibration process has been completed, the microprocessor 63 stores the "gain" and "scale" values and the required drive current for LED 24 in non-volatile memory. These values would be set to zero for a detector assembly that has not yet been calibrated. Therefore, the microprocessor 63 can determine whether a detector assembly requires calibration.

If a detector assembly is installed that has not yet been calibrated this is detected by microprocessor 63, which will then refuse to process any banknotes. In this case, the "gain" and "scale" values of digital potentiometers 56a, 56b, 58a and 58b are set to their midpoint value of 128. The calibration process must be completed before notes may be processed.

Once the system has been calibrated, the user may adjust the "scale" values of either the fluorescence or reflectance sensors to alter the respective sensitivities to optimise the detector for the notes that are being processed.

Claims

1. A detector assembly for use in sheet document processing comprising:

at least one sensor for sensing resultant radiation received from the sheet document; and

2. A detector assembly according to claim 1 , wherein the window is made from glass.

3. A detector assembly according to claim 2, wherein the glass is non-fluorescent

4. A detector assembly for use in sheet document processing comprising:

at least one sensor for sensing resultant radiation received from the sheet document;

a window, disposed between each of the emitter and the at least one sensor and the sheet path, through which the beam and resultant radiation pass; and

5. A detector assembly according to claim 4, wherein the window is separated from the opaque barrier by the gap.

6. A detector assembly according to either of claims 4 to 5, wherein the beam of radiation passes through the gap as it travels from the emitter to the window.

7. A detector assembly according to any of claims 4 to 6, wherein the emitter emits ultraviolet radiation.

8. A detector assembly according to any of claims 4 to 7, wherein the emitter comprises a light emitting diode.

9. A detector assembly according to any of claims 4 to 8, wherein the at least one sensor is sensitive to fluorescence of the sheet documents.

10. A detector assembly according to claim 9, further comprising a filter that passes only blue light, thereby adapting the at least one sensor to detect only fluorescence that is blue in colour.

11. A detector assembly according to any of claims 4 to 10, wherein the at least one sensor is sensitive to radiation reflected from the sheet documents.

12. A detector assembly according to claim 11 , further comprising a filter that passes only ultraviolet light, thereby adapting the at least one sensor to detect reflected radiation that is of ultraviolet wavelength.

13. A detector assembly according to claim 11 or claim 12, further comprising a hemispherical lens for focussing the at least one sensor on the document.

14. A detector assembly according to any of claims 4 to 13, wherein the at least one sensor comprises a photodiode.

15. A detector assembly according to any of claims 4 to 14, wherein the window is made from glass.

16. A detector assembly according to claim 15, wherein the glass is non-fluorescent.

17. A detector assembly according to any of claims 4 to 16, wherein the window is provided with a conductive layer which may be electrically connected to an electrical ground in order to dissipate any static charge that builds up in the window.

18. A detector assembly according to any of 4 to 17 claims, further comprising a housing within which the emitter and at least one sensor are mounted.

19. A detector assembly according to claim 18, wherein the emitter and at least one sensor lie in a cavity within the housing over which the window is mounted.

20. A detector assembly according to claim 19, wherein the housing is mounted on a printed circuit board which lies beneath the cavity, thereby ensuring that radiation may only enter the cavity through the window.

21. A detector assembly according to any of claims 18 to 20, wherein the opaque barrier forms part of the housing.

22. A detector assembly according to any of claims 18 to 21 , wherein detector assembly comprises two sensors and the housing comprises a barrier interposed between the two sensors.

23. A detector assembly according to claim 22, wherein the two sensors are responsive to reflected ultraviolet and fluorescent radiation from the sheet documents respectively.

24. A detector assembly according to any of claims 18 to 23, wherein the housing is made from polycarbonate.

25. A method of detecting a counterfeit sheet document comprising:

a) irradiating the sheet document;

b) sensing radiation reflected from a plurality of regions on the sheet document and generating corresponding reflected radiation signal values; c) calculating the average of the reflected radiation signal values;

26. The method of claim 25, wherein the sheet document is irradiated with ultraviolet radiation.

27. The method of claim 25 or claim 26, wherein the reflected radiation is of ultraviolet wavelength.

28. The method of any of claims 25 to 27, wherein the reflected radiation signal values are generated for each of the plurality of regions by:

i) sensing the reflected radiation with a photodiode;

ii) amplifying the signal from the photodiode;

iii) integrating the amplified signal; and

29. The method of claim 28, wherein the radiation reflected from the plurality of regions is detected by causing the sheet document to move relative to an emitter for irradiating the document and a sensor for receiving the reflected radiation, and periodically performing steps (iii) and (iv).

30. A method of calibrating a detector assembly according to any of claims 1 to 24, the detector assembly further comprising a signal processing amplifier connected to the at least one sensor, the method comprising:

a) irradiating a reference document having known values of fluorescence;

31. A method of calibrating a detector assembly according to any of claims 1 to 24, the detector assembly further comprising a signal processing amplifier connected to the at least one sensor, the method comprising:

a) irradiating a reference document having known values of reflectance;

32. The method of claim 30 or claim 31 , wherein the reference document is irradiated by ultraviolet radiation.

33. The method of any of claims 30 to 32, wherein the gain of the signal processing amplifier is defined by the ratio of a first resistor to a second resistor, step (c) comprising setting the value of the second resistor depending on the reflectance or fluorescence value of the reference document, and then setting the value of the first resistor to adjust the gain of the signal processing amplifier such that the output is equal to a predetermined value or lies within a predetermined range of values corresponding to the reflectance value of the reference document.

34. The method of claim 30, further comprising adjusting the drive current to an emitter for irradiating the reference document in step (c).

35. The method of any of claims 30 to 34, further comprising subtracting a value representative of any extraneous reflected radiation or fluorescence that does not emanate from the reference document from the output value in step (c) before adjusting the gain.