Field of the invention The present invention relates to a method and apparatus for counting banknotes. Background

Plural banknotes are assembled in bundles. In order to establish the value of the bundle of banknotes, it is convenient to use a counting apparatus rather than count manually.

Known banknote counters generally fall into two categories; friction counters and counters by weight. Friction counters count each banknote individually by moving each banknote past a sensor. In this arrangement each banknote has to be moved individually past the sensor and the count register has to be incremented. This means that friction count machines have many moving parts that move at high speed, are subject to wear and can jam or clog. The present invention seeks to provide a counting method and apparatus that does not require the banknotes to be moved individually and has a greatly reduced moving component part count.

Count by weight counters weigh a bundle of banknotes and divide that weight by the known weight of a single banknote to give a result for the number of banknotes in the bundle. This means that it is difficult to achieve accuracy when banknotes of differing weights exist within a denomination population or when banknote weight is altered by addition of tape, a torn and missing section exists, notes absorb ambient moisture etc. The present invention seeks to provide a counting method and apparatus that counts banknotes accurately irrespective of differing note weights, size, thickness or material.

Other apparatus and methods exist for counting sheets by image processing that rely on either statistical analysis of a processed image or on recognition of a repeated pattern. The apparatus that use a statistical analysis method are not accurate enough to consistently count used banknotes and the apparatus that use a pattern recognition method are primarily used for counting corrugated sheets and are not suitable for counting used banknotes where no repeat pattern occurs. All these apparatus and methods possess commonality in terms of apparatus in that they must all comprise a means to capture an image, a means to process the captured image and a means to count the edges in the image. Significant differences only exist in the method employed and particularly in the means by which the captured image is processed and the means by which the edges in the image are counted. The present invention seeks to provide a counting method and apparatus that counts banknotes accurately. Prior art

United States Patent US 5 534 690 (Goldenberg et al) and European Patent EP0805992 disclose a method for counting banknotes employing an optical sensor comprising: counting the number of banknotes in the stack characterized in that the mutual orientation of the banknotes relative to said optical sensor is substantially maintained. The counting step includes: employing at least one optical sensor for generally simultaneously viewing at least two separate columns along a surface defined by edges of the banknotes in the stack; and receiving an output from said optical sensor and providing an output indication of a number of banknotes in the stack. This output indication is an average of the different values provided by the counts from each column viewed. A count is determined by analysing the number of non zero pixels within a column.

However, this method utilises a statistical approach to counting the edges in an image and no products exist in the market place that use the present Method and Apparatus for counting banknotes. Beyond Technologies Limited, the owners of these patents, have no products that count banknotes and have stated on their website, as recently as 2009, that within a few months it would also be possible to count paper, but in 2010, this statement was removed and there are no existing products for counting banknotes in their product range, which focuses on counting of flat cards.

European Patent Application EP 1473665 discloses a method for counting stacks of sheets whereby the imaged side of the stack is split into a number of columns and a separate count is made for each column. A count is made based on the number of lines that appear in each column. A statistical treatment is then performed on the result and this determines whether the count is deemed valid and what the final count is determined to be. JP Patent Application JP10134162 discloses a method of and apparatus for counting cards where the problem of operation noise and vibration are completely eliminated by adjusting and closely stacking the cards using a card pusher prior to illuminating and then enlarging the reflected light by employing a lens before guiding the light to a charge- coupled device. Processing software converts the information gathered into a card count.

Another area where counting stacks of a material is important is in corrugated boards, one solution is disclosed in US Patent 6173607. The corrugated boards have a distinct flute or wave pattern repeated along the cross-section with a substantially uniform wavelength. Processing software isolates this flute or wave pattern to provide a count. The apparatus which is used includes a vertically displaceable imaging device; this solves the problem of obtaining a clear enough image to enable counting of the boards without having to use multiple imaging devices which has a higher cost implication.

US Patent 6091792 is also concerned with counting corrugated boards and uses an imaging device which, during a counting process is displaced with respect to the stack of boards. There is no physical contact between the counting apparatus and the boards. This invention is concerned with the problem of counting individual boards more than once i.e. that individual boards appear in consecutive images resulting in an inaccurate count. To mitigate this problem, images are compared and where there is an overlap this is identified and removed from the count. Summary of the invention

According to a first aspect, the present invention consists in an apparatus for counting the number of banknotes in a bundle of banknotes, the apparatus comprising: an imaging device for capturing an image of an edge of the bundle of banknotes; and a controller for processing the image whereby the image is scanned, grey block data recorded and analysed wherein, when the controller has processed the image, a count of the number of banknotes in a bundle is produced. The captured image typically comprises more than one adjacent pixel along the edge of the bundle of banknotes.

Preferably, the captured image comprises a section of the edge of the bundle of banknotes. A section means a continuous length of the edge of the bundle of banknotes that comprises a plurality of adjacent pixels which are analysed in the same process.

Preferably, said imaging device comprises an image sensor and a lens arrangement. In a preferred embodiment, the apparatus comprises an object lighting array, operable to illuminate the edges of one or more banknotes in the apparatus with white, coloured, ultra-violet or infra-red light.

Preferably, the apparatus comprises an enclosure for enclosing at least the image sensor and the lens arrangement, the enclosure having an aperture, said aperture being disposed in line with and within the depth of focus of said image sensor and said lens arrangement.

In a preferred embodiment, said controller is a microprocessor.

An apparatus according to any of the preceding claims, comprising a banknote holder configured to provide optimum orientation and presentation of banknote edges. Preferably, said banknote holder comprises a sealing arrangement configured to substantially eliminate ingress of ambient lighting to the enclosure via the image aperture.

Preferably, the apparatus includes a banknote sensor, operable to detect the presence of one or more banknotes in the banknote holder. Preferably, the apparatus includes a banknote settlement device, operable to settle the banknote edges against the aperture. Preferably, said settlement device comprises a tamper or vibrating device.

In a preferred embodiment, the apparatus includes a separation device comprising: a clamping device; and a former having a generally curved surface whereby in use, banknotes are clamped by the clamping device and moulded by the former facilitating separation of the banknotes.

In a second aspect there is provided a banknote separation device comprising: a clamping device; and a former having a generally curved surface whereby in use, banknotes are clamped by the clamping device and moulded by the former facilitating separation of the banknotes.

Preferably, the clamping device includes two pairs of pinch rollers one of which is moveable in two generally orthogonal directions with respect to a bundle of bank notes.

According to a third aspect, the present invention consists in a method for counting the number of banknotes in a bundle of banknotes, the method comprising the steps of: capturing an image of an edge of a bundle of banknotes; processing the image; and counting the image of the edge of each banknote in the processed image.

Preferably, the captured image comprises a section of the edge of the bundle of banknotes.

In a preferred embodiment when the image is scanned, an array of grey block data is produced.

Preferably, the method comprises the step of orienting the banknotes to provide optimum orientation and presentation of banknote edges.

In a preferred embodiment, prior to the step of analysing the data, the image is enhanced by the steps of: setting a threshold based on the grey block data; modifying the data based on the threshold value; and re-scanning the modified data to produce an enhanced image. Preferably, the method comprises the step of applying a filter to the enhanced image. Preferably, at least one processing array is created and initialised to record image pixel colour, position and relational data.

In a preferred embodiment, the step of analysing the data of a section of the edge of a bundle of banknotes includes one or more of the following: identifying matching grey blocks; identify joins; identify splits; identify found notes; identify lost notes; and identify steps in the data.

Preferably, the method includes the step of applying algorithms to identify and/or correct anomalies.

Preferably, the method includes the step of identifying and recording each continuous and discontinuous note segment in a segments array.

Preferably, the method includes the step of assessing the quality of the image against acceptance criteria to ascertain confidence in count accuracy. If acceptance criteria are not met, processing is abandoned, the banknotes are re-positioned and a new image is captured and processed.

In a preferred embodiment each note segment in the segments array is assessed and compared relative to other segments to match note segments. Preferably, the position of each unmatched note segment is assessed and compared to steps in other segments to identify matching steps. In a preferred embodiment, segment positions are updated in a positions array to reflect changes from applying algorithms to identify and/or correct anomalies, identifying and recording each continuous and discontinuous note segment in a segments array, matching note segments, and matching steps. Preferably, the method comprises the step of detecting the presence of one or more banknotes in the apparatus.

Preferably, a count is displayed visually and/or stored in a memory for future access. The image sensor preferably comprises of a camera including a Complementary metal-oxide-semiconductor (CMOS) or Charge-coupled Device (CCD) image sensor. The associated, fixed focal length optic arrangement, fixed subject lighting and banknote presentation holder/ambient light eliminator are preferably disposed and fixed with reference to an enclosure.

The method can comprise processing an image of a section of the edge of a bundle of banknotes, and can include some or all of the following steps: saving a captured image to a bitmap file; scanning the captured image and recording grey block data; enhancing the contrast of the image; re-scanning the image, updating grey block data and recording black block data; applying a filter to the image; initialising and populating a set of processing arrays for the storage and manipulation of image data; analysing the image and recording "Join", "Split", "Lost", "Found" and "Step" data; applying corrective algorithms to correct shadow and distortion anomalies; initialising and populating a note segments array; assessing count confidence; matching note segments; matching steps; processing segment positions and updating image; displaying a count result.

The method of counting is by image processing, preferably by microprocessor and bespoke software that analyses the image data and identifies individual banknote edges thus ascertaining the number of banknotes in the bundle.

The invention provides a banknote edge counting method and apparatus in which a bundle of banknotes are presented to an imaging device, an image of a section of the edges of the bundle of banknotes is captured, processed and then analysed to detect the number of banknotes in the bundle.

The invention can be applied to any arrangement to count banknotes of any thickness whether flat or distorted. The banknote presence sensor is preferably operable to detect presence of a banknote and, in response to such detection, operable to trigger image capture.

The microprocessor circuit board is preferably incorporated in the enclosure. The image sensor, optic arrangement and subject lighting are preferably incorporated in the enclosure.

The enclosure preferably incorporates a banknote presence sensor fitted to the enclosure within the bounds of the banknote holder.

The enclosure preferably incorporates a digital display for display of options, results and any other user information. The enclosure preferably incorporates a method of selecting banknote denomination comprising selection buttons or options on a touch screen.

The enclosure preferably incorporates communication port(s) and/or wireless connectivity for communicating results to a microprocessor directly or over a network.

The invention is further explained, by way of an example, by the following description, to be read in conjunction with the appended drawings, in which: Brief description of the drawings

Figure 1 is a cross sectional view of an apparatus according to the invention showing the arrangement and disposition of the various image capture components; Figure 2 shows a first face of an enclosure suitable for use with the apparatus of Figure 1 , and illustrates the user operable controls and display;

Figure 3 shows a second face of the enclosure of Figure 2, illustrating power and communication arrangements;

Figure 4 shows a method for improving banknote separation that could be employed prior to image capture; and Figure 5 is a flow diagram detailing of the method employed to processing images to arrive at a banknote count total.

Detailed description of preferred embodiments of the invention Attention is first drawn to Figure 1, showing a cross section of an apparatus according to the present invention and illustrating the arrangement and disposition of the various image capture components.

An image sensor 18 is arranged in line and at a fixed distance from an optical lens arrangement 20 to give a fixed distance to subject and a fixed depth of field. The image sensor 18 and optical lens arrangement 20 are rigidly mounted within an enclosure 50 such that the distance from the front of the optical arrangement 20 to an image aperture 30 cut into the enclosure 50 ensures the image aperture 30 is positioned at the fixed subject distance and within the depth of field. In an alternative arrangement a plurality of image sensors 18 can be arranged at a fixed distance from a plurality of related optical lens arrangements 20 to give a fixed distance to subject and a fixed depth of field. The plurality of image sensors 18 and plurality of optical lens arrangements 20 are rigidly mounted within an enclosure 50 such that the distance from the front of the plurality of optical arrangements 20 to a plurality of image apertures 30 cut into the enclosure 50 ensures the plurality of image apertures 30 is positioned at the fixed subject distance and within the depth of field. This arrangement can allow simultaneous capture of a plurality of images of different areas of the edge of a bundle of banknotes consisting of a greater number of banknotes than the maximum number of banknotes that can be captured by a single image sensor 18 and single optical lens arrangement 20, or simultaneous capture of a plurality of images of different areas of the edge of a bundle of banknotes for count validation purposes. In an alternative arrangement the image sensor 18 is arranged in line and at a fixed distance from an optical lens arrangement 20 to give a fixed distance to subject and a fixed depth of field.

The image sensor 18 and optical lens arrangement 20 can be rigidly mounted to a motorised carriage running on a rigidly mounted track, mounted within an enclosure 50, that runs parallel and at a fixed distance from the subject such that the distance from the front of the optical arrangement 20 to an image aperture 30 cut into the enclosure 50 ensures the image aperture 30 is positioned at the fixed subject distance and within the depth of field and to allow capture of a scanned image of the edge of a bundle of banknotes consisting of a greater number of banknotes than the maximum number of banknotes that can be captured in a single image capture by the fixed arrangement. The optical lens arrangement 20 includes a minimum of one object lighting array for illumination of the banknote edges. In an alternative arrangement that has a plurality of lens arrangements 20 each lens arrangement 20 includes a minimum of one object lighting array for illumination of the banknote edges. In an alternative arrangement the lighting array or arrays can be separate from the optical arrangement 20 or plurality of optical arrangements 20 and independently mounted within the enclosure 50. A controller 70 is fitted within the enclosure 50. The controller 70 is operable to acquire and analyse captured image data. The controller 70 is preferably implemented in the form of a microprocessor 70, the microprocessor operating according to a program for acquisition of image data and subsequent analysis of the captured image data as hereinafter described. In an alternative arrangement the image sensor 18 and microprocessor 70 can be situated on a common printed circuit board.

The enclosure 50 is sealed to mitigate and preferably prevent the ingress of ambient light.

A banknote holder 40 is positioned on the top of the enclosure 50 around the image aperture 30 such that the edges of banknotes inserted into the banknote holder 40 are positioned in line with the image aperture 30 and within the depth of field of the optical lens arrangement 20. A rubber, brush or alternative sealing arrangement (not shown) is incorporated in the banknote holder to prevent the ingress of ambient light to the enclosure 50 through the image aperture 30. The bundle of banknotes to be counted is manually loaded into the banknote holder 40. In an alternative arrangement the bundles of banknotes to be counted can be automatically loaded to and unloaded from the banknote holder 40 consecutively by a banknote handling system (not shown) to facilitate fast processing of large quantities of banknotes. The controller 70 would exercise control and co-ordination of the banknote handling system.

A banknote sensor 60 is installed in the enclosure 50 and within the bounds of the banknote holder 40. Attention is next drawn to Figure 2, which shows a first face of an enclosure suitable for use with the apparatus of Figure 1 , and illustrates the user operable controls and display.

A digital display screen 90 is fitted to the enclosure 50 and displays operator options, messages and count results.

A set of mode selection buttons 80 are fitted to the front of the enclosure 50 for operator selection of count mode options. In an alternative arrangement the display screen 90 can be a touch sensitive graphical display screen for accessing operator mode selection choices through a menu system instead of or as well as the mode selection buttons 80.

Attention is next drawn to Figure 3, which shows a second face of the enclosure of Figure 2, illustrating power and communication arrangements.

A battery compartment 120 is incorporated in the enclosure 50 for alternative battery powered operation. A power supply socket 100 and various communication ports 1 10 are fitted to the enclosure 50 for attachment of an external power supply and adapted to transfer of data to and from the unit to a computer, either directly or via a local area network or wide area network. In an alternative arrangement the unit may also be fitted with a wireless transceiver for wireless communication.

Attention is next drawn to Figure 4, which shows a method and apparatus that can be used for separation of individual banknotes, in the bundle of banknotes to be counted, prior to image capture. The apparatus could be incorporated in the banknote holder 40. The separation device 160 includes two pairs of pinch rollers, a top pair 130 which are located on one side of the bundle and a bottom pair 140 which are located on the distal side of the bundle of bank notes. A former 150 having a generally curved surface is also provided on one side of the bundle.

Once the bundle of banknotes to be counted has been manually or automatically loaded into the banknote separation apparatus (Figure 4) in the banknote holder 40, as indicated in Figure 4, Schematic 5.1 : the top pinch rollers 130 move generally vertically with respect to the bundle of banknotes to lightly compress the bundle of banknotes between the top pinch rollers 130 and the former 150 as indicated in Figure 4, Schematic 5.2; the top pinch rollers 130 roll in opposing directions over the surface of the bundle of banknotes and parallel to the curved surface of the former 150 until they reach a position where the banknotes are trapped between the top pinch rollers 130 and the bottom pinch rollers 140 as indicated in Figure 4,

Schematic 5.3 shows top pinch rollers 130 being forced into firm contact with the banknotes gripping the bundle of banknotes firmly between the top pinch rollers 130 and the bottom pinch rollers 140, the former 150 withdraws and the bottom pinch rollers 140 and top pinch rollers 130 pivot in unison until perpendicular to the bundle of banknotes. The top pinch rollers 130 and the bottom pinch rollers 140 then move outwardly in unison until the bottom banknote in the bundle of banknotes becomes taught and the other banknotes bow by varying degree dependant upon their position within the bundle of banknotes.

Attention is finally drawn to Figure 5 which shows a flow diagram 300 detailing of the method employed to capture and process images of banknote edges in order to arrive at a banknote count total.

At commencement of operation, a bundle of banknotes is introduced into the banknote holder 40. In a first stage (indicated by the arrow 1 in Figure 5) the controller 70, responding to detection of banknote presence by the banknote sensor 60, co-operates with the image sensor 18 and the object lighting array to capture a greyscale image of a section of the edge of a bundle of banknotes. The captured image is saved to the microprocessor 70 as a bitmap file at a resolution required by subsequent image processing.

In a second stage, (indicated by the arrow 2 in Figure 5) the captured greyscale image is then scanned pixel column by pixel column and for each column an array of grey block data is produced and the minimum and maximum grey shade existent within each grey block is derived.

In a third stage (indicated by the arrow 3 in Figure 5) the grey block data is assessed for each grey block and a threshold value defined from the minimum and maximum grey shade values for that specific block. Using the defined threshold value rules are applied and certain pixels within the grey block that have grey shade values lower than the threshold are given a grey shade value of 0 (Black) in the greyscale image.

In a fourth stage (indicated by the arrow 4 in Figure 5) the now modified grey scale bitmap image is re-scanned pixel column by pixel column and for each column the array of grey block data is updated and an array of black block data is produced.

In a fifth stage (indicated by the arrow 5 in Figure 5) a filter is then applied to the modified grey scale image or the recorded grey scale image to identify random isolated single grey pixels and random isolated single black pixels and change their colour value to match their surrounding pixels.

In a sixth stage, (indicated by the arrow 6 in Figure 5) a suite of processing arrays is now created and initialised to record image pixel colour, position and relational data. The suite comprises a "Notes" array to hold grey pixel block data and three further match arrays, "JoinsLostMatch", "SplitsFoundMatch" and "StepsMatch" to hold pixel block relational data. In a seventh stage (indicated by the arrow 7 in Figure 5) the processing arrays are next populated with data derived from initial analysis of the modified grey scale image.

In an eighth stage (indicated by the arrow 8 in Figure 5) the modified grey scale image grey block data held in the Notes array is analysed column by column and block by block by comparing relational information between each grey block and the grey blocks in the previous and next adjacent columns. Where a relationship exists it is categorised as a "Join", a "Split", a "Lost" note or a "Found" note and recorded in the Notes array and relevant match array. Significant differences in matching grey block size are also assessed and recorded in the StepsMatch array. On completion a representative digital image is derived from the data held in the processing arrays.

In a ninth stage (indicated by the arrow 9 in Figure 5) numerous algorithms relating to differing anomalous conditions are applied to the modified grey scale image to identify anomalies caused by shadow and banknote distortion and the digital image is corrected accordingly.

In a tenth stage (indicated by the arrow 10 in Figure 5) the digital image is scanned pixel row by pixel row from top to bottom and each continuous and discontinuous note segment is individually identified and recorded in a "Segments" array with its position relative to adjacent note segments.

In an eleventh stage (indicated by the arrow 1 in Figure 5) the quality of the image is assessed based on the number of anomalies identified and recorded segment data. The result is compared to acceptance criteria and processing is abandoned, banknotes are repositioned and a new image is captured and processed, if criteria are not met. In an twelfth stage (indicated by the arrow 12 in Figure 5) the position of each segment start and end points in the Segment array relative to other segments end and start points is assessed against qualifying criteria and segments deemed to be associated with the same note are identified and recorded as matching segments in the Segments Array.

In a thirteenth stage (indicated by the arrow 13 in Figure 5) the position of any segment, with un-matched start point or end point is assessed, relative to steps in other segments, against qualifying criteria and segment start or end points deemed to match with a step are modified to appear as a "Split" or a "Join" and the Segments Array and digital image are modified accordingly. In a fourteenth stage (indicated by the arrow 14 in Figure 5) the initial Note positions are recorded in a Position Processing Array. In a fifteenth stage (indicated by the arrow 15 in Figure 5) the segment positions are updated in the Positions Array to reflect changes recorded during application of Corrective Algorithms, Segment Matching and Step Matching.

In a sixteenth stage (indicated by the arrow 16 in Figure 5) the digital image is updated to reflect segment position changes imposed during application of Corrective Algorithms, Segment Matching and Step Matching. A count of notes is derived from the number of occupied lines in the digital image.

In a seventeenth stage (indicated by the arrow 17 in Figure 5) the final note count derived from the final digital image is displayed.

It is to be understood that, while the processing is here described as taking place within the microprocessor 70, the output of the image sensor 18 or plurality of image sensors 18 can be communicated to a remote processor such as a PC and the image processing, which can include some or all of the steps: saving a captured image to a bitmap file; scanning the captured image and recording grey block data; enhancing the contrast of the image; re-scanning the image, updating grey block data and recording black block data; applying a filter to the image; initialising and populating a set of processing arrays for the storage and manipulation of image data; analysing the image and recording "Join", "Split", "Lost", "Found" and "Step" data; applying corrective algorithms to correct shadow and distortion anomalies; initialising and populating a note segments array; matching note segments; matching steps; processing segment positions and updating image; displaying a count result, can be conducted there.

It is to be appreciated that these Figures are for illustration purposes only and other configurations are possible.

The invention has been described by way of several embodiments, with

modifications and alternatives, but having read and understood this description further embodiments and modifications will be apparent to those skilled in the art. All such embodiments and modifications are intended to fall within the scope of the present invention as defined in the accompanying claims.