GB2487350A - Method of detecting a fault on an imaging device - Google Patents

Abstract

A method of detecting a fault on an imaging device, such as a scanner window, comprising the steps of: (a) capturing an image of an article viewable through an imaging device 102, the imaging device being divided into one or more cells, with each cell being assigned to a portion of the image; (b) processing the image to determine whether any of the cells contain computer-recognisable data; (c) If a cell does contain computer-recognisable data, analysis the data to determine if it is recognisable to a pre-determined probability of success 104; (d) maintaining a first count 106 for each cell when the computer-recognisable data contained in that cell is recognised and a second count 108 when it is not. Steps (a) to (d) are repeated for a plurality of articles and an error signal is generated when a threshold based on the first and second counts is reached during a fixed number of successive repetitions of steps (a) to (d).

Description

METHOD OF DETECTING A FAULT ON AN IMAG1NG DEVICE The present invention relates to a method of detecting a fault on an imaging device, in particular, a method of detecting a fault on a scanner window.

Traditional imaging devices and systems, for example still cameras using silver halide photographic films, or traditional video cameras or super-8 video films, are progressively being replaced by digital imaging systems. These digital imaging systems allow for more flexible processing of the captured images using digital methods, often automatic, and for individually-designed enhancement of still images and moving pictures.

Digital imaging systems are used both in private and in industrial applications. For example, in the consumer sector, digital cameras are available as photo or video cameras, and they are also integrated into mobile phones and laptop computers. In industry, automatically-triggered digital cameras are being used, for example, for quality control in automated assembly lines, in police radar traps, and automatic license plate recognition in security-related applications.

A specific application of modem digital imaging systems is in the area of document imaging, one example of a document imaging system is the office scanner. Recent years have seen a drastic increase in the speed and quality of the scanning process, where automatic feeders, high resolution and colour depth have contributed to ease of use and high quality of the result. Data generated during the scanning process may be sent to a printer, if a hard copy of the original is required, and/or to an external computer, from which the scanned data might be e-mailed.

Document imaging, however, is not limited to the scanning of office material. Systems have, for example, been developed that automatically image security-related documents including, but not limited to: travel documents, such as passports, ID cards, visas; identity documents, such as passports, ID cards, driver's licenses; and fiduciary documents, such as bank books, cheques and banknotes; and other so-called value or valuable documents. These systems may be found, for example, at land borders, airports, or sea ports, in banks or at reception desks. Albeit often smaller in format, these systems often provide illumination in predominantly infrared, visible (i.e. white) and ultraviolet wavelength intervals, or may require retroreflective light to be viewed.

Regardless of the environment in which they are used, imaging systems will often pick up dip or dust from various sources including the ambient air, the skin of the operators, and, in case of document imaging systems, from the documents themselves. Such contamination may result in artefacts in the captured image data, and have a negative impact on any further outputting or processing of this image. Operating manuals often ask for periodic cleaning of the system, but in reality this is often neglected, and sometimes systems are operated without cleaning until they fail completely.

Imaging systems may also have optical defects that arise during their use. For example, metal clips or staples on documents may, over time, scratch the imaging window of a scanner, or jewellery worn by an operator may scratch the window of a scanner. Optical elements in the optical path of an imaging system may even break when the imaging system is exposed to mechanical shock or sudden temperature fluctuations, which may cause cracks in the optical elements. The deformation of an optical element (for example, a slight curvature of an imaging window) can also be considered to be an optical defect.

Efforts have been made in the past to detect faults, i.e. contamination or optical defects, in the optical path of imaging systems automatically, so that the operator -or a service person -may receive notice that a system requires cleaning or repair. An example of a system to measure cleanliness of a glass panel is found in US patent application U52005/0 170097 (PPG), wherein a template is placed behind the glass panel, the glass panel and template are optically scanned, the scan data fed into a computer, and the evaluation for cleanliness is based on the scanned image of the glass panel and the template.

European Patent Application EP 1667434 (Nikon) refers to an image-capturing diagnostic device, that obtains an image of a uniform surface (a reference image), photographed through an optical system with a variable aperture, wherein the device may comprise a monitoring unit that monitors foreign matter present in an optical path, based upon the image.

The use of templates or reference images is not a preferred way of detecting contamination of an imaging system, as this detection process requires an interruption of the sequence of imaging processes in order to image the reference image. A fixed schedule for imaging a reference image may be established, but the operator may be tempted to skip the step of imaging the reference image when workload is high. A further issue is that reference documents must be kept clean and calibrated, which means that to be used in real life they must be kept carefully, which may not S always be possible. With a fixed schedule for imaging the reference image, imaging of the reference image may be done too frequently, i.e. dirt has not accumulated yet, or too rarely, i.e. dirt has accumulated and affected the imaging results. In both cases, operator time is unnecessarily spent, in the former case for running the dirt detection process too early, in the latter case for correcting the effects of bad imaging results.

European patent application EP 1596568 (Xerox) mentions a streak detection method in a scanning device, comprising a step of redundantly scanning an image with at least one full colour spectrum channel set of sensor rows, and at least one clear channel sensor row. Further steps comprise integrating image data from the full colour channel set of rows to obtain an estimate of clear channel image data, and comparing the estimate to actual image data from the clear channel row of sensors, thus deriving clear channel error data. Presence of non-image data is detected by comparing the derived clear channel error data to a predetermined threshold. This document is only concerned however with systems and methods for detecting dirt particles, paper fibres and contaminants in the field of view of an imaging array in digital scanning systems.

Utilizing a further row of sensors in an imaging system to detect contamination is associated with considerable extra cost and presents an additional element of ambiguity, since a sensor in the further row of sensors may fail and lead to unjustified error messages. The use of an additional sensor would also require provision of additional equipment or hardware, making it difficult to employ such a method in existing equipment.

There is therefore a need to be able to detect such faults in imaging devices, such as scanners, reliably and repeatedly, and with minimal or no additional equipment needed to ensure full back compatibility with existing systems.

The present invention aims to address these issues by providing a method of detecting a fault on an imaging device, comprising: (a) capturing an image of an article viewable through an imaging device, the imaging device being divided into one or more cells, each cell being assigned to a portion of the image; (b) processing the image to determine whether any of the cells contains computer-recognisable data; (c) if a cell does contain computer-recognisable data, analysing the computer-recognisable data to determine whether it is recognisable or not to a pre-determined probability of success; and (d) maintaining a first count, for each cell, of whether the computer-recognisable data contained in that cell is recognised and a second count for whether it is not; and (e) repeating steps (a) to (d) for a plurality of articles and generating an error signal when a threshold based on the first and second counts is reached during a fixed number of successive repetitions of steps (a) to (d).

Preferably the imaging device is a scanner and comprises a scanner window.

The present invention also provides a method of detecting a fault on a scanner window, comprising: (a) capturing an image of an article viewable through a scanner window, the scanner window being divided into one or more cells, each cell being assigned to a portion of the image; (b) Processing the image to determine whether any of the cells contains computer-recognisable data; (c) if a cell does contain computer-recognisable data, analysing the computer-recognisable data to determine whether it is recognisable or not to a pre-determined probability of success; and (d) maintaining a first count, for each cell, of whether the computerrecognisable data contained in that cell is recognised and a second count for whether it is not; and (e) repeating steps (a) to (d) for a plurality of articles and generating an error signal when a threshold based on the first and second counts is reached during a fixed number of successive repetitions of steps (a) to(d).

By utilising a count of whether data has been recognised or not it is possible to determine from the number of times data is not recognised over a series of repeated activations of an individual cell it is possible to determine remotely whether a fault is present on the imaging device or scanner window. This enables the user of the device to have confidence in the quality of data retrieval and for maintenance to be carried out easily. Such a method can also be combined easily with existing protocols and software within such devices.

The threshold may be based on a running average of the first count and a running average of the second count. Alternatively the threshold may be based on a historical average of the first count and a historical average of the second count. As a second alternative, the threshold may be based on a time average of the first count and a time average of the second count. As a third alternative, the threshold may be based on a ratio of the first count to the second count.

The method may further comprise the step of storing the number of error signals generated in a memory.

The cells are preferably of equal size. Alternatively, cells of a first size are assigned to a first type of computer recognisable data, and cells of a second size are assigned to a second type of computer recognisable data, the first and second types of computer recognisable data appearing in different positions on the article.

Preferably, the article is an identity document, a travel document or fiduciary document.

Preferably the computer recognisable data is one of: a character, image object or symbol. If this is the case, preferably the character is in a style acceptable under ISO and/or ICAO standards.

The present invention also provides a computer program that when run on a computer causes the computer to perform the method steps outlined above.

The present invention will now be described by way of example only, and with reference to the accompanying drawings, in which: Figure 1 a is a schematic representation of a typical passport biodata page used as a travel document; Figure lb is a schematic representation of the front side of a typical identity card used as an identity document; Figure lc is a schematic representation of the rear side of a typical identity card used as an identity document; Figure 2 is a schematic diagram showing a scanner used to image a passport biodata page or identity card; Figure 3 is a schematic representation of a scanner window divided into a number of equally sized cells; Figure 4 shows two lines of OCR B, typically found in an MRZ, un-obscured; Figure 5 shows two lines of OCR B, typically found in an MRZ, part of which is obscured; Figure 6 is a flow chart of the steps used to increment the counters in each cell as required; and Figure 7 is a schematic representation of a scanner window divided into a number of different sized cells.

The present invention adopts the approach that a software-based method can be used to detect whether there is a fault present on an imaging device, such as a scanner window. A computer program can be loaded into an existing system as a patch or an upgrade, and utilises components already present within an existing device or system.

Figure 1 a is a schematic representation of a typical passport biodata page used as a travel document. The biodata page 1 is provided with regions of text 2, indicating personalisation data comprising details of the holder's name, place of birth, nationality, passport number, expiry date, and issue information, data headings 3, an image 4 of the holder and a region of text known as a machine readable zone or MRZ S containing characters in OCR-B font as stipulated by both ISO and ICAO standards. The biodata page 1 may be illuminated using visible, infrared (IR), ultraviolet (UV) light or retroreflective light, depending on whether it is desired to see overt (seen under visible light) or covert (seen under IR, UV light or retroreflective light) features.

Figure la is an illustration of how the biodata page 1 appears when viewed under visible light.

In travel documents conforming to ICAO and ISO standards, the personalisation information 2, headings 3, photograph 4 and MRZ S are provided in similar regions of the biodatapage 1, with the version shown in Figure 1 being a general example. In terms of covert security features, such as ultraviolet sensitive characters, information about the issuing authority contained in the MRZ S can be used to access data indicating where such covert security features are printed and what they consist of for a particular type of travel document, such as a passport. Figure lb is a schematic representation of the front side of a typical identity card used as an identity document, and Figure 1 c is a schematic representation of the rear side of a typical identity card used as an identity document. The identity card 6 is also provided with regions of text 7 indicating personalisation data comprising details of the holder's name, nationality and issue details and an image 8 of the holder on the front side, and an MRZ S on the rear side. The identity card 6 may be illuminated using visible, infrared (IR), ultraviolet (liv) light or retroreflective light, depending on whether it is desired to see overt (seen under visible light) or covert (seen under IR LV light or retroreflective light) features. Each of the text images and MRZ S text forms computer recognisable data in the form of characters, image objects (such as pictures, photographs, patterns or drawings) or symbols, that can be recognised by a scanner forming part of an imaging system typically used to read travel and identity documents. Characters and symbols may be formed from any language or typeface or font that may be scanned and pattern matched as described below.

Figure 2 shows an imaging system in accordance with the present invention. The imaging system 9 comprises a scanner 10, having an optical image CMOS (complementary metal-oxide semiconductor) sensor 11, and a computer 12. The scanner 10 has a housing 13, which is generally box-like, having first and second opposing rectangular faces forming the top and bottom of the housing 13. A scanner window 14 is provided on the top of the housing 13, such that a biodata page 1 or identity card 6 may be placed in contact with the scanner window 14 in order for an image of the page 1 to be captured. A first set of light-emitting diodes (LED5) iSa is adapted to illuminate the biodata page 1 with predominantly infrared light; a second set of LEDs is adapted to 1 Sb illuminate the biodata page 1 with predominantly visible light; and a third set of LEDs is adapted to 1 Sc illuminate the biodata page 1 with predominantly ultraviolet light. Ahhough not shown, an arrangement of minors may be included if desired to allow illumination under retroreflective light.

Light reflected by the biodata page 1 (and illustrated by ray trace lines Ri, R2, R3) is captured by the CMOS sensor ii after travelling through an optical element 16 comprising a multi-element glass lens to focus the reflected light onto the CMOS sensor ii. To ensure that the housing is made with as small a footprint as possible, light focused by the optical element 16 is reflected onto the surface of the CMOS sensor 11 by a flat minor 17. The scanner 10 also comprises an auxiliary electronics unit 18, comprising an analogue-to-digital converter and a shift register, within the housing 13, to convert the electrical signals of the CMOS sensor ii into a raw digital image data, and an image processing unit 19 to organise and assemble the raw digital image data into an image data format readable by the computer 12. The scanner 10 is connected to the computer 12 by means of cabling 20.

The computer 12 is provided with computer software enabling it to carry out various image processing tasks to be able to recognise the computer recognisable data within the image data captured by and transferred from the scanner 10. A data recognition processing means 21 is provided as part of this software, operable to recognise computer recognisable data within all or part of the image data. The CMOS sensor 11 comprises an array of sensor cells in an individual sensor cell array of 2048 x 1536 cells, giving a pixel density of approximately 3,000,000 pixels with each single sensor cell approximating to one pixel. This creates a one-to-one mapping between sensor cells and pixels.

To enable the software to function the computer 12 is provided with: a first counter 22a for maintaining a count of each time computer recognisable data is recognised to a pre-determined probability of success; a second counter 22b for maintaining a count of each time computer recognisable data is not recognised to a pre-determined probability of success; and an error signal generator 23. A section 24 of the computer memory 25 is assigned to store data representing the status of each counter 22a, 22b, and the signal generator 23 is linked to the computer 12 to provide a visible alarm when an error is generated. The data recognition processing means 21 is the same as that used in the imaging system 9 before the software is installed.

Figure 3 is a schematic representation of a scanner window divided into a number of equally sized cells. The scanner window 14 is divided by the software into a plurality of equally sized cells as part of the initial set up of the imaging system 9. In this example, each cell comprises an array 200 x 75 sensor cells, which is physically approximately the same size as a group of five characters and associated spacings within the MRZ 5 region of the biodata page 1. In this example, each cell comprises the same number of sensor cells regardless of the position it occupies on the sensor window 14 and its relative position with data provided on the biodata page 1.

The operation of the software comprising a computer program that when run on the computer 12 causes the computer to execute a number of method steps will now be described using a biodata page 1 as an example. Initially the software is loaded onto the computer 12 either by means of transferring from a computer program product, such as a disc, or by downloading from a secure server. During set-up of the software the cell size is pre-selected along with an appropriate threshold for error generation. The first 22a and second 22b counters for each cell are set to zero. In the following example, the threshold chosen is based on a running average of a first count of each time the computer recognisable data is recognised to a pre-determined probability of success maintained by the first counter 22a, in conjunction with a running average of a first count of each time at the computer recognisable data is not recognised to a pre-determined probability of success maintained by the first counter 22b.

In order to process the image data, each part. of the image representing a letter or symbol is recognised using a probability of success based on known image processing and pattem matching techniques, which are not part of this invention. Based on the highest probability calculated, the data recognition processing means returns a match for the letter or symbol in the image data.

This can be illustrated as follows. Figure 4 shows two lines of OCR B, typically found in an MRZ, un-obscured. The upper line contains the text < < D E B, and the lower line U T L 7 8. In the case where there is no fault or defect in the scanner window 14, the data recognition processing means 21 retums <<D E B for the upper line U T L 7 8 for the lower line. This is achieved by calculating the probabilities of success for each of the letters and symbols using known image processing and pattern matching techniques, and outputting the letter or symbol with the highest probability of a match in each position.

Figure 5 shows two lines of OCR B, typically found in an MRZ, part of which is obscured. The same set of letters and symbols is illustrated with an obscuration over the letter D. This may be caused by a defect or error on the scanner window 14. In this situation there is a potential error in the letter/symbol returned at this position, due to the fault. There is now doubt over the unique identity of the character in the image data. The data recognition processing means 21 will obtain a very low confidence score for the D as is has been distorted.

With this process in mind, each time a biodata page 1 is scanned by the imaging system 9 the following takes place in at least one cell. Figure 6 is a flow chart of the steps used to increment the counters in each cell as required. Each of steps 102, 104 106 and 108 is carried out per cell.

Each time a biodata page 1 is scanned computer recognisable data may not be present in every single cell. Cells in the region where the MRZ is typically positioned on the scanner window 14 may image computer recognisable data each time a biodata page 1 is positioned on the scanner window 14. Cells corresponding to other regions of the biodata page 1 may not image computer recognisable data each time a biodata page 1 is placed on the scanner window 14. The counters for each cell are therefore incremented each time that cell is activated, that is images computer recognisable data, rather than each time a biodata page 1 is scanned.

At step 100, the imaging system 9 recognises that a biodata page 1 has been placed on the scanner window 14. This allows the capturing an image of the biodata page 1 viewable through the scanner window 14, where, as discussed above, the scanner window is divided into one or more cells (in this case a plurality of cells), and each cell being assigned to a portion of the image. At step 102, the imaging system 9 scans the biodata page 1, processing the image to determine whether any of the cells contains computer-recognisable data. In this example, for a cell n in the region of the scanner window 14 corresponding to part of the MRZ S of the biodata page 1, computer recognisable data is imaged. At step 104, the image processing means 21 performs the probability of success calculation outlined above to determine whether it is recognisable or not to a pre-determined probability of success. A successful determination of the identity of the computer recognisable data is made if the computer recognisable data is recognised to a pre-determined probability of success -"if match = yes". In this case at step 106 the first counter 22a is incremented by one, forming the first count. A successful determination of the identity of the computer recognisable data is not made if the computer recognisable data is not recognised to a pre-determined probability of success -"if match = no". In this case at step 108 second counter 22b is incremented by one, forming the second count. These steps are repeated for a plurality of biodata pages 1. Furthermore, each of steps 102 to 108 are repeated for each biodata page 1 imaged in each cell where computer recognisable data is found.

To determine whether the error signal generator 23 needs to be activated a threshold based on the first and second counts needs to be reached during a fixed number of successive repetitions of steps 100 to 108. In this example a running average based on the first count is compared with a running average of the second count, based on a threshold being set for a successive number of repetitions of steps 100 to 108 outlined above. For twenty successive activations of one cell, the threshold is set as being if ten or more have incremented the second counter 22b (the counter is at ten or higher) the scanner window 14 is classed as containing a fault. The use of such a running average reduces the likelihood that a biodata page 1 error can cause a false error to be generated.

A second check is to provide criteria to determine if a biodata page is faulty. An MRZ 5 typically contains 88 characters. To overcome issues with the signal-to-noise ratio making a statistically significant number of faulty characters difficult to determine, if ten or more (11%) of those characters in a single document return an increased count for the computer recognisable data is not recognised to a pre-determined probability of success the biodata page 1 is classed as faulty.

As an alternative to keeping the counters 22a, 22b incremented over the last 20 activations of a cell a buffer of data representing the last 100 characters read can be kept. This buffer is constantly updated every time a cell is activated. This then enables a historical average to be used instead of the running average to determine if the error signal generator needs to be activated. Consequently the threshold is based on a historical average of the first count and a historical average of the second count. The threshold set is then whether more than 50% of the last 100 characters read have caused the second counter 22b to be incremented, and if so, the scanner window 14 is faulty. By adding an extra step into the flow chart shown in Figure 6, step 110, data relating to the first and second counts can be stored in the memory of the computer 12.

This may then be used with information relating to where a cell n is located on the scanner window to determine a time-based average, where the number of times the second counter 22b is incremented for a cell over a set time period, for example, three months, is counted.

Consequently the threshold is based on a time-based average of the first count and a time-based average of the second count. The threshold is set that if 50% of the activations of that cell over three months increment the second counter 22b, the scanner window 14 is faulty. A further alternative is to count the number of times the cell is activated, and then to use a first count to determine whether it has been read, and a second count to determine whether it has not been read, thus creating a running average of the activation and read data. Data relating to the physical proximity of cells where errors are generated may also be useful to understand the type of fauh that has been detected, or to discount faulty biodata pages 1, as with the signal-to-noise considerations outlined above.

In the above example, a plurality of equally-sized cells was used in conjunction with a running average. However, the scanner window 14 may be divided into one or more cells, such as a large single cell or a plurality of cells as described here. For two or more cells therefore equally sized cells may be chosen. However, for two or more cells it may be desirable to choose cells of different sizes, depending on where they lie on the scanner window 14 in relation to the information contained on the biodata page 1. The size of the cells is determined by the need to have a statistically significant indicator of whether a biodata page 1 is faulty at that position, to ensure that the signal generated in imaging in that cell is higher than the signal-to-noise ratio.

Figure 7 a schematic representation of a scanner window divided into a number of different sized cells. Cells of a first size are assigned to a first type of computer recognisable data, and cells of a second size are assigned to a second type of computer recognisable data, the first and second types of computer recognisable data appearing in different positions on the article. In Figure 7, cells in the region of scanner window 14 where the MRZ 5 would be positioned are small in size, whereas those in the region of the scanner window 14 where the image 4 is positioned, are larger in size. A typical OCR-B character can be contained within 200 x 75 sensor cells, whereas an image may be contained within 400x 600 sensor cells. Sensor cell arrays forming individual cells need not be rectangular in shape, but may be square, for example, 200 x 200 sensor cells.

Monitoring the increment of the first 22a and second 22b counters for adjacent cells may also be an indicator that the scanner window 14 is faulty. Data for cell n may be used in conjunction with data for cell n+1 and cell n-i. Again this information is compared with that used to determine whether the biodata page 1 is faulty before the error signal generator 23 is activated.

The above example is concerned with the type of imaging system 9 employed as a page reader in travel document reader systems, such as passport reader systems. However, the software may be used in conjunction with a swipe reader, typically used for reading only the MRZ 5 of a biodata page. Typical swipe readers, such as the RTE 6700 and RTE 6701 passport swipe readers, are available from 3M in the United Kingdom (3M Rochford Thompson (a trading name of 3M UK plc), The Votec Centre, Hambridge Lane, Newbury, Berkshire, RG14 5Th, United Kingdom).

Compared with Figure 2, the CMOS sensor 11 is replaced with a line scan CMOS sensor, again with pixels grouped to form sets of sensor cells in vertically spaced horizontally oriented groups.

The imaging system 9 shown in Figure 2 may be part of a static system or may be embodied within a mobile system. The scanner window 14 may be provided in various positions to enable flexibility in the configuration of the imaging system 9, such as on the top, front, side or bottom of the scanner 10. To create a mobile system certain of the various components shown in Figure 2 (the data processing recognition means 19, the first 22a and second 22b counters) may be placed in a server accessed wirelessly by the mobile system. Even within a static system it may be desirable to have certain components present on a client terminal, adjacent to or forming part of the scanner 10, or on a remote server accessed when necessary. The visible alarm generated by the error signal generator 23 may be a signal such as the illumination of an LED, a pop-up box, a network signal, a message to a maintenance or call centre, an email, text or SMS message, for example, and may be complimented by an audible alarm..

Although the invention is intended for use in relation to imaging systems used primarily to automatically image security-related documents including, but not limited to, passports, ID cards, drivers' licenses, visas; fiduciary documents, such as bank books, cheques and banknotes; and other so-called value or valuable documents, it may be applied to other imaging devices, such as cameras, copiers and scanners. A fault may be present in or on such an imaging device in the same way as on a scanner window, and require detection to enable the imaging device to be cleaned and/or checked as necessary.

Claims (14)

1. Claims 1. Method of detecting a fault on an imaging device, comprising: (a) Capturing an image of an article viewable through an imaging device, the imaging device being divided into one or more cells, each cell being assigned to a portion of the image; (b) Processing the image to determine whether any of the cells contains computer-recognisable data; (c) If a cell does contain computer-recognisable data, analysing the computer-recognisable data to determine whether it is recognisable or not to a pre-determined probability of success; and (d) Maintaining a first count, for each cell, of whether the computer-recognisable data contained in that cell is recognised and a second count for whether it is not; (e) repeating steps (a) to (d) for a plurality of articles and generating an error signal when a threshold based on the first and second counts is reached during a fixed number of successive repetitions of steps (a) to (d).
2. 2. Method of claim 1, wherein the imaging device is a scanner and comprises a scanner window.
3. 3. Method of detecting a fault on a scanner window, comprising: (a) Capturing an image of an article viewable through a scanner window, the scanner window being divided into one or more cells, each cell being assigned to a portion of the image; (b) Processing the image to determine whether any of the cells contains computer-recognisable data; (c) If a cell does contain computer-recognisable data, analysing the computer-recognisable data to determine whether it is recognisable or not to a pre-determined probability of success; and (d) Maintaining a first count, for each cell, of whether the computer-recognisable data contained in that cell is recognised and a second count for whether it is not; (e) repeating steps (a) to (d) for a plurality of articles and generating an error signal when a threshold based on the first and second counts is reached during a fixed number of successive repetitions of steps (a) to (d).
4. 4. Method of any of claims 1 to 3, wherein the threshold is based on a running average of the first count and a running average of the second count.
5. 5. Method of any of claims 1 to 3, wherein the threshold is based on a historical average of the first count and a historical average of the second count.
6. 6. Method of any of claims 1 to 3, wherein the threshold is based on a time average of the first count and a time average of the second count.
7. 7. Method of any of claims 1 to 3, wherein the threshold is based on a ratio of the first count to the second count.
8. 8. Method according to any of claims 1 to 7, further comprising the step of storing the first and second counts in a memory.
9. 9. Method of any of claims 1 to 8, wherein there are at least two cells and the cells are of equal size.
10. 10. Method of any of claims 1 to 8, wherein there are at least two cells, and cells of a first size are assigned to a first type of computer recognisable data, and cells of a second size are assigned to a second type of computer recognisable data, the first and second types of computer recognisable data appearing in different positions on the article.
11. 11. Method of any of claims 1 to 10, wherein the article is an identity document, a travel document or a fiduciary document.
12. 12. Method of any of claims 1 to 11, wherein the computer recognisable data is one of: a character, a symbol or an image object.
13. 13. Method of claim 12, wherein the character is in a style acceptable under ISO and/or ICAO standards.
14. 14. A computer program that when run on a computer causes the computer to perform the method steps of any of claims 1 to 13.