The invention relates to a method for monitoring documents, for example documents of value such as banknotes.
Document monitoring is used in a number of different fields to obtain information about different characteristics of documents, particularly documents being fed along a transport path. These characteristics include the document condition, size and, in the case of documents of value, characteristics such as authenticity and denomination. In order to determine these characteristics, different information must be obtained from the document and traditionally the document is fed past a number of is different detectors and processing environments to enable each aspect or characteristic to be determined. This is inefficient in terms of cost and space required and limits the extent of information which can be obtained.
Documents of value need to be inspected at least to determine their classification, for example their denomination and to confirm authenticity. One approach is to compare images of part or parts of the document under test obtained under predetermined illumination conditions with sets of predetermined images and determine which predetermined image most closely matches the image under test. However, the matching process is complex and thus can take a significant time while it is desirable to increase the speed of processing of documents of value, for example in banknote sorters and counters.
In accordance with the present invention, a method of inspecting documents of value comprises
In this invention, we provide a two stage inspection process. In particular, following the analysis in step b), the analysis of step c) is performed using only data relating to members of the first type of class identified in step b). In general, this will constitute a much smaller number of sets of data, typically just one, so speeding up the matching process.
The wavelength bands may be overlapping or non-overlapping, both in the visible or both in the non-visible spectra or, preferably, one in the visible and one in the non-visible e.g. infrared or ultraviolet.
Typically, the first type of class defines one or more of the denomination, face and orientation of the document of value. In this case, preferably the image used in step b) is obtained from visible radiation received from the document.
The second type of class may define the authenticity of the document of value. In this case, the image used in step c) may be defined by radiation received from the document in a waveband outside the visible wavelength range, for example the infrared wavelength range.
It should be understood, however, that these class types could be reversed so that initially an authenticity classification is carried out followed by a denomination or other classification.
As an alternative, the second type of class may define the degree of soiling of the document of value.
In a further, preferred case, the method further comprises
The second and third types of class can then be authenticity and degree of soiling respectively. Steps c) and d) could be carried out in parallel or sequentially and in one case, the predetermined data used in step d) could relate only to members of the second type of class identified in step c).
The images can be derived from different areas of the document since the discriminative and/or authenticating regions of the document may well be in different places. The data however is usually gathered at the same time.
The images can be obtained using conventional equipment such as described in EP-A-0660277, GB-A-2107911 and GB-A-1470737.
Preferably, however, step (a) is performed using a document monitoring system comprising illumination apparatus for illuminating a region, at which part of a document is located in use, with radiation in at least two different wavelength bands; and detection apparatus for detecting any of said radiation which is reflected by or transmitted through substantially the same said part of the document and for generating corresponding output signals.
This apparatus allows information to be obtained from a document by both reflection and transmission. By irradiating in at least two different wavelength bands, information relating to both authentication and classification can be achieved while reflected and transmitted radiation provides information about soiling and other conditions.
The illumination apparatus may generate the radiation at the at least two different wavelength bands simultaneously, the detection apparatus having separate detectors responsive to radiation in the different wavelength bands. Alternatively, the illumination apparatus may generate radiation in the different wavelength bands sequentially, the detection apparatus having one or more detectors responsive to radiation in all the wavelength bands.
In the preferred example, the illumination apparatus includes a radiation source located on the opposite side of the region to the detection apparatus for transmitting radiation through the document. However, as an alternative, the illumination apparatus may include a reflective surface in the said region and on which a document is located in use, wherein any radiation impinging on the reflective surface is reflected towards the detection apparatus.
In some cases, the system may include a single set of illumination apparatus and detection apparatus and the region may be chosen in accordance with the document to be monitored. Thus, the region could comprise a small region within the document or a region extending across the full dimension of the document. In the preferred case where the document is fed through the region, the region preferably extends across the full dimension of the document transverse to the feed direction.
In other cases, more than one set of detection and illumination apparatus may be provided to monitor different regions of the document.
The invention is particularly concerned with inspecting documents of value such as banknotes but is also applicable to visas, passports, licences, cheques, identity cards, plastic cards, bank notes, tickets, bonds, share certificates, vouchers, passes, permits, brand authentication labels, serial numbering slips, quality control certificates, bills of lading and other shipping documentation, legal documents and tamper evident labels and the like.
Some examples of methods and apparatus according to the invention will now be described with reference to the accompanying drawings, in which:
As mentioned above, the invention can be used in a wide variety of apparatus and in this particular example we will describe its application to banknote handling apparatus, for example a banknote sorter. In this sorter, the banknotes are fed from a stack (not shown) to a feed system 1 (
The sources 12,13 are arranged such that light will be reflected/scattered from the surface of the document 14 back to the detector(s) 16. No direct light path exists between the illumination sources and the detector(s). Dependant upon the choice of detector 16, the illumination sources 12,13 can either be broad band polychromatic devices, generating illumination from the IR and visible band (eg fluorescent tubes with appropriate coatings) or, can be collections of monochromatic sources (eg LEDs) that are modulated to provide pulses of light spread across the IR and visible spectra.
The detector(s) 16 could be a set of narrow band detectors for detecting radiation in respective wavebands, or a broadband receiver. In the latter case, the sources 12,13 will need to be modulated on and off to ensure that only one frequency or frequency band of light is illuminating the note at any one time. The modulation would be controlled from a processor or a piece of hardware that would successively turn on each illumination source whilst turning off each of the others.
The transmissive, visible radiation source 5 is positioned such that it is directly opposite the detector(s) 16. The transmissive source 5 can either be monochromatic or polychromatic to match the form of detector 16 but does not need to have any output in the IR.
The relative brightnesses of the sources 12,13 and the source 5 are arranged such that the image received by the detector(s) 16, when a document is present, is primarily created from the light reflected from the document, thus creating visible and IR reflective images of the document. The transmissive source 5 does not provide any substantial contribution to the received light when a document is present. However, when no document is present the transmissive source is set to be brighter than any reflected image.
Although two sources 12,13 are shown in
The detector(s) 16 shown in
The use of polychromatic sources requires one detector for each point in the spectrum of interest e.g. red, green, blue and IR for colour and IR imaging, or visible and IR for grey scale and IR imaging.
The use of multiple modulated illumination sources 5,12,13 allows the use of a single broadband detector as mentioned above.
The spectra in the example above is not meant to convey any limit to the range of usable spectra, the device can be constructed to work in any areas of the electromagnetic spectrum providing suitable illumination sources and receptors are available.
Before a document arrives at the detector system 3 (
The image thus created across all pixels of the detector is therefore one that has a very bright (or even saturated) background (from the transmissive illumination) within which is a reflective image of the document. The reflective image contains bright (or saturated) regions created by the transmissive illumination shining through any holes in the document. An illustration of this is shown in
Further, it is important to realise that the detector head will be creating multiple images of each document, one for each illumination or reception spectra. The minimum for a gray scale and IR device will thus be two images and for a red, green, blue, IR device four images will be created. The number of images being equal to the number of points in the spectrum being used.
The or each detector element within the detector(s) 16 generates an analogue signal which is fed to the A/D convertor 6 for conversion into digital form, the digital signals then being fed to the microprocessor 7. The microprocessor 7 stores the digital signals in a conventional manner so as to define an image for each of the received wavebands.
Each image created by the detector(s) 16 can be either separately processed or processed in conjunction with other images. A typical process sequence for a grey scale and IR device is shown in
The incoming stream is the raw data arriving from the detector head and is initially segregated 30 to form images 31,32 for each point in the spectrum.
After this, visible image 32 is processed 33 in a conventional manner to determine the skew, this process also involving finding the edges of the document. The skew and positional information is then used 34 to de-skew and position all images such that they have a common coordination system with the templates that are used later in the processing 35,36.
The de-skewed visible image 36 is then processed 37 in order to determine its classification 43 by reference to predetermined templates 39 stored in a store 40. This comprises the denomination (the face value for a banknote) the face, (the top or bottom image) and the orientation (which way up the document is). A double sided document, such as a banknote, has one denomination, two faces and two orientations, a total of four classes. The determination of this is usually carried out on a normalised image 38 that has had the contrast varying effects of wear and print variations removed. The recognition means applied to the document, such as double threshold correlation using templates 48 are well known and will not be disussed here since the choice of algorithm will depend upon many factors such as image resolution and whether or not the image is from all or part of the document. A typical example is the technique used on the De La Rue 2700 Banknote Counter.
Once the document classification has been determined, this is then used to inform and reduce the processing requirements remaining. It should also be noted that whilst the processing up to this point has been of a serial nature the remaining processing can be carried out either serially or in parallel depending upon the choice of processing environment.
The authentication process 42 is carried out on the de-skewed IR image 35 and is essentially a comparison between the acquired image and a single template 41 stored in the store 40, for the class 43 of document being examined. The single template 41 is the one indicated by the classification of the visible image. The use of the visible classification 43 to select the template 41 significantly reduces the processing required and allows features that may not distinguish two different classes to still be useful in the authentication decision. The simultaneous existence of the visible and IR images will also allow verification that a feature found in the IR domain for authentication also exists in the appropriate form in the visible domain. A version of this device that produces colour images may also be able to check the colour of the specific features. An example of an authenticatable feature is a portrait where the printing inks are arranged such that they appear to be a single colour when viewed in visible light but when viewed in the infra-red split into reflective and absorbing blocks.
The remaining processes 44 establish the condition (fitness) of the document and take place on the visible image 36. It should be noted that all of the processes for condition detection are performed on an image that has not been normalised.
The establishment of the transmissive illumination to be such that all holes 21 etc within the document become bright spots (brighter than the reflective image could ever reach) provides a means by which all such defects in the document can be determined using a suitable threshold. (See
Similarly the use of edge tracing algorithms can be applied to the image to establish the position and size of any folds around the document edges. These algorithms establish the line equations for each portion of the document edge and then establish the intersection point of each line. The folds can be found by checking for orthogonality of the lines and by aligning the image within the appropriate rectangle for the document as informed by its class.
It is well known that as a document (especially a banknote) is handled the dirt build up tends to evenly cover it making the white areas less white and the dark areas less dark, in other words the overall contrast of the image is reduced. The level of dirt on the document (Soil level) is also determined from the visible image 36 and achieves this by a two pass process on the image. Again the process 46 is informed from the already established classification 43. The first pass is an overall comparison of the image with its template 48 for variations in the contrast of the image. This will give an overall measure and will also detect any large stains that may be present. The second pass uses the appropriate template 47 to just examine the unprinted areas of the document to determine its absolute brightness relative to its template. The template will have been developed from a mint note. The combination of these two measures will be used to report the overall soil level.
The templates 41,47,48 will typically be discriminant functions as used in the De La Rue 2700 Banknote Counter. The IR templates would be a binary image of the note face showing the areas containing the controlled IR features. The soil templates would be grey scale image of the note showing the contrast levels of a clean bank note. It would particularly illustrate areas of un-printed paper.
Number | Date | Country | Kind |
---|---|---|---|
0001561 | Jan 2000 | GB | national |
Filing Document | Filing Date | Country | Kind | 371c Date |
---|---|---|---|---|
PCT/GB01/00254 | 1/23/2001 | WO | 00 | 8/14/2002 |
Publishing Document | Publishing Date | Country | Kind |
---|---|---|---|
WO01/54076 | 7/26/2001 | WO | A |
Number | Name | Date | Kind |
---|---|---|---|
3679314 | Musteri | Jul 1972 | A |
4723072 | Naruse | Feb 1988 | A |
4875589 | Lacey et al. | Oct 1989 | A |
5498879 | De Man | Mar 1996 | A |
6040584 | Liu et al. | Mar 2000 | A |
6269169 | Funk et al. | Jul 2001 | B1 |
6661910 | Jones et al. | Dec 2003 | B2 |
Number | Date | Country |
---|---|---|
198 40 482 | Mar 2000 | DE |
0 660 277 | Jun 1995 | EP |
2 107 911 | May 1983 | GB |
WO 8502928 | Jul 1985 | WO |
WO 9909382 | Feb 1999 | WO |
Number | Date | Country | |
---|---|---|---|
20030030785 A1 | Feb 2003 | US |