WO2002073545A1

WO2002073545A1 - Item identifying apparatus and method

Info

Publication number: WO2002073545A1
Application number: PCT/GB2002/001201
Authority: WO
Inventors: Raglan Horatio Andrew Harold Tribe; Phillip Michael William Ireland; Anthony Lloyd Robinson; Barry Scowen
Original assignee: De La Rue International Limited
Priority date: 2001-03-14
Filing date: 2002-03-14
Publication date: 2002-09-19
Also published as: GB0106288D0

Abstract

Item identifying apparatus comprises one or more detectors (50) for obtaining data defining characteristics of items; and a processor (50) for monitoring the data obtained by the detector(s) and for performing a first identification process (510) using the data to identify an item, wherein when the first identification process does not identify the item with sufficient confidence, the processor performs at least one second identification process (550) using the said data from the detector(s), the item being identified if the identity generated by the second identification process or processes has a sufficient level of confidence.

Description

ITEM IDENTIFYING APPARATUS AND METHOD

The invention relates to item identifying apparatus and a method of identifying items. The invention is particularly concerned with identifying items of value such as coins or banknotes but is also applicable to a wide variety of other items.

Cash handling machines that can recognise the value of a particular note or coin need to draw an appropriate balance between achieving a correct classification and rejecting currency or coins as unrecognisable. The primary aim is to ensure misclassification is at a low enough level but high enough to ensure an acceptable reject rate.

Many identification processes have been developed for identifying items, particularly items of value such as coins and banknotes. In the case of banknotes, these can involve determining the size of the banknote and/or part or all of a pattern carried on the banknote. This data is then compared with predetermined, master data and if the comparison indicates a level of confidence exceeding a threshold then the item is identified as corresponding to the master pattern with the highest correlation.

Problems arise, however, where it is desired to achieve rapid throughput of items and yet high confidence levels of identification. Complex matching algorithms have been developed but these take time to operate and thus can reduce throughput rates .

EP-A-0101115 describes a banknote pattern recognition system in which one half of a banknote is analysed and if the results are dubious or negative then the other half of the banknote is analysed, if desired with a different correlation threshold. This suffers from the problems mentioned above in that it requires that different parts of the banknote are used for the two different comparisons. EP-A-0805408 describes a pattern recognition process in which a first test is carried out to limit the number of possible patterns which could correspond to the banknote under test following which a second pattern recognition process is carried out using the preselected subset of patterns. It is therefore necessary to utilize both processes to arrive at a result which again is undesirable in view of the time taken. US-A-4179685 has a similar disclosure.

DE-A-2723078 describes a system for authenticating banknotes. If an initial test suggests that the banknote is counterfeit then it is transferred back to a detector for renewed testing and clearly this requirement to refeed the note is time consuming and undesirable.

In accordance with a first aspect of the present invention, item identifying apparatus comprises one or more detectors for obtaining data defining characteristics of items; and a processor for monitoring the data obtained by the detector (s) and for performing a first identification process using the data to identify an item, wherein when the first identification process does not identify the item with sufficient confidence, the processor performs at least one second identification process using the said data from the detector (s), the item being identified if the identity generated by the second identification process or processes has a sufficient level of confidence.

In accordance with a second aspect of the present invention, a method of identifying items comprises obtaining data from an item and performing a first identification process on the item using the data to identify the item; and, when the first identification process does not identify the item with sufficient confidence, performing at least one second identification process using the said data obtained from the item, the item being identified if the identity generated by the second identification process or processes has a sufficient level of confidence. We have developed a new method and apparatus which can enable high item throughput to be achieved but with small risk of misclassification or identification. This is achieved by performing a first identification process, which can be of any conventional type, but which can also work rapidly, and only if that process is unable to identify an item with sufficient confidence, carrying out at least one second identification process. This second process or processes may take more time to process the data than the first process but since it should not be needed frequently, this does not adversely affect the overall throughput rate. Furthermore, this second identification process is carried out on the same data as used by the first identification process which increases overall processing speed. In addition, each identification process is complete in its own right. The second does not rely on results from the first. Typically, when a set of gathered data is compared with a stored "reference" a "score" is calculated to represent the degree of fit between the gathered and reference data. The threshold or confidence level that is then applied, to judge whether the data is acceptable, is set to provide a statistical measure of the likelihood of the document being correctly identified. That is, it is set high enough to limit the number of false attributions

(known as "cross-over" when talking about denomination discrimination) but low enough to prevent a high number of rejections. Thus in this invention, each identification process attempts to identify the item with a level of confidence consistent with the minimum level of false attribution of identity, to confirm a particular level of product performance . Typically, a single second identification process will be performed. However, more than one could be performed, the item being identified from the results of the last or a combination of two or more of the first and second processes. Examples of the way in which the second identification process can be utilized are as follows: 1) The original data could be used to better effect. In many pattern recognition systems, the data is often sub-sampled to reduce the processing overhead. Thus, for the second identification process, in a correlation based classification system a second set of discriminant function templates could be trained using the full resolution rather than the sub- sampled image. The initial sub-sampling could have been area or grey scale based.

2) The quality of de-skewing a captured image to enable a match with a Master Pattern Template could be improved by using more accurate techniques using floating-point calculations for the trigonometric functions.

3) Extended hierarchy which performs additional checks on the classification carried out to see if the classification is valid or attempt to classify a note or coin that is indeterminate. 4) Use a completely different type of algorithm, for example the first identification process uses a discriminant function as in WO-A-00/26861 while the second identification process uses an analogue representation of the note or coin (grey scale) or binary vectored images.

5) Improve the correlation quality by finding the best match between template and image captured by repeated small movement in x, y and z, also known as sliding correlation. 6) Improved feature selection, for a note this could consist of accurately locating the print image using edge finding techniques.

7) In note counters that purely use size to determine value then a best-fit parallelogram technique could be used to improve the multiple measurement of length and width of a note. This has the benefit of providing some degree of resistance to note misclassification (denomination) due to note damage. 8) In a coin sorter, a best circle, arc or B-spline fit could be undertaken to provide a more accurate view of coin diameter which was originally determined by optical sensor array obscuration.

In the preferred examples, the first and second identification processes are of different types, for example different mathematical processes on the data at the same or different resolution. Alternatively, the same mathematical process could be performed on the data but at a higher resolution during the second identification process.

Some examples of banknote handling apparatus and methods according to the invention will now be described with reference to the accompanying drawings, in which :-

Figure 1 is a schematic diagram showing the primary transport components of a first example;

Figure 2 is a view similar to Figure 1 but of a second example; and,

Figure 3 is a flow diagram illustrating the identification procedure used in each apparatus. Figure 1 illustrates a banknote counter 100 having an input hopper 2 mounted beneath an inlet opening 3 in an enclosure 1 which comprises upper and lower parts la, lb normally screwed together. Contained within the enclosure 1 is an internal chassis assembly (not shown for clarity) which itself has side members between which the sheet feeding and transport components to be described herein, are mounted. Two conventional feed wheels 5 are non-rotatably mounted on a shaft 7, which is rotatably mounted to the chassis assembly, and have radially outwardly projecting bosses 6 which, as the feed wheels rotate, periodically protrude through slots in the base of the hopper 2. A pair of stripper wheels 15 are non-rotatably mounted on a drive shaft 16 which is rotatably mounted in the chassis assembly. Each stripper wheel 15 has an insert 17 of rubber in its peripheral surface. Shaft 16 is driven clockwise by a motor (not shown) to feed notes individually from the bottom of a stack of notes placed in the hopper 2. Transversely in alignment with, and driven from the circumferential peripheral surface of the stripper wheels 15, are pressure rollers 30 which are rotatably mounted on shafts 31 spring based towards the stripper wheels 15. Downstream of the wheels 15 is a pair of transport rollers 19 non-rotatably mounted on a shaft 20 rotatably mounted in the chassis assembly. Shaft 20 is driven clockwise from a second motor (not shown) to transport the note in the transport arrangement, in conjunction with pairs of pinch rollers 21,23 into stacking wheels 27 and hence output hopper 105. Pinch rollers 21, rotatably mounted on shafts 22 spring based towards the transport rollers 19, transversely align with rollers 19 and are driven by the peripheral surface of the rollers 19. The rollers 23, rotatably mounted on shafts 24 are in alignment with the transport rollers 19, and are essentially caused to rotate by the note passing between the adjacent peripheral surfaces of the rollers 19 and 23. Situated between the pressure rollers 30 and pinch rollers 21 are separator roller pair 25, non-rotatably mounted on shaft 26 adjustably fixed to a top moulding assembly 32, having a circumferential peripheral surface which is nominally in alignment with the peripheral circumferential surface of, but transversely separated from, the stripper wheels 15.

Also forming part of the top moulding assembly 32, is a curved guide surface 8 extending partly around the circumference of the rollers 15, 19 which, when the top moulding is lifted allows the operator access to the note feed and transport path so that a note jam can be cleared. A surface 37 provides note guiding from the end of the curved guide surface 8 to the conventional stacking wheels 27.

The drive shaft 16 is continuously driven, and this, via a belt and pulley arrangement from shaft 16, causes the auxiliary drive shaft 7 rotating the feed wheel 5 also to be driven. Drive shaft 20, rotating the transport rollers 19, is driven by the other drive motor. A further pulley and belt arrangement (not shown) between shaft 20 and shaft 28, on which the stacking wheels 27 are non-rotatably mounted, provides the drive to the stacking wheels 27.

A guide plate 9 extends as a continuation of the base of the hopper 2 towards the nips formed between the transport rollers 19 and the rollers 23.

A pattern recognition system 50 including a linear photodiode array is mounted adjacent to the transport path. The array extends across the full length of the banknotes (transverse to the feed direction) , so as to detect light originating with a light source 50A transmitted through the banknotes as they pass beneath the detector. (Other known detectors could be used which, for example, only scan a portion or portions of the banknotes. Also, a reflective system could be used, as an alternative or in addition, to the transmissive system.) The system 50 includes an onboard processor which samples the photodiode outputs regularly.

Sampled photodiode output signals are digitized and stored and then compared with a set of prestored images . Where the counter 100 is to count a single denomination then the prestored images will correspond to the appearance of that denomination in all possible orientations. If the counter 100 is to form a "value balancing" count of mixed currency then the prestored images will define the appearance of each currency in all its orientations.

In addition, the pattern recognition system will also determine whether or not the note is of the expected denomination (in the case of single denomination counting) or one of the expected denominations or . currency or denomination series (in the case of mixed denomination counting) and will increment a count accordingly. The count will typically be a count of the people value of the notes being counted although in some cases it could simply represent the number of notes .

If the microprocessor in the pattern recognition system 50 determines an unacceptable condition such as a corrupted image, an indefinable image or the like then it will cause one or both of the drive motors to stop so that the unacceptable note or notes will be stacked at or near the top of the stack formed in the output hopper 105. The entire stack can then be refed or the operator, in some cases, can adjust the count either to decrement it if the notes are to be removed or to increment it as necessary if the notes are, in fact, acceptable.

The counter 100 shown in Figure 1 has a single output hopper 105. The invention is also applicable, however, to counters/sorters having multiple output hoppers and Figure 2 illustrates such an example with two output hoppers. The Figure 2 counter 300 has an input hopper 401 having a base

402 with an aperture 403, through which a high friction portion 404 of a nudger wheel 405 can project. The base 402 optionally has a second aperture 406 in alignment with a barcode reader 407 for reading data on note separators. Bank notes are supported in a stack on the base 402 against a front wall 426, and are fed intermittently by rotation of the nudger roller 405 into a nip 408, between a high friction feed roller 409 and a separator, counter rotating roller 410. The nudger 405 and roller 409 are driven by a motor 200 (not shown) . The documents pass through pinch rollers 411, 412 into a pattern detection region 413 in which a sensor of a transmission pattern recognition system 414A,414B (414B indicating a radiation source) scans the bank note as it is fed and passes information back to a microprocessor of the system 414A (not shown) . Each bank note is then fed through pinch rollers 416, 417 onto a drive belt 418 which conveys the bank note around various rollers 419 to a diverter 420. At least one of the rollers is driven by a motor (not shown) . The position of the diverter 420 is controlled by the microprocessor of the system 414A, so that bank notes are guided either towards an output pocket 421, where they are stacked using a rotating stacking wheel 422 in a conventional manner, or to a reject bin 423.

As can be seen, the bank notes are stacked on the base 402 and are urged forward against the front wall 426. A small gap 427 is provided at the base of the front wall, through which individual bank notes and separators can be nudged.

The pattern recognition system 414A,414B operates on the detected image data in an exactly similar way to the pattern recognition system 50 of the previous example. In this case, however, instead of stopping the transport when an unsatisfactory condition is. determined such as a double note feed or the like, the diverter 420 is operated so that the unacceptable notes are fed to the reject bin 43. The operation of the pattern recognition systems 50,414A will now be described with reference to Figure 3. Initially, in step 500, the note is presented to the corresponding detector. The detector will obtain data from the note, as previously described, and typically this will be sub-sampled and then compared with one or more master patterns using a conventional comparison algorithm. If the degree of correlation between the data from the note and one of the master patterns exceeds a predetermined threshold indicating a sufficient level of confidence in the identification (step 510) , the note is identified as being that denomination (step 520) and it will be forwarded and processed as an identified note.

If, however, the denomination cannot be determined with sufficient confidence then drive to the feed wheels 5 (Figure 1 example) or nudger wheel 405 (Figure 2 example) will be stopped (step 530) . A second identification algorithm known as "second look" is then run (step 540) and typically this will involve looking at the pre-sampled data or in other words data at a higher resolution using the same, but preferably a different, pattern matching process. Other combinations of different or the same mathematical process and the same or different resolution data could be used as described above. The important point is that no more data needs to be obtained from the note. If the result of this second process results in a level of correlation which provides a sufficiently high confidence level to identify the note denomination (step 550) then the drive to the wheels 5,405 is restarted (560) and the note continues to be processed in a conventional manner. If its denomination cannot be determined then the note is rejected

(step 570) , again in a conventional manner. In the case of the Figure 1 example, this would involve starting the transport for a short while to feed the note to the top of the stack and then stopping the transport to allow the note to be extracted by the operator. In the Figure 2 example, the note would be fed to the reject bin 423. In the latter case, in most circumstances, drive to the wheel 405 would also be started.

Suitable comparison algorithms are well known in the art, one example being described in WO-A-00/26861.

Claims

1. Item identifying apparatus comprising one or more detectors for obtaining data defining characteristics of

5. items; and a processor for monitoring the data obtained by the detector (s) and for performing a first identification process using the data to identify an item, wherein when the first identification process does not identify the item with sufficient confidence, the processor performs at least 0 one second identification process using the said data from the detector (s), the item being identified if the identity generated by the second identification process or processes has a sufficient level of confidence.

2. Apparatus according to claim 1, wherein the processor 5 is adapted to implement the second identification process using data obtained from the detector (s) at a resolution higher than that used in the first identification process.

3. Apparatus according to any of the preceding claims, wherein the second identification process performs more 0 tests on the data than the first identification process.

4. Apparatus according to any of the preceding claims, wherein the second identification process is of a different type from the first identification process.

5. Apparatus according to any of the preceding claims, 5 wherein the identification processes are chosen from one or more of pattern and size matching processes.

6. Apparatus according to any of the preceding claims, wherein the second identification process is carried out on data obtained from one or more predetermined regions of the 0 item.

7. Apparatus according to any of the preceding claims, further comprising a transport system for conveying items past the detector (s).

8. Apparatus according to claim 7, wherein the processor 5 is adapted to stop the supply of items to the transport system while performing the second identification process.

9. Apparatus according to any of the preceding claims, wherein the processor performs only one second identification process.

10. Apparatus according to any of the preceding claims, the apparatus being adapted to handle items of value, such as coins and banknotes .

11. A method of identifying items, the method comprising obtaining data from an item and performing a first identification process on the item using the data to identify the item; and, when the first identification process does not identify the item with sufficient confidence, performing at least one second identification process using the said data obtained from the item, the item being identified if the identity generated by the second identification process or processes has a sufficient level of confidence.

12. A method according to claim 11, wherein the data used by the second identification process is at a higher resolution than the data used by the first identification process.

13. A method according to claim 11 or claim 12, wherein the second identification process performs more tests on the data than the first identification process.

14. A method according to any of claims 11 to 13, wherein the second identification process is of a different type from the first identification process.

15. A method according to any of claims 11 to 14, wherein the identification processes are chosen from one or more of pattern and size matching processes.

16. A method according to any of claims 11 to 15, wherein the second identification process is carried out on data obtained from one or more predetermined regions of the item.

17. A method according to any of claims 11 to 16, wherein the second identification process is carried out only once.

18. A method according to any of claims 11 to 14, wherein the items comprise items of value such as coins and banknotes .