This invention relates to the reading of codes on a package as it passes along a conveyor. In particular, the invention may be used for the aggregation of unique cigarette pack codes into a bundle.

It is increasingly important for valuable or safety related items such as pharmaceuticals or tobacco products to be able track articles through manufacture and onwards to sale. Thus as packages are aggregated and wrapped during packaging for example from small packages into small groups of packages and then into cases and ultimately into pallets, the labelling needs to reflect the contents accurately at each layer.

This is achieved by reading codes of items that are being aggregated and then producing an aggregated code to be placed on the wrapping of the grouped items. This must have cross-checks to ensure that the items packed together are accurately reflected in the codes on the outer layers of packaging.

The accuracy of the code reading of the individual packets is thus important, as any inaccurate codes cause an aggregated pack to be rejected. This then results in inefficiencies in the packing line as the rejected packages must then be analysed for the cause of the error and manually corrected. In the prior art, code reading is carried out using a single camera as a package passes along the conveyor. This invention aims to improve the accuracy of the prior art arrangements in order to reduce the volume of rejected aggregated packages.

In a first aspect, the invention provides a code reader for a conveyor carrying items having a code extending along the direction of the conveyor movement, comprising at least two image readers arranged to read different parts of a code at different times as an item passes along the conveyor, and also arranged to read at least some of the same parts of the code at different times as the item passes along the conveyor whereby two images of at least part of the code are available for error correction in the code reading.

The invention will now be described by way of example with reference to the drawings in which:- Figure 1 shows schematically, codes which are gradually obscured as extra wrapping is placed over them;

Figure 2 shows a typical camera and lighting arrangement for reading the codes of Figure 1 ;

Figure 3 shows a typical conveyor arrangement for the present invention;

Figure 4 shows schematically how the cameras read the code as it passes along the conveyor; and

Figure 5 shows the code reading in more detail.

With reference to Figure 1 , codes on the individual items are typically printed in human and machine readable form as shown in the image marked 2A.

When the package is wrapped, typically in a transparent material, this becomes slightly more obscured and open to reflections as shown in image 2B. When these packages are themselves grouped together and wrapped again, the code becomes even more obscured as shown in Figure 2C. Thus as packages are aggregated into yet larger groups, the problem of reading codes accurately becomes ever harder.

Figure 2 shows a typical arrangement for reading codes in which dispersed illumination is used to try to reduce reflections from the packaging caused by layers of overwrapping film. Red LED lights 4 direct light to a reflector 6 which then directs multiple rays towards the part 8 which is passing along the conveyor. An opening in the reflector 6 allows a code reading apparatus, typically a camera 10, to receive an image of the part 8 illuminated by the multiple reflected incidences of red light. In this way, direct reflections into the code reader/camera 10 are largely avoided. Nevertheless, since it will be noted that the wrapping film is not necessary perfectly smooth or flat, it is still possible for some reflections to obscure the code from the camera 10.

Accordingly, an improved arrangement is described below which helps improve the accuracy of code reading and avoid the problem described above, of rejected packages. With reference to Figure 3, a conveyor 20 carries items (not shown) past a reading apparatus 22. The reading apparatus has 3 cameras 24A, 24B, 24C aligned along the direction of movement of conveyor 20 with illumination (not shown) into a reflector hood 30. Opposite the cameras, triggers 26A, 26B, 26C recognise the passing, typically of a leading edge, but possibly instead a trailing edge, of an item.

With reference also to Figure 4, a notional code 28 is shown passing along the conveyor 20 from left to right, at times t1 , t2 and t3. The optical triggers are triggered as items pass along the conveyor. In the preferred embodiment, at trigger 26A, camera 24A is activated to read the first two segments of the code 28. At trigger 26B, all three cameras 24A, 24B and 24C are activated to read the entire code 28. At trigger point 26C the second 2 cameras 24B and 24C are triggered to read the rear three portions of the code 28.

In this way, when all three cameras are triggered, an entire image of the code is obtained and this, in most instances, will be enough. However, if some errors are found and the code cannot be read properly, trigger points 26A and 26C provide a second image of the entire code which may well have enough additional data to reconstruct the code accurately and thereby avoid a rejected package.

The general principle is to trigger a plurality of cameras at different times, to provide duplicated images of the code, but these duplicate images are not necessarily all entire images, but rather may be made from leading or trailing portions which reflect the moving nature of the items and an economic balance between numbers of code readers which must be provided along the passage of the items. It would of course be possible to provide two separate reading heads to read the entire code twice, but the present arrangement takes a different path and avoids the need for complete duplication of the reading heads by instead providing multiple and selective triggering of the reader heads as the same item passes. This reduces the number of cameras required which reduces costs and complexity. Error correction

The system is designed to read pack codes in bundles with, typically, up to 10 packs. Although the system also works for bundles with other numbers of packs such as 8 or 6 packs, the majority of applications are for bundles containing 10 packs.

The codes on all the packs in the bundle are read when the second trigger causes all three cameras to read. This has the effect of generating the "Primary List" of codes for the bundle. The first trigger reads causes the first camera to operate and reads the first four codes. The third trigger causes the second and third cameras to operate which reads the last six codes. The combination of codes read from the first and third trigger are grouped into the "Secondary list".

If a code has not been read in the "Primary list", the "Secondary list" is looked at and if it has been read its value is used. If the code does not exist in either primary or secondary list then no code is recorded. If the code in the primary list is different from the corresponding code in the secondary list then no code is recorded. When no code is recorded, the pack is rejected.

Figure 5 shows this similar process in more detail in which master pack codes and secondary pack codes for the smaller internal items are aggregated into a single code which is read in the way described above in connection with Figure 4. Again, it will be seen that the first trigger allows reading of the first two segments, i.e. codes 1 to 4, the second trigger triggers all 3 cameras which allows reading of all the codes 1 to 10 and then the third trigger causes triggering of cameras 2 and 3 (24B and 24C) which allows reading of codes 5 to 10. Thus all codes 1 to 10 are read twice.

As a further enchantment, a master list of codes may be prepared using a separate code reader, typically located upstream of the aggregating machinery. Use of a separate remote camera to generate a Master List

To improve aggregation efficiency further, a separate standalone camera may be placed at an earlier point in the production process. This camera can be mounted as far back as the creation of the codes or just before the packs enter the previous machine. Indeed, if there is feedback from the printing device regarding which codes have been printed there may be no need of a camera at all.

The purpose of the Master List is to provide a comprehensive list of codes processed before any pack wrapping takes place or at a position where although already wrapped it is easier to read the codes. It is important that even if the camera fails to read a code, a blank is inserted into the Master list, to maintain the correct order and spacing of codes relative to real packs passing through the system.. When the aggregation of the codes take place, and a code is missed, the Master List is checked to determine whether the missing code can be filled in from the data in the Master List.

It is possible for the codes to become out of sequence, because packs have not moved through the system in the same sequence, and so checking rules are applied to ensure that the correct code is filled from the Master List. Unless the rules are satisfied, no fill in takes place.

Typically, the first check is determine that the codes either side of the missed code are the same via both systems. This enables a single missed code to be filled in. For the second check, if the first and last code in the bundle have been read, then this, together with the positions of the other read codes, is compared with the Master List. As long as all read codes are found in the same positions within the Master list, then the missing codes are filled in. Separate system checks are also made to ensure that a filled in code has not been used previously and if so the corresponding bundle is rejected from the line.

In a cigarette pack application, the system reads a "DotCode" on a cigarette pack and then aggregates the codes together within a bundle and links them to label that is applied to the bundle.

The bundle is wrapped and a pair of driven rollers slows down the wrapped bundle and ensures that a gap exists between them. Each gapped wrapped bundle passes the three camera array and lightbox. The three sensors 26A 26B, 26C trigger the cameras to ensure that each pack code is read twice as described above. The wrapped bundle is labelled and a Unique Identification code is incorporated into a 2D DataMatrix code applied to the wrapped bundle.

After labelling, a confirmation camera (not shown) confirms that the correct wrapped bundle is associated with the wrapped bundle 2D code. If aggregation has failed or validation has failed then the wrapped bundle is rejected into a bin. The reasons for reject can then be analysed.

In this way, the system takes an image of the code several times, typically, twice, with two separate cameras and if one camera fails to read accurately, the reading from the second camera image is used. In this way although each camera may only be, e.g, 99.9% efficient, the overall reading efficiency could be at 99.99% - this is a significant improvement and a significant advantage as every bundle with a missing code must be rejected and re-processed.

The system works dynamically as the wrapped bundles pass by, and it is this movement of the wrapped bundle past the cameras, as opposed to the wrapped bundle being static in front of a camera, as in the prior art, that enables taking of at least two reads. It should be noted that an arrangement using an indexer in which the packs are temporarily halted just to take an image is intended to be encompassed by the present description and claims. For example, it would be possible to index the wrapped bundle three times and take the image whilst the wrapped bundle is stationary or a combination of both i.e. dynamic first image (the first 4 codes), holding the wrapped bundle static for the second (all 10 codes) image and then indexing again dynamically for the last image (6 codes)

In Summary:- Method of Operation

Three separate triggers. Each trigger activates camera and lighting

1 st trigger activates first camera which reads the first 4 codes

2nd trigger activates all three cameras which reads all 10 codes

3rd trigger activates last two cameras which reads the last 6 codes

Each code is read twice by a separate camera helping to increase overall reading efficiency.