System and method for tracking items

Abstract

A system and method for identifying a package comprises capturing an initial scan of the exterior of an initial package and receiving an associated initial package record S11. An initial fingerprint i

Description

Field of the invention

[0001] The present invention relates to a system and method for identifying a package. Background and the problem to be solved

[0002] Usually, when a parcel or envelope (hereafter package) is sent via a package processor network, such as the Royal Mail in the UK, proof that payment has been made for the use of that package processor network’s services is evidenced via a physical mark e.g. a franking mark for business post or paper-based stamp (e.g. a first-class stamp) for privately sent mail. Equally, private courier services will use some method of identification, such as a label including a bar code attached to the package, to confirm that payment has been made before accepting a package. Any service that provides some form of package handling or processing desires a method of confirming that payment has been made.

[0003] The use of a payment identifier can be inconvenient to the end user, requiring a franking machine or printer to be owned or a stamp to be sourced before a package can be sent. There is also a risk of the mark being damaged or detached from the package if not properly secured. Finally, only packages with a surface suitable for attachment of a stamp or frank can be used. The surface of the package must be smooth and receptive to ink or glue. Summary

[0004] It is an aim of the present invention to provide a system and method of package identification that is more accessible and convenient to the user.

[0005] According to an aspect of the invention, there is provided a method for identifying a package comprising: capturing an initial scan of the exterior of an initial package and receiving an associated initial package record, creating an initial fingerprint comprising information derived from the initial scan and an associated initial package record and storing the initial fingerprint, initial scan and the associated initial package record in a datastore that may be configured to store one or more initial fingerprints, one or more initial scans and one or more associated initial package records, capturing a tracking scan of the exterior of a tracked package in order to retrieve its initial associated package record, creating a tracking fingerprint representing the tracked package, comparing the tracking fingerprint to the one or more fingerprints stored within the datastore, and returning the associated initial package record of the initial package if the tracking fingerprint matches one or more of the initial fingerprints stored within the datastore.

[0006] The present invention enables a package processor network to verify that a payment for delivery has been made without requiring any stamp, frank mark or any other physical representation of payment on the package being sent. The present invention (sometimes referred to as NoStamp) aims to establish proof of payment as the Package progresses through the package processing system without the use of a payment identifier.

[0007] The “NoStamp” sender app enables postage fee payment for use of a package processor network to be made using only a mobile phone app; proof of payment does not require any stamp, frank mark or any other physical representation of payment on the package or envelope being sent. The NoStamp sender app captures an ‘initial scan’ (which may include video, still images, LIDAR, data about the scanning device such as its orientation etc) of the package usually before the package is put into the package processing network. The app then saves that ’initial scan’ in a datastore, together with data that may include extracted ‘fingerprint’ information derived from the scan as well as associated data such as payment information. Later, as the package progresses through the package processing network, proof of payment is established using the datastore information about the package that is identified in the datastore from a new ‘tracking scan’ that is taken at the time the package is being processed. 10008] Thus, a method is provided that allows for a package to be identified using a scan taken of the package at any point in its journey from before posting, to being delivered to the final recipient and even after delivery if required. Relevant information about the package and its scan may be stored in a datastore and the associated package record can also be accessed and used to assist with the routing and sorting of the package, for example by allowing for the location of the package to be tracked and relayed to the package sender / recipient as the package is transported. The package can be identified without the need for a physical mark, such as a stamp or barcode, providing a method that is far more convenient to use compared to alternative methods of making payment for a package to use a Package Processing Network.

[0009] The system further allows for a package to be sent without a physical symbol of payment attached, for example a stamp, a written identifier code or a barcode. Instead, the package can be identified at any point using another scan (a “tracking scan”) to generate a “tracking fingerprint” which is then compared with the initial fingerprint for that package 2 that exists in the datastore. Once a package is identified, a check can be made to ensure that payment has been received for sending the package. The package can therefore be sent without a proof of payment, such as a stamp, attached. The payment information of the package can instead be verified using the scan and fingerprint system to ensure that the correct payment has been made.

[0010] In an embodiment the initial and tracked packages comprise package text information positioned on the package surface, and both the initial and tracking fingerprints further comprise initial and tracking package text data respectively. The method further includes extracting the package text information from the initial and tracking scans and storing the package text information within the initial and tracking fingerprints. Optionally the package text information is address information. The package fingerprint information can then be readily accessed when later required, such as for routing purposes (if it is delivery information) or for use in comparing two fingerprints to determine if they were derived from the same package.

[0011] In an embodiment, both the initial and tracking fingerprints further comprise bounding box data comprising the position, size and orientation of the bounding box around each letter and word of the package text information positioned on the package surface and the position, size and orientation of the bounding box around each line of the package text information positioned on the package surface. The bounding box data provides generally unique information about the location, orientation and sizing of the package text information that may be positioned on the package surface; this, combined with the package text information used to derive the bounding boxes, is able to uniquely identify the package. The bounding box data stored in the initial and tracking fingerprints may be compared to determine whether the two fingerprints were derived from the same package.

[0012] In an embodiment, comparing the initial fingerprint with the tracking fingerprint involves comparing the diagonal lengths of each bounding box of the initial fingerprint with the bounding box for the equivalent text in the tracking fingerprint. The diagonal length of a bounding box is a straightforward value to calculate, while also accurately representing the shape of the bounding box. An efficient comparison can therefore be made between the bounding boxes of each fingerprint to identify whether the two fingerprints correspond. The diagonals may be computed when the initial fingerprint is created, rather than when the comparison is made. Computing the diagonals in advance allows for the amount of computation required by the package processor network to be 3 reduced, increasing the speed at which it can process packages. When a tracking fingerprint is created by the package processing network, the bounding boxes are again calculated to compare with the bounding boxes of the initial fingerprint.

[0013] In an embodiment, both the initial and tracking fingerprints further comprise image data, wherein the image data is a digital image representation of the initial or tracked package and its unique features identified from a scan. In an embodiment, comparing the initial fingerprint with a tracking fingerprint involves comparing the initial scan’s image data and the tracking scan’s image data. The image data derived from the scan provides an alternative method of identifying packages and comparing fingerprints. The image data derived from the scan is unique to a specific package, with a scan of the same package usually producing multiple, slightly different image data from the frames of the scan’s video data. Two different packages that use a printed label with the same text in the same position, and which appear to be identical, could be identified as different packages through use of the scan derived image data. 10014] In an embodiment, the method further includes receiving a user provided unambiguous digital package delivery text for the initial package, and wherein the associated package record comprises the digital package delivery text for the initial package. When packages are later sorted, software is used to read the package text address information located on the package. This software may sometimes not be able to parse the package delivery address text accurately enough to sort the package, requiring a manually sorting process. If the package has been identified, the package delivery text (if it is delivery information) stored within the associated package record can instead be used to sort the package, thus increasing the efficiency of the sorting process. The package can be identified even when the package delivery text cannot be parsed correctly.

[0015] In an embodiment, the method further includes alerting the sender when the address information cannot be located from a list of known addresses or cannot be successfully parsed and extracted into ASCII or UNICODE formatted data by AICR. The sender of the package can then be alerted that there may be an issue with the address. For example, the address may be misspelt or may no longer exist and the sender can be prompted to enter a digital, ‘unambiguous’ address.

[0016] In an embodiment, the associated initial package record includes the payment status of the initial package. Including the payment status of the package in the associated records allows for the payment status of the package to be quickly checked once the package is identified.

[0017] In an embodiment, the method further includes receiving the weight of the initial package, and wherein the initial fingerprint further comprises the weight of the initial package. In an embodiment, the method further includes receiving the weight of the tracked package, the tracking fingerprint further comprises the weight of the tracked package and the comparison further includes comparing the weight of the initial package and the weight of the tracked package. Including the weight of the package in the comparison allows for an additional identifier to be used when comparing the packages, thereby allowing for a more accurate comparison between two fingerprints.

[0018] In an embodiment, the initial and tracking scans are one or more of videos, one or multiple still images, LIDAR data showing the package exterior, or other data showing the package exterior. In an embodiment, the initial and tracking scans are videos showing the exterior of the initial and tracked package respectively, and the method further comprises converting the video into a sequence of still digital images, wherein the package text information text is identified using text fragments consistent across each of the still digital images in the sequence of still digital images derived from the initial and tracking scans. When just a single image is used to identify the package text information, there is greater chance of an error occurring. The use of a series of still images allows for the package text information to be analysed multiple times. The series of still images may also show the package from different angles. The text that is most consistently identified can then be used to identify the package text information. The use of multiple images allows for a more consistent identification of the package text information text when compared to the use of a single image.

[0019] In an embodiment, capturing the initial scan and tracking scan includes capturing an initial and tracking set of pose information about the position of the scan devices used to capture the initial and tracking scans, and wherein this initial and tracking set of pose information is stored in the initial and tracking fingerprints respectively, once the fingerprints are created. In an embodiment, the initial and tracking scans are normalised using the initial and tracking pose information respectively, wherein normalising the initial and tracking scans involves adjusting the images from the scans based on the location and orientation of the scan device.

[0020] When two scans are produced using different devices operated by different users (e.g. the initial scan and the later tracking scan), there may be a difference between the scans produced by the two - for example, the devices may be at different position or in a different rotation. Normalising the scan data through the use of the pose information 5 reduces the variation in data derived from the scans produced. Reducing the variation between scans allows for a more consistent fingerprint to be produced for a package. Providing a reduction in variation reduces or avoids the generation of fingerprints that cannot be matched for a package.

[0021] In an embodiment, the datastore is remotely located on a cloud platform, at a datacentre or on a server capable of persistently storing information. The use of a remote datastore allows for the fingerprint of a package to be readily accessed from multiple locations, regardless of where the scan of a later fingerprint is captured, allowing for the package to be identified using the system at multiple locations. List of figures

[0022] Embodiments of the disclosure will now be further described, merely by way of example, with reference to the accompanying drawings.

[0023] Fig. 1 is a flow chart illustrating the method steps taken when performing the method.

[0024] Fig. 2 shows an example of package text information (which is delivery information in this particular case, but does not have to be) positioned on the surface of the package, as well as the bounding boxes positioned around each word of the package text information, and the extracted address.

[0025] Fig. 3 is an example of printed package text information.

[0026] Fig. 4a is an example of good handwritten package text information.

[0027] Fig. 4b is an example of good handwritten package text information with bounding boxes positioned around each word of the package text information.

[0028] Fig. 5 is an example of bad handwritten package text information.

[0029] Fig. 6 is a flow chart showing the overall process of sending a package according to an embodiment of the method of the present invention.

[0030] Fig. 7a shows an overview of the data structures produced by an initial scan

[0031] Fig. 7b shows an overview of the process of searching the datastore for an initial scan corresponding to a tracking scan. Detailed description

[0032] The NoStamp app is an application that can be downloaded onto a capable device, such as a computer, tablet or mobile phone and can be used to pay for package processing 6 services without using a physical payment identifier, such as a paper stamp, on the package being sent. NoStamp also uses software that a package processing network (e.g. the Royal Mail in the UK) can use to identify and process packages that have been paid for (or not) by the No Stamp app.

[0033] The method used to identify packages will be described in reference to the flow chart of Figure 1. The first stage of the method Sil involves capturing an initial scan of the exterior of a package and receiving an associated initial fingerprint and package record. The initial scan may be captured using a mobile phone camera for example. A mobile phone app may be used to receive the associated initial package record or it may be received in a cloud or similar platform. A scan may be one or more images, videos, LIDAR, or similar digital representations of a package, taken by a scanning device. The associated initial package record may be received from the user, and contain information relating to the package. For example, the associated initial package record may contain a payment identification number unique to that package that can be used to verify whether a payment has been made for the package.

[0034] A scan of the package is how its address and other information are extracted and processed. An “interactive” user-interface may be used that helps the user correctly position the camera over the package when scanning e.g. using visual prompts to surround key information such as the address, and overlaying the address text that has been “extracted” from the scan as text on the screen.

[0035] The next stage of the method involves creating an initial fingerprint comprising information derived from the initial scan and the associated initial package record and storing the initial fingerprint in a datastore configured to store one or more initial fingerprints. Every initial fingerprint corresponds to a specific package such that a different scan (e.g. a tracking scan) of the same package will provide a fingerprint that can be used to uniquely identify the scanned package’s associated initial package record in the datastore.

[0036] The datastore may be located locally on the app device or, more likely, on a cloud platform or technology of similar functionality. The technology underpinning the datastore may be a SQL database, document or any other suitable data storage and retrieval technology such as Postgres, Mongo, SQL Server, but a globally distributed, massively scalable technology like Azure Cosmos DB could be a very good fit.

[0037] The datastore may be backed by additional information processing services such as cloud computing, APIs, security services, monitoring, logging, failover, redundancy with scaling capabilities.

[0038] The initial fingerprint may be created locally on the scanning device, for example a mobile phone app may be used to capture an initial scan and then create the initial fingerprint, alternatively the initial scan may be sent to a remote location, such as a cloud platform, for transformation into the initial fingerprint. The newly created initial fingerprint may then be transmitted to a datastore for later searching and retrieval. Processing of the initial scan may involve notifying an event grid process of the scan, use of a cloud function or some other computation method to create and store the initial fingerprint derived from the initial scan.

[0039] In some arrangements, the package comprises package text information 20 positioned on the package surface. The package text information 20 may be handwritten text 22, 23 or printed text 21 and may identify the address that the package should be delivered to. Both the initial and tracking fingerprints may further comprise package text data. The method further includes extracting the package text 20 into a digital form from the initial and tracking scans and storing the package text information 20 within the initial and tracking fingerprints.

[0040] In some arrangements, both the initial and tracking fingerprints further comprise bounding box data comprising the position, size and orientation of the bounding box 24 around each word of the package text information 20 positioned on the package surface and the position, size and orientation of the bounding box 24 around each line of the package text information positioned on the package surface.

[0041] For handwritten text 22, 23 or printed text 21, the fingerprint may include all of the bounding boxes 24 for the text positioned on the package, and the results of calculating the diagonal length of each individual bounding box 24, and / or the diagonal length of the bounding box 24 for the entire text. The ratio of the diagonal of the bounding box 24 for the entire text to the diagonal of the bounding box 24 of the largest word or largest phrase may be calculated and form part of or the entirety of the initial or tracking fingerprint. A bounding box 24 for a line of text is defined as the position and size of the smallest rectangle able to fit all parts of the line of text within it.

[0042] Figure. 4A shows an example of good handwritten text 22. Figure. 4B shows the same good handwritten text 22 with the bounding boxes 24 shown. The bounding box information identifies the position and orientation of each word (e.g. the words “Taylor’ 8 and “slow”). The bounding boxes 24 surround each word. The position and orientation of each bounding box 24 therefore provides information about the position and rotation of each word in the package text information. There may additionally be bounding box information for each line of text e.g. “Taylor slow”, or for the package text information as a whole. These additional bounding boxes 24 are not shown in Figure. 4b.

[0043] Fig. 5 shows an example of bad quality handwritten text 23. The A ICR (Al character recognition) confidence level in handwritten text is typically much lower than that of printed text 21, but can still be high for clearly identifiable handwriting, even when the handwriting is viewed at an angle (such as a 90-degree rotation). For bad handwritten text 23, confidence is normally low or very low. An understanding of the format and semantics of addresses can be used to assist with the text extraction process.

[0044] For printed text 21, the creation of the initial and tracking fingerprints may include extracting all words from the package text information 30 at very high accuracy. The extracted words may then be stored within the initial and tracking fingerprints. The extracted words contained within the initial and tracking fingerprints can be compared to determine whether two fingerprints were created from initial and tracking scans of the same package.

[0045] The initial and tracking fingerprints may also include one or more of: image information extracted from the scan, the initial scan data, the sender information, recipient information, information about the computer equipment used to create the scan, sensor data from the device used to create the scan (e.g. pose information used to normalise the images), or any other payload data, which can comprise any additional information that an application requires.

[0046] The next stage of the method S13 involves capturing a tracking scan of the exterior of a tracked package that the user - who may be the package processing network -wants to identify. The tracking scan may or may not have a corresponding initial scan in the datastore. The tracking scan may be performed using a different device than the device that was used to create the initial scan.

[0047] The next stage of the method S14 involves creating a tracking fingerprint from the tracking scan of a package. This tracking fingerprint is used to identify whether the package has a matching initial scan stored in the datastore.

[0048] The next stage of the method S15 comprises comparing the tracking fingerprint to the one or more initial fingerprints stored within the datastore. Comparing an initial fingerprint with the tracking fingerprint may involve comparing the diagonal lengths of 9 each bounding box 24 of the initial fingerprint with the bounding box 24 for the equivalent word in the tracking fingerprint. The calculation of the diagonal lengths of each bounding box 24 may be done during the creation of the initial fingerprint. The initial and tracking fingerprints may be considered to match if the diagonal lengths of each bounding box 24 are similar. Alternatively or additionally, any of the information stored within the initial and tracking fingerprints may be used to match an initial and tracking fingerprint. This may include the image of the package, the package text information 20, or the package weight. Alternatively, other methods of performing the comparison may be used.

[0049] The bounding box information may be unique for any handwritten text and for any printed text and can be used to distinguish between packages that have the same text on them (e.g. two packages sent to the same address).

[0050] Finally, step S16 of the method comprises returning the associated package record(s) of an initial package if the tracking fingerprint is identical to one or more of the initial fingerprints stored within the datastore. 10051] The associated initial package record may include, but is not limited to, one or more of: • initial and tracking scans data • package text id - unique package text id • date-taken • package-text-string - string representing the text on the package • extracted address - a list or array of the best address extractions ordered by AICR confidence • user-corrected-address - address as corrected by user and entered via the app or alternative means • package-image - photo or photos of entire package (this may be in the form of a bmp, jpg, png, or other file type) • package-size - estimate of package size • package-weight - if entered, either by a machine or by the user • status - contains the id of each package processing network office that the package traverses • delivery status • sender &recipient information • payment id - payment reference id 10 • amount - how much was paid • has-been-used - has the payment for this package already been used i.e. the package has been processed by NoStamp technology at one or more package processing network nodes.

[0052] If the fingerprint of a scanned package is not found in the fingerprint datastore when it is searched, it is most likely that the package has not had its fingerprint registered within the datastore. If an initial fingerprint is found within the datastore, the initial fingerprint can be checked to confirm that the correct payment was made for the package. When a package cannot be identified in the datastore, it is most likely that no payment has been made for the package and an initial scan of the package does not exist in the datastore.

[0053] If no payment has been made for the package, then many actions could be taken at this point to obtain payment. If the package’s destination address matches that of a registered user of NoStamp, then a message can be sent to the recipient via the NoStamp app to complete payment and delivery. If no account for the sender can be found in any of these ways, handwriting recognition could be used to identify an account from the package text handwriting.

[0054] In some arrangements the initial and tracking fingerprints further and respectively comprise the actual initial and tracking scans. It will be possible to use NoStamp to match handwriting on any package to handwriting on any other package i.e. to determine if the handwritten text on two different packages was written by the same person. This capability of recognising and matching handwriting will enable functionality such as recognising that the handwritten text on a package sent without payment was written by someone who already has a NoStamp account (e.g. the handwritten text on the unpaid for package matches the handwritten text on a previously sent package that was sent using a NoStamp account); this technique can also be used to match packages of traditionally sent mail i.e. mail that uses physical proof of payment to NoStamp accounts.

[0055] As the data used to derive the fingerprints is retained, when there is a change in the way a fingerprint is calculated, such as an update or improvement to the algorithm used to compute the fingerprints, the initial fingerprint may be re-created. Storing the packages initial scan data in the datastore allows for a fingerprint to be re-created without having to take another scan of the package it represents. The system used to implement the method can then be easily updated at any point with no requirement to rescan a package. Allowing 11 the fingerprints to be updated means that the software used to create the fingerprints can be continuously improved and revised to provide better fingerprints. A system that did not maintain the scan information would not be able to continuously update the derived fingerprints, instead requiring all packages within the package processing network to be rescanned.

[0056] The method is able to identify a package even when multiple packages with the same address are posted. If two packages are scanned, each with the same destination address, and both addresses are printed addresses or high-quality handwritten addresses, then both of the separate fingerprints created for the packages could have the same package text data.

[0057] For example, using the printed text 21 shown in Figure. 3, two packages with printed addresses could end up with identical bounding boxes 24 and the following address: Her Majesty’s Passport Office Freemans Reach Riverside Place Durham DH1 1SL

[0058] A company, such as a British Gas (a utility company), will send multiple packages to the same address using the same, identical printed text 21. Fingerprints derived from a scan of these packages will have apparently identical package text information (derived from AICR), for example the bounding boxes 24 and delivery address for that package will be the same. It is not possible, using only the fingerprint derived from the package, to distinguish between identically addressed packages sent from a particular company that uses exactly the same package text information 20 printed in the same position. In some arrangements, both the initial and tracking fingerprints further comprise image data, wherein the image data is a representation of the unique artefacts identified in a scan. The use of the image data within the fingerprints provides an alternative way to distinguish between packages that have otherwise identical package text information and / or bounding box information. There are many well established algorithms, toolkits, API and approaches that can be used to compare images for similarity, even if the distinct images are taken with different cameras, under different lighting conditions and from different angles or distances. One example of a technique is named SIFT (Scale Invariant Feature Transform) in which keypoints (features) of objects are first extracted from a set of reference images and stored in a datastore. An object is recognized in a new image by individually comparing each feature from the new image to this database and finding candidate matching features that may be based on Euclidean distance of their feature vectors. Implementations and source code for this image matching technique are available multiple libraries and languages. SIFT is just one approach to matching different images. Optimisations of that approach and other applicable techniques may be used instead to derive an efficient algorithm for matching otherwise identical package text and package bounding box fingerprints. Equally, image matching may be used to perform all package identification, and not just when packages have identical package text and / or bounding boxes. It may be that image matching emerges as the optimal technique for efficient package identification.

[0059] The image data produced using techniques like SIFT may comprise a set of ‘features’ i.e. distinct points of the initial image that can be used to match against a tracking image.

[0060] In some arrangements, NoStamp’s scan may contain video (from which multiple still images can be extracted), and so the algorithm for comparing the initial and tracking scan will have a large volume of images to use for comparison. In this arrangement the chances of accurately finding a match using SIFT, or any other image comparison techniques, is greatly improved.

[0061] In some arrangements, image data is only required for matching package scans when both the package bounding box and package text fingerprints from the tracking scan match multiple initial scans in the datastore. In alternative arrangements the use of image data is used as the primary way of identifying packages in NoStamp.

[0062] The use of image artefacts to identify the package has the advantage of always providing unique image data and therefore a unique fingerprint to identify the package for initial scans of packages that have identical text on their exterior, e.g. two utility bills sent to a customer from the same utility company at the same time which will likely both have exactly the same text and bounding box data in their fingerprints.

[0063] When fingerprints further comprise image data, comparing the initial fingerprint with the tracking fingerprint may involve comparing the initial image data and the tracking image data. If the features identified by one or more techniques used (e.g. SIFT as described above) in the initial image data and the features identified in the tracking image data are sufficiently similar, the image fingerprints will be found to match. The tracked package scan will then be identified as matching an initial package.

[0064] Two apparently identical printed characters on two different packages will in fact have significant differences between them, due to artefacts and imperfections in the printed character that emerge during printing. These differences will be sufficient for a suitable algorithm, such as SIFT, to distinguish (or match) package images if text &bounding box fingerprints don’t identify a single package. This means that, at a detailed image level, even seemingly identical package addresses will have significant differences and can therefore be distinguished.

[0065] In some arrangements, the text stored in the address data is extracted from the initial scan using optical character recognition (OCR). Artificial intelligence character recognition (AICR) may be used to perform the optical character recognition. An example of AICR would be the use of various AI techniques to improve the recognition of text detected using conventional OCR. Examples of cloud-based AICR services are Azure Video Indexer, Azure Vision Cognitive Services, or AWS Textract.

[0066] When the optical character recognition is performed to extract text from a scan, information about the bounding box 24, for both handwritten text 22, 23 and printed text 21 of each word and line of text in the image may additionally be extracted. An estimation of the confidence level (e.g. 0% -100% confidence) that the identified text was correctly identified may also be provided. The confidence levels for handwriting recognition can vary significantly depending on the technology and algorithms used, the quality of the handwriting, the written language, and other factors. In general, the confidence for OCR on clean and clear handwriting might range from 70-95%, but for more challenging or messier handwriting, the confidence could drop significantly, often lower than 50%.

[0067] The confidence level in the extracted text is a numerical value assigned, for example by an AICR engine, to package text information. The confidence level indicates the degree to which the engine is certain that it has recognised the package text information correctly. The confidence level can apply to any portion of the recognised package text information, from single characters to words, lines or complete pages. AICR technologies for printed text can achieve high accuracy rates, often in the range 90-99%.

[0068] When the comparison of two fingerprints is performed, an accurate conversion of the package’s initial and tracking text information 20 is not required, only a consistent reproduction of the initial scan at the tracking scan stage is necessary. If the text extracted is consistent between the initial and tracking scans, the initial package record is likely to be later found from the tracking scan, as the same incorrectly converted text will be produced in both initial and tracking scans, allowing for an accurate comparison between the two 14 fingerprints derived from this information. The method is able to function even on challenging handwriting, as long as a consistent reproduction is maintained between initial and tracking scan stages.

[0069] The initial fingerprint and tracking fingerprint may further comprise one or more of the following: resolutions of the image, an indicator of whether the text was printed, the confidence level of each recognised word, and other text values. The fingerprint may additionally comprise one or more of: an array of all the words in that image; an array of the lines of text in that image; bounding box information for each word, line and paragraph; or rotation angle information for some or all parts of the package text. 10070] When a comparison between an initial fingerprint and a tracking fingerprint is performed, these values may then be considered. For example, words with a low confidence value may be given less weight when comparing initial and tracked package text. Alternatively, the fingerprints may be viewed as more similar if the same word has a similar confidence level in each fingerprint. Alternatively, another method of performing the comparison may be used. Initial and tracking scans of a package may each contain video from which multiple still images of that package may be extracted and compared against each other; comparing multiple still images improve the chances of successfully identifying an initial package record from a tracking scan.

[0071] In an arrangement, the method further includes receiving package text information that may comprise a package delivery address for the initial package. The package delivery address is the address that the package should be delivered to, and may comprise the entire address or a subset of the address such as a house number and / or postcode. Additionally, the associated initial package record may comprise the package text for the initial package. When the comparison between an initial fingerprint and a tracking fingerprint is successful, and the AICR converted package text is recognisable by NoStamp as a full or partial address, the associated initial package record may include the package delivery address.

[0072] The same address recognition software may be used for both the package sender’s “initial scan” and by the “tracking scan” that may be performed by the package processing network’s computer equipment, that may also check postage has been paid. Any address that cannot be correctly read by the NoStamp recognition software used to identify the package delivery information text 20 in an “initial scan” may also not be capable of being correctly read by the package processing network’s “tracking scan”. Allowing or requiring for any address that cannot be accurately read from the initial scan data, an alternative, manually entered, unambiguous “digital” address receiving method is provided (usually as a feature of the NoStamp app). The method allows a user to enter the address through a specifically designed address entry user interface. Entering a manual address ensures the packages are easier to route by machines and less likely to require inefficient hand-sorting.

[0073] The method may further involve receiving the package delivery address when the OCR, AICR or alternative method of character recognition is unable to read the address on the package. The method may also highlight an incorrectly AICR “extracted” address, allowing the user to manually type in the correct the address - in most cases only a house or flat number and postcode are required to correctly deliver a package, so the method may only receive that minimal information. Receiving a “digital” package delivery address means that a package that would previously have to be sorted by hand because the address could not be automatically read by the package processing service sorting technology, can now be automatically routed by the package processing service’s machines using the custom user-entered “digital” address instead. The package can then be delivered in a shorter period, as there is no requirement for the package to be manually sorted.

[0074] In some arrangements, the method further includes sending a message that alerts the package sender when the address information cannot be located from a list of known addresses. The cause of failing to locate the address may be that the address no longer exists, or the address is very new and is not yet in the address list, or because the address on the package was incorrectly entered or some other reason. Providing an alert that the address information is incorrect allows for the package delivery address to be provided manually by the sender. Without a manually entered “digital” address, the package would require manual sorting and may not even be deliverable.

[0075] In an arrangement, the method further includes receiving the weight of the initial package. Additionally, the initial fingerprint further comprises the weight of the package. The method may further include receiving the weight of the package at the time of a tracking scan. The initial weight may be received via an app that allows a user to input the weight of a package. Additionally, the tracking fingerprint may further comprise the weight of the package and the comparison further includes comparing the weight of the initial package and the weight of the tracked package. The weight of certain packages may differ significantly, such as parcels. The consideration of the weight of each package when comparing the fingerprint allows for a more detailed and accurate comparison. The likelihood that a package is mis-identiTied is therefore reduced.

[0076] Users may be able to register an account on NoStamp by providing their own residential, work or other address, allowing for them to be alerted when a package is sent to them. A user can also use the app without proving their identity and address, so that they can easily send packages and receive SMS / WhatsApp updates of a package’s progress. A user may further use the app completely anonymously without entering any personal information and pay using some anonymous currency like bitcoin. A user may be required to prove one or both of their own identity and their own address to use the full features of the app, as they will be able access sensitive information about packages being sent to / by them, and control delivery destination &timing of mail via the app. 10077] In some arrangements, to prove where they live, a user may present sufficient valid address information and documents via the app when requested. The app uses its usual address scanning to read the address from proof documents and then verify their authenticity.

[0078] After proving their own identity and address information, users can be notified of ANY mail that is being delivered to their registered address, regardless of whether the package uses the presently described method or is stamped / franked, has some other payment mechanism.

[0079] This functionality is available because once the user has verified their address, the package processor can match the address on any package against a recipient’s stored addresses (if it exists) from the datastore and then may send that recipient’s app / mobile or other destination device messages about the progress of or issues with any package that is addressed to them.

[0080] A package sender (“sender”) may be encouraged to add the package recipient’s (“recipient”) mobile number to become a part of the package record information. If the sender adds the recipient’s mobile number, and the recipient agrees, the recipient may be sent a message (via SMS, WhatsApp or some other channel as required) saying that the sender has sent a package to them, the estimated delivery date / time and may include other information such as a link to the NoStamp app with information about it. This mechanism for informing the recipient about NoStamp’s service may be present to encourage virality and uptake as well as providing a useful service to track a package’s progress.

[0081] NoStamp may save the recipient’s address and mobile number (likely, but not necessarily after informing the Recipient), so in future if anyone sends mail to that recipient’s address (either via NoStamp, or with a traditional stamp or frank of some kind) the recipient may be informed by SMS / Whatsapp etc as well as via NoStamp. The 17 recipient may have to go through an address authentication processes to authenticate their address with the app.

[0082] Once payment for the package being sent has been made successfully, a message confirming this may be displayed to the user together with a link that takes the user to a ‘Track” screen. The ‘Track’ screen may show that a package that has been paid for, together with a status (initially, for example, “Ready to Post”), the amount paid, the package weight and a globally unique package identifier that may be used to identify the package and perhaps track its location. The track screen may show the progress of the Package is as it traverses the package processing network. 10083] A refund button may also be displayed at this point to undo the NoStamp payment and refund it (less, if required, an amount to cover any refund charges). If the refund button is used before posting the package, the package may require an alternative form of payment, e.g. via the NoStamp app again, a stamp or a frank or letting the recipient pay. The sender may then post the package in any usual way of posting mail, e.g. via a post box or at a post office. The app may also help direct the sender to the place geographically closest to them where they can post a package.

[0084] In an arrangement, the NoStamp app may enable a party (the “payee”) other than the package sender to pay for the package being sent; the package sender may use the app to send a payment link to the payee which enables the payee to download and open the NoStamp app and pay the necessary fee to successfully send the package. The NoStamp app may also allow recurring billing where a bill is charged after a set period. A user may be billed weekly or monthly for the level of postage service they have used, rather than paying for each item sent individually. At the end of each billing period a bill for the amount of postage may be issued, rather than the user having to pay for each item sent ahead of time. When each posted item is scanned, the payment is aggregated and is only paid at the end of the billing period.

[0085] In an arrangement, the initial and tracking scans are videos showing the package exterior. This method may further comprise converting the video into a sequence of still images, wherein the package text information 20 is identified using text fragments consistent across each of the still images in the sequence of still images. When OCR, AICR, or an alternative character recognition method is used to extract the package text information 20, the character recognition methods (such as AICR) may be performed on multiple of the still images. Text extraction performed in this way, especially for handwritten addresses, will provide a more accurate extraction of the text when compared with extracting information from just a single image.

[0086] The package text information 20 can be more easily extracted because a video contains multiple images of the same package from different camera positions. The variation of images greatly assists with the accurate determination of address text, as a word that cannot be identified in one image will be able to be identified using another that shows the word from a different angle. The chance of an aberration caused by a defect in the scan device, such as a scratch on a lens or distortion of the image at the edges, impacting the extracted text is also reduced. The variation in image angles allows for all of the text to be accurately identified. A scan can also contain a plurality of images, as well as the single frame image previously described.

[0087] A single frame image provides the benefit of being simple, and allows for data to be processed cheaply and at faster speeds when compared with multi-image video. A single frame image (not extracted from video) can be combined with the normalisation techniques described below to deliver results that are almost as good as using video, especially for printed text 21. Multi-frame images can also be used to extract address information. Multi-frame images allow for multiple views of the package from different angles, allowing for a higher quality fingerprint to be produced whilst still having a fast processing speed and reduced cost. Video based scans provide multiple and varied representations of the same image. Video is a much richer source of data for text processing and gives better results especially for handwritten text 22, 23.

[0088] The use of a video scan of the package greatly assists in producing a high-quality fingerprint of a package. A video scan produces, especially for difficult handwriting, a wide variety of different package text images in the fingerprint. The overall variation of the text extraction is sufficiently invariant to give high confidence matching between a tracking fingerprint and an initial fingerprint retrieved from the datastore entry. This matching can be performed either through machine learning, algorithmically or both or some other method.

[0089] An example of this will be described with reference to Figure. 5. Attempts to extract the bad handwriting example shown from multiple frames of a video delivered different results. The consistent text fragments identified in these images can then be used to form the text data of the fingerprint. As these fragments are consistently identified, this allows for “incorrect” (i.e. not a 100% accurate text extraction) representation of the package text to be used when comparing an initial and tracking fingerprint. The use of a 19 video allows for more attempts to be made to extract the text, resulting in a more accurate determination of the text fragments commonly extracted from the text. In the case of Figure. 5, the address extracted was: MaximLodg 72 Groove Tunbridg wells TBI2A

[0090] This address is extracted consistently between the initial and tracking scans, and so two distinct fingerprints created from the package text information shown in Figure. 5 can be identified as coming from the same package even though the address itself cannot be parsed correctly.

[0091] In some arrangements, capturing the initial scan and tracking scan includes capturing an initial and tracking set of pose information about the position of the scan devices used to capture the initial and tracking scans that are stored in the initial and tracking fingerprints. The pose information stores the device’s pose. The device’s pose is the position and orientation of the camera as it is videoing the package. The pose information can be obtained from the scanning device’s equipment, such as gyroscopes, accelerometers etc. When the pose information is captured, it is synchronised with the scan. The pose information can be used to determine the position and orientation of the device used to capture the scan throughout the creation of a Scan. The scanning device may be a mobile phone.

[0092] Sensor data may include one or more of: gyroscopic readings to determine the camera “pose” (rotation in x, y, z axes) as images were being recorded by the app; accelerometer readings to detemiine the movement of the camera as images were being recorded by the app; GPS information to assist with package identification (e.g. in the package processing system), to determine the nearest location the sender can post the package; operating system version including major, minor, revision values; camera information such as resolution, front / back camera.

[0093] In some arrangements, the initial and tracking scans are normalised using the initial and tracking pose information, wherein normalising the initial and tracking scans involves adjusting the scans based on the location of the scan device. The pose information enables positional and rotational transformations to be applied to the scan data and to “normalise” images before a fingerprint is created using information derived from the scan. The orientation of the camera can be compensated for to produce a more consistent scan. Normalising the scan improves the accuracy of address recognition.

[0094] In some arrangements, the method further comprises calculating the correct postage to pay to send the package. The weight of the package may be included when calculating the correct postage. Once the correct postage has been calculated, the user is able to pay. The method may be implemented using an app, with the user able to make a payment through the app. The payment information may be stored in the associated initial package record. The payment information can then be accessed at a later point if it is required. For example, when the package is later scanned to produce a tracking fingerprint, the initial fingerprint can be identified and the corresponding payment information can easily be verified without the requirement of any form of mark identifying payment on the physical package.

[0095] If the package sender is unsure whether or not they have already paid to send their package, they can check easily by using the app and looking at their package scan and / or payment history, or by scanning the package again to confirm whether a payment for that package has already been made.

[0096] The payment screen of the app shows the user how much it costs to send the package, to the address shown, if its weight and dimensions are less than a certain amount for that package processing service (e.g. A4 size, 100g weight). If the package weighs more or has greater dimensions than a predetermined amount, the user may type (or enter in some other way) the package’s weight and dimensions. Once the user has correctly supplied the package’s weight and dimensions and checked the package’s destination address, they press to confirm and pay. Payment may be made via one or more of the usual platforms such as Stripe, crypto, PayPal etc. or by some other method including requesting a third-party to pay for the package’s sending costs. If the sender’s account is anonymous the payment is a process that may have to be repeated each time a package is sent, but if the account has been registered the user may be able to set up a reusable payment mechanism. This simple payment UI will eliminate a lot of over and underpayment of postage and is a benefit to all the different participants in the system.

[0097] In some arrangements, the method further comprises automatically identifying, the recipient from their address and notifying that recipient that a package is being sent to their address. If required information is not currently known about the recipient, the sender may be prompted to provide additional information.

[0098] At its limit, the method does not require even an address on the physical package - a user could use the initial scan of the package and type in the recipient’s address in the app (or just postcode and house number) into the app as described above. Images of the physical package’s surface (without text) will then be associated with the recipient address the user entered into the app directly. When a tracking scan occurs on a package with no address, an image fingerprint match would be used to find the package in the datastore and extract the typed address information. A mobile phone and NoStamp app can be used to allow the deliverer (e.g. postman) to easily obtain a recipient’s address from an “addressless” package, to allow for the final stage of delivery to take place.

[0099] In some arrangements, the method can be applied to franking machines. The use of the method allows for the construction of franking machines to be simplified, as well as reducing its running cost by eliminating printing components and costs. Additional tracking of packages could be provided that is not currently offered by existing franking machines. By using NoStamp companies could charge third-parties for package postage payment if required.

[0100] A user might set up a NoStamp account to send a package using the system, because of all the benefits it provides, though a user can instead choose to send their package completely anonymously. The amount of personal information in a user’s account may determine the level of functionality that NoStamp can deliver and the complexity of registration as described below. A user of the app can use it without registering at all (anonymously), and use crypto currency to pay so that, like physical stamps paid for in cash, their package sending is essentially anonymous. It is anticipated that registration-free and anonymous use of the app is vital for an application that can be used instead of traditional stamps, which are an anonymous payment method. Alternatively, quick registration may be used, where the package sender enters just their email address and / or mobile number which enables SMS / WhatsApp / email etc. notification messages to be delivered to them. In this case, payment can be made via a credit card, crypto currency, Stripe or similar mechanism and can be saved on the app for “one-click” payment. Alternatively, the user can fully register their account. With full app registration, a user may enter their email, phone number &password, social media credentials etc. and the can then receive text and email notifications from NoStamp and add a payment method that is setup once and can be subsequently reused.

[0101] A user may also have to verify their address. A NoStamp app user can use the app without proving their identity and address, so that they can easily and quickly send NoStamp Packages and get SMS / WhatsApp updates of progress. Users may have to prove one or both of their own identity and their own address to use the full features of the app, as they may be able access sensitive information about Packages being sent to / by them, and control delivery destination &timing of mail via the app. To prove an address, a user may have to present sufficient valid address information and documents via the app when requested. The app may use address scanning technology to read the address from proof-of-address documents and then verify their authenticity. Potential third party identity services that could be used to verify addresses in the UK include ‘Trulioo’ and ‘Zinc app’.

[0102] After proving their own identity and address information, users may be notified by the app about any mail that is being delivered to the address they registered, whether the package uses NoStamp or is stamped / franked or has some other payment mechanism. This functionality is available because once the user has verified their address on the NoStamp system, a tracking scan can match the address that is on any mail (even if it isn’t using NoStamp) against the NoStamp datastore and then send that user’s NoStamp app / mobile messages or email about the progress of or issues with any mail that is being sent to their physical address. NoStamp can be used to manage “virtual” addressing. Mail that has one physical delivery address on the package surface can instead be delivered to a different physical address as the recipient instructs.

[0103] NoStamp Payment depends on the registration method a Sender uses (described above) and includes, but is not limited to credit card, crypto currency, Stripe or other similar mechanism. Users may be able to save their payment details to enable “one-click” payment for any Packages they send. Once they have setup one-click payment, users can pay for sending a Package with a single click.

[0104] NoStamp should be an immediately useful tool. Many people understand how stamps work - buy some stamps, put one, or a few, onto a package and post it. To deliver all of the NoStamp benefits, a user may have to register with NoStamp and validate their identity and address, but this takes some time and effort and an appreciation of why they’re doing it. Anonymous use of the app enables a user to download the NoStamp app and immediately use it (without any form of registration) to send a package. A NoStamp user must still provide payment (unless some kind of “free” usage is provided, perhaps as a trial to incentivise NoStamp uptake) and a crypto payment option will be provided to assist anonymity. For convenience and anonymity, it will be possible to pay for NoStamp using crypto currency with a wallet such as MetaMask, Coinbase or Ledger.

[0105] For a package processing service, such as the UK’s Royal Mail, to use NoStamp it would have to make some changes to its current technology infrastructure. The package processing service may need to change existing systems so that packages sent without traditional stamps or franks are not rejected but may be further processed by NoStamp technology. The NoStamp technology required for tracking scanning may be a camerabased system capable of photographing packages and determining, by using the NoStamp datastore and API, if a package has been paid for via NoStamp. The UK Post Office already uses photographic systems and envelope address reading technology, so it may be that adding NoStamp features is not a very complicated enhancement to the existing process.

[0106] If the package processing service tracking scan matches a Package image with a paid record entry in the datastore, and the user and / or recipient permit notifications, a message may be sent, informing them about the package’s progress through the package processing service. If a package fingerprint is found in the datastore, but the package has not been paid for (either with stamp, frank or NoStamp) the sender (if they have a NoStamp app) may be sent a message asking for payment that may be via a link to a payment page or via the app. Anytime the package has a tracking scan as it progresses from initial scan to being delivered, the sender and recipient may be updated about its progress.

[0107] Package processing services are already able to provide customers with services such as redirecting mail to an alternative address, or not delivering it for a certain period, if the user requests it. NoStamp can enable rerouting via the app (because the NoStamp app user has already proven the address of a registered user). So, NoStamp may provide more “dynamic” delivery services such as redirecting packages for a very short period and even provide programatic control of redirection and other features via an API so opening up an entirely new range of services.

[0108] As discussed above, the various embodiments can be implemented in a wide variety of operating environments, which in some cases can include one or more user computers, computing devices, or processing devices which can be used to operate any of a number of applications. User devices can include any of a number of general-purpose personal computers, such as desktop or laptop computers running a standard operating system, as well as cellular, wireless, and handheld devices running mobile software and 24 capable of supporting a number of networking and messaging protocols. Such a system also can include a number of workstations running any of a variety of commercially-available operating systems and other known applications for purposes such as development and datastore management. These devices also can include other electronic devices, such as thin-clients, gaming systems, and other devices capable of communicating via a network, provided the devices include a fde system to protect.

[0109] Various aspects also can be implemented as part of at least one service or Web service, such as may be part of a service-oriented architecture. Services such as Web services can communicate using any appropriate type of messaging, such as by using messages in extensible markup language (XML) format and exchanged using an appropriate protocol such as SOAP (derived from the "Simple Object Access Protocol"). Processes provided or executed by such services can be written in any appropriate language, such as the Web Services Description Language (WSDL). Using a language such as WSDL allows for functionality such as the automated generation of client-side code in various SOAP frameworks. Alternatively, more modern technology stacks such as JSON messages and Open API standards may be used to define the APIs used.

[0110] In embodiments utilising a Web server, the Web server can run any of a variety of server or mid-tier applications, including HTTP servers, FTP servers, CGI servers, data servers, Java in servers, and business application servers. The server(s) also may be capable of executing programs or scripts in response requests from user devices, such as by executing one or more Web applications that may be implemented as one or more scripts or programs written in any programming language, such as JavaL. C, C# or C++, typescript javascript or any other language, such as Node, Go, Perl, Python, or TCL, as well as combinations thereof. The server(s) may also include datastore servers, including without limitation open source datastores such as or MySQL or PostgresSQL or those commercially available e.g. from Oracle®, Microsoft®, Sybase®, and IBM®.

[0111] The environment can include a variety of data stores and other memory and storage media as discussed above. These can reside in a variety of locations, such as on a storage medium 36 local to (and / or resident in) one or more of the computers or remote from any or all of the computers across the network, or on a third-party provided cloud platform such as Azure or AWS. In a particular set of embodiments, the information may reside in a storage-area network (“SAN”) familiar to those skilled in the art. Similarly, any necessary files for performing the functions attributed to the computers, servers, or other network devices may be stored locally and / or remotely, as appropriate. Where a system 25 includes computerised devices, each such device can include hardware elements that may be electrically coupled via a bus, the elements including, for example, at least one central processing unit (CPU), at least one input device (e.g., a mouse, keyboard, controller, touch screen, or keypad), and at least one output device (e.g., a display device, printer, or speaker). Such a system may also include one or more storage devices, such as disk drives, optical storage devices, and solid-state storage devices such as random access memory (“RAM”) or read-only memory (“ROM”), as well as removable media devices, memory cards, flash cards, etc.

[0112] Such devices also can include a computer-readable storage media reader, a communications device (e.g., a modem, a network card (wireless or wired), an infrared communication device, etc.), and working memory as described above. The computer-readable storage media reader can be connected with, or configured to receive, a computer-readable storage medium 36, representing remote, local, fixed, and / or removable storage devices as well as storage media for temporarily and / or more permanently containing, storing, transmitting, and retrieving computer-readable information. The system and various devices also typically will include a number of software applications, modules, services, or other elements located within at least one working memory device, including an operating system and application programs. It should be appreciated that alternate embodiments may have numerous variations from that described above. For example, customized hardware might also be used and / or particular elements might be implemented in hardware, software (including portable software, such as applets), or both. Further, connection to other computing devices such as network input / output devices may be employed.

[0113] Storage media and other non-transitory computer readable media for containing code, or portions of code, can include any appropriate media known or used in the art, such as but not limited to volatile and non-volatile, removable and non-removable non-transitory media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules, or other data, including RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium which can be used to store the desired information and which can be accessed by the a system device. Based on the disclosure and teachings provided herein, a person of ordinary skill in the art will appreciate other ways and / or methods to implement the various embodiments.

[0114] Computational tasks performed as part of the above description may be performed by more than one computing device serially or concurrently. If performed on a computer system comprising multiple computers, these computers may be co-located or distributed. The invention can be implemented wholly in a client computer, on a server computer, or a 5 combination of client and server side processing.

[0115] Having described the invention it will be appreciated that variations may be made on the above described embodiment, which is not intended to be limiting. The invention is defined in the appended claims and their equivalents. 10

Claims

1. A method for identifying a package comprising:capturing an initial scan of the exterior of an initial package and receiving an associated initial package record,creating an initial fingerprint comprising information derived from the initial scan and an associated initial package record and storing the initial fingerprint, initial scan and the associated initial package record in a datastore that may be configured to store one or more initial fingerprints, one or more initial scans and one or more associated initial package records,capturing a tracking scan of the exterior of a tracked package in order to retrieve its initial associated package record,creating a tracking fingerprint representing the tracked package, comparing the tracking fingerprint to the one or more fingerprints stored within the datastore, andreturning the associated initial package record of the initial package if the tracking fingerprint matches one or more of the initial fingerprints stored within the datastore.

2. The method of the previous claim, wherein the initial and tracked packages comprise package text information positioned on the package surface, and both the initial and tracking fingerprints further comprise initial and tracking package text data respectively, and wherein the method further includes extracting the package text information from the initial and tracking scans and storing the package text information within the initial and tracking fingerprints, optionally wherein the package text information is address information.

3. The method of the previous claim, wherein both the initial and tracking fingerprints further comprise bounding box data comprising the position, size and orientation of the bounding box around each letter and word of the package text information positioned on the package surface and the position, size and orientation of the bounding box around each line of the package text information positioned on the package surface.

4. The method of the previous claim, wherein comparing the initial fingerprint with the tracking fingerprint involves comparing the bounding box data of the initial fingerprint with the bounding box data for the equivalent package text information in the trackingfingerprint, optionally wherein comparing the initial fingerprint with the tracking fingerprint involves comparing the diagonal lengths of each bounding box of the initial fingerprint with the bounding box for the equivalent text in the tracking fingerprint.

5. The method of any of the previous claims, wherein both the initial and tracking fingerprints further comprise image data, wherein the image data is a digital image representation of the initial or tracked package and its unique features identified from a scan, optionally wherein comparing the initial fingerprint with the tracking fingerprint involves comparing the initial scan’s image data and the tracking scan’s image data.

6. The method of the previous claims, wherein the text stored in the address data is extracted from the initial scan using AICR and optical character recognition, optionally wherein the initial fingerprint and tracking fingerprint further comprise one or more of the resolution of the image, an indicator of whether the text is handwriting or print, the confidence level of each recognised word, and other image metadata.

7. The method of the previous claims wherein the method further includes receiving a user provided unambiguous digital package delivery text for the initial package, and wherein the associated initial package record comprises the digital package delivery text for the initial package, optionally wherein the method further includes alerting the sender when the address information cannot be located from a list of known addresses or cannot be successfully parsed and extracted into ASCII or UNICODE formatted data by AICR.

8. The method of the previous claims wherein the associated initial package record includes the payment status of the initial package.

9. The method of the previous claims, wherein the method further includes receiving the weight of the initial package, and wherein the initial fingerprint further comprises the weight of the initial package.

10. The method of the previous claim, wherein the method further includes receiving the weight of the tracked package, the tracking fingerprint further comprises the weight of the tracked package and the comparison further includes comparing the weight of the initial package and the weight of the tracked package.2911. The method of the previous claims, wherein the initial and tracking scans are one or more of: videos, one or multiple still images, LIDAR data showing the package exterior, or other data showing the package exterior.

12. The method of the previous claim, wherein the initial and tracking scans are videos showing the exterior of the initial and tracked package respectively, and the method further comprises converting the video into a sequence of still digital images, wherein the package text information is identified using text fragments consistent across each of the still digital images in the sequence of still digital images derived from the initial and tracking scans.

13. The method of the previous claims, wherein capturing the initial scan and tracking scan includes capturing an initial and tracking set of pose information about the position of the scan devices used to capture the initial and tracking scans, and wherein this initial and tracking set of pose information is stored in the initial and tracking fingerprints respectively, once the fingerprints are created.

14. The method of the previous claim, wherein the initial and tracking scans are normalised using the initial and tracking pose information, wherein normalising the initial and tracking scans involves adjusting the images from the scans based on the location and orientation of the scan device.

15. The method of the previous claims wherein the datastore is remotely located on a cloud platform, at a datacentre or on a server capable of persistently storing information.

16. The method of the previous claims, further comprising calculating the correct postage to pay to send the initial package.

17. The method of the previous claims, further comprising identifying, from a list of recipients, the recipient with the address corresponding to the package address and notifying that recipient that a package is being sent to their address.

18. The method of the previous claims, wherein the initial fingerprint further comprises the actual image of the initial package.

19. A system configured to identify a package, comprising:a first image capture device;a second image capture device optionally the same as the first image capture device;a memory; andone or more processors configured to:capture an initial scan of the exterior of an initial package using the first image capture device and receive an associated initial package record,create an initial fingerprint comprising information derived from the initial scan and the associated initial package record and store the initial fingerprint in a datastore configured to store one or more initial fingerprints,capture a tracking scan of the exterior of a tracked package using the second image capture device,create a tracking fingerprint representing the tracked package,compare the tracking fingerprint to the one or more fingerprints storedwithin the datastore, andreturn the associated initial package record of the initial package if the tracking fingerprint is identical to one or more of the initial fingerprints stored within the datastore.

20. A computer program comprising code means that, when executed by a computer system, instruct the computer system to perform the method of claims 1 to 18.

Images