Fraudulent activity detection at a barcode scanner by verifying visual signatures
System and method for detecting a fraudulent activity at a barcode scanner is disclosed. The method issues an alert when the fraudulent activity is confirmed by comparing the visual signature of the item being transacted over the checkout terminal to the model visual signature. The model visual signature is obtained by averaging the collection of visual signature of the item gathered over a period of time. A human validation via a remote processor is employed to confirm the fraudulent activity verified by a computer.
1. A method, comprising:
identifying, from an image, interesting points surrounding an item identifier for an item, wherein the interesting points are on a packaging of the item;
calculating a signature for the item identifier based on the interesting points;
verifying the signature matches a model signature for the item identifier; and
raising an alert during a transaction associated with the item when the signature does not match the model signature based on the verifying.
2. The method of claim 1 further comprising, recording a video of the transaction as the item is processed at a terminal during a checkout.
3. The method of claim 2 , wherein recording further includes creating a video snippet from the video comprising a portion of the video when the item was processed during the transaction.
4. The method of claim 3 , wherein creating further includes linking the video snippet to the alert.
5. The method of claim 1 , wherein verifying further includes retaining the signature with a group of acceptable signatures for the item identifier when the signature matches the model signature, wherein the group of acceptable signatures are accumulated for previous transactions associated with the item, and updating the model signature based on adding the signature to the group of acceptable signatures.
6. The method of claim 1 , wherein verifying further includes comparing the interesting points against an average of other interesting points represented by the model signature.
7. The method of claim 1 , wherein verifying further includes comparing the interesting points against a threshold boundary of a range represented by the model signature.
8. The method of claim 1 , wherein verifying further includes comparing the interesting points against a predefined variance represented by the model signature.
9. The method of claim 1 , wherein raising further includes sending the alert to an alert system.
10. The method of claim 1 , wherein raising further includes activating a light, generating a visual signal, playing an audible alarm, presenting a message on a display associated with a terminal, sending a text message, sending an email message.
11. A method, comprising:
monitoring items being processed by a terminal as the items are scanned at the terminal during a checkout;
capturing images of the items as they are scanned and for each image:
identifying an item identifier associated with the corresponding item and interesting points that surround the item identifier, wherein the interesting points are on a packaging of the item;
calculating an item signature from the interesting points;
comparing the item signature against a model item signature for the corresponding item; and
generating an alert during the transaction when the item signature does not match the model item signature.
12. The method of claim 11 , wherein identifying the item identifier for each image further includes selecting the interesting points from the image as features associated with a corner, an edge, and an area of high texture depicted in the image for the corresponding item.
13. The method of claim 11 , wherein identifying the item identifier for each image further includes adding other interesting points included within the item identifier from the image to the interesting points.
14. The method of claim 11 , wherein identifying the item identifier for each image further includes selecting the interesting points from the image as scale-invariant feature transform features associated with the interesting points.
15. The method of claim 11 , wherein identifying the item identifier for each image further includes selecting the interesting points from the image as affine invariant features associated with the interesting points.
16. The method of claim 11 , wherein identifying the item identifier for each image further includes selecting the interesting points from the image as speeded up robust features associated with the interesting points.
17. The method of claim 11 wherein identifying the item identifier for each image further includes selecting the interesting points from the image as image histogram features associated with the interesting points.
18. The method of claim 11 , wherein the alert for each image further includes linking a video snippet to the alert, wherein the video snippet corresponding to the image and representing when the corresponding item was scanned at the terminal during the transaction.
19. A device, comprising:
a processor;
a non-transitory computer-readable storage medium having executable instructions; and
the executable instructions when executed by the processor from the non-transitory computer-readable storage medium cause the processor to:
obtaining interesting points that surround an item identifier from an image captured of the item during a transaction at a terminal when the item is scanned at the terminal for the transaction, wherein the interesting points are on a packaging of the item;
combining the interesting points with other points associated with the item identifier;
calculating a signature for the item based on the interesting points and the other points;
comparing the signature against a model signature associated with the item identifier; and
raising a transaction alert during the transaction when the signature does not match the module signature for transaction intervention.
20. The device of claim 19 , wherein the item identifier is a barcode, a stock keeping unit number (SKU) or a product lookup number (PLU).