Digital fingerprinting track and trace system
Methods and systems for tracking a physical object to identify or authenticate it utilizing digital fingerprints which are based on natural features extracted from a digital image of the object. Digital fingerprints do not require or rely on any labels, tags, integrated materials, unique identification characters, codes or other items that may be added to the object specifically for the purpose of identification. Consequently, the disclosed digital fingerprint techniques help to detect or prevent unauthorized alterations of documents, apparel, drugs and pharmaceuticals, etc. Further digital fingerprints can be used to better track and trace a wide variety of objects throughout the distribution chain to demonstrate their provenance and to detect counterfeit objects.
1. A method for tracking a physical object comprising the steps of:
acquiring digital image data of an image of at least a portion of a physical object;
identifying a class of objects to which the physical object belongs;
accessing a stored feature template associated with the identified class of objects, wherein the feature template defines a first authentication region and defines at least one feature located within the first authentication region;
analyzing the image data to form a digital fingerprint of the object, wherein the digital fingerprint is responsive to natural structure of the object at least in that, first, it does not rely upon or require any labels, tags, integrated materials, unique identification characters, codes or other items that were added to the object specifically for the purpose of identification, and second, the digital fingerprint does not include text or symbols recognized from the image data;
and wherein analyzing the image data includes—
selecting a subset of the image data corresponding to the authentication region defined by the feature template;
locating the at least one feature located within the first authentication region;
for each feature, extracting at least one feature vector, the feature vector comprising an array of numeric values based on a portion of the image data corresponding to the feature;
combining the extracted feature vectors to form the digital fingerprint, without including in the fingerprint any text or symbols recognized from the image data and storing the digital fingerprint in a database record.
2. The method of claim 1 and further comprising adding initialization data about the object to the database record, so that the initialization data is associated with the digital fingerprint without changing the digital fingerprint itself.
3. The method of claim 1 and further comprising:
subsequent to storing the digital fingerprint in a database record, acquiring second digital image data of an image of at least a portion of a target physical object;
extracting at least one feature vector from the second image data to form a second digital fingerprint in accordance with the stored feature template;
querying the database to seek a matching record based on the second digital fingerprint;
wherein querying the database includes calculating a Euclidean vector distance between corresponding feature vectors of the second digital fingerprint and digital fingerprints previously stored in the database;
returning a matching record based on the corresponding vector distances;
in the case that a matching record is returned; updating the matching record with an indication that the second digital fingerprint was matched to it; and
in the case that a matching record is not returned; updating the matching record with an indication that the second digital fingerprint was not matched to it.
4. The method of claim 3 wherein matching a record comprises fuzzy matching.
5. The method of claim 3 wherein matching a record comprises identifying at least one record having a digital fingerprint that matches the second digital fingerprint within a predetermined threshold of similarity.
6. The method of claim 3 and further comprising adding second data to the database in association with the database record, wherein the second data is associated with the target physical object but the second data is not itself a digital fingerprint.
7. The method of claim 3 including, based on identifying a matching record, updating the matching record with at least one of an indication of a current location of the target physical object and an identification of an entity having possession of the target physical object.
8. The method of claim 3 and further comprising storing the second digital fingerprint in the database and associating the second digital fingerprint with the database record.
9. The method of claim 3 wherein the object comprises
Government documents, legal and financial documents, mail pieces, parcels, art, photographs, coins, currency, precious metals, gems, jewelry, apparel, mechanical parts, consumer goods, integrated circuits, weapons, pharmaceuticals, drugs, alcohol, tobacco and food and beverages.
10. A method comprising:
identifying a class of objects to which a physical object belongs;
accessing a stored feature template associated with the identified class of objects, wherein the feature template defines plural authentication regions;
acquiring digital image data of an image of at least a portion of a physical object that includes the plural authentication regions;
for each of the plural authentication regions, extracting at least one feature vector, the feature vector comprising an array of numeric values based on a correspond portion of the image data;
combining the extracted feature vectors to form a digital fingerprint, without including in the fingerprint any text or symbols recognized from the image data;
storing the digital fingerprint in a database;
acquiring a second digital fingerprint of a target object that belongs to the identified class of objects;
querying the database based on the second digital fingerprint to find a matching fingerprint, wherein a matching record requires that the second digital fingerprint match the stored extracted feature vectors for each and every one of the plural authentication regions defined by the stored feature template within a predetermined tolerance.
11. The method of claim 10 wherein extracting the feature vector from the authentication region includes identifying a location and shape of a distinctive aspect with the selected region based on the image data.
12. A method for authenticating an object comprising:
identifying a class of objects;
accessing a stored feature template associated with the identified class of objects, wherein the feature template defines a first authentication region, a second authentication region, and a position of the second authentication region relative to the first authentication region;
acquiring digital image data of an image of at least a portion of an object that includes the first and second authentication regions;
selecting a first subset of the image data corresponding to the first authentication region;
extracting at least a first feature vector from the first subset to form a first digital fingerprint associated with the first authentication region, the first feature vector comprising an array of numeric values based on the first subset of the image data;
selecting a second subset of the image data corresponding to the second authentication region;
extracting at least a second feature vector from the second subset to form a second digital fingerprint associated with the second authentication region, the second feature vector comprising an array of numeric values based on the first subset of the image data;
determining a position of the second authentication region relative to the first authentication region in the image data and querying a database of stored digital fingerprints to find a record in which—
(1) the stored digital fingerprint includes a first stored digital fingerprint that matches the first digital fingerprint associated with the first authentication region within a predetermined threshold difference tolerance;
(2) the stored digital fingerprint includes a second stored digital fingerprint that matches the second digital fingerprint associated with the second authentication region within a predetermined threshold difference tolerance; and
(3) the defined position of the second authentication region relative to the first authentication region matches the determined position of the second authentication region relative to the first authentication region in the image data within a predetermined threshold difference tolerance; and
responsive to finding a record in the database in which at least said three criteria are met, reporting an indication that the object matches an object that corresponds to the database record.
13. The method of claim 12 wherein the extracted first feature vector represents the image data corresponding to the first authentication region in a compacted form.
14. The method of claim 12 wherein the first extracted feature vector comprises an array of color or gray scale numeric values corresponding to areas within the first authentication region.
15. The method of claim 12 wherein the first extracted feature vector comprises at least one numeric value that represents a selected mathematical function of the individual values of a corresponding group of pixels forming a sub-part of the first authentication region.
16. The method of claim 12 including associating the matching database record to the acquired digital image data.