Granted Patent B2

US 8,750,624 · App. 12/907,226 · Granted Jun 10, 2014

Abstract

Claims (48)

1. A method of detecting duplicate document content, the method comprising:

inputting a query document page to a document detection system;

comparing the query document page with pages of documents stored in a document collection by using visual fingerprints of a two-dimensional visual fingerprint detection process, wherein the visual fingerprints are at least one of affine and perspective invariant visual fingerprints; and

automatically highlighting duplicate or different document content of the query document page and at least one document page of the stored documentation collection based on the results of using the two-dimensional visual fingerprint detection;

wherein the automatically highlighting is accomplished by calculating a mean and standard deviation for two-dimensional locations of each cluster of non-corresponding fingerprints of the query document page and one of the document pages of the document collection; and

wherein the method is implemented using at least an electronic processor.

2. The method according to claim 1 wherein a list of the documents of the document collection having matching fingerprints to the query document page is returned and the returning of the list of documents having matching fingerprints to the query document page includes forming the list from only documents of the document collection having a predefined required minimum corresponding number of matching fingerprints.

3. The method according to claim 1 wherein clusters of corresponding identical fingerprints in the query document page and one of the documents of the document collection represent regions of matching document content, while clusters of non-corresponding fingerprints represent document content differences.

4. A method of detecting duplicate document content, the method comprising:

inputting a query document page to a document detection system;

comparing the query document page with pages of documents stored in a document collection by using visual fingerprints of a two-dimensional visual fingerprint detection process; and

automatically highlighting duplicate or different document content of the query document page and at least one document page of the stored documentation collection based on the results of using the two-dimensional visual fingerprint detection, wherein the automatically highlighting includes generating a geometric element for each cluster of non-corresponding fingerprints on at least one of the query document page and one of the document pages of the returned documents of the document collection,

wherein the method is implemented using at least an electronic processor.

5. The method according to claim 4 wherein the geometric element is centered at a cluster mean and a size of the geometric element is proportional to estimated second moments of fingerprint location distribution within the cluster.

6. The method according to claim 5 wherein the size of the geometric element is further controlled in proportion to the number and location of any corresponding fingerprints occurring inside or in close proximity to the cluster, and by cluster complexity, shape, and fingerprint density.

7. The method according to claim 4 wherein the geometric element is drawn or super-imposed on top of the visual image display in a highlight color, or is saved in a separate image layer of metadata.

8. The method according to claim 1 wherein the document collection includes at least one of electronically created documents and/or scanned versions of hardcopy documents.

9. The method according to claim 1 wherein the document collection is an abstract logical document collection that may be physically distributed in a set of sub-collections across any one of a number of folders, drives systems, networks, remote servers, web-servers, and/or cloud storage facilities.

10. A method of detecting duplicate document content, the method comprising:

inputting a query document page to a document detection system;

comparing the query document page with pages of documents stored in a document collection by using visual fingerprints of a two-dimensional visual fingerprint detection process;

storing the visual fingerprints in a fingerprint lookup index and incrementally updating the fingerprint lookup index when documents are added to the document collection or when documents are deleted from the document collection, wherein the fingerprint lookup index is formed by processing a sequence of page images, computing the visual fingerprints; and adding the visual fingerprints and their locations to the fingerprint lookup index, and

automatically highlighting duplicate or different document content of the query document page and at least one document page of the stored documentation collection based on the results of using the two-dimensional visual fingerprint detection

wherein the method is implemented using at least an electronic processor.

11. A method of detecting duplicate document content in a large document collection and automatically highlighting duplicate or different document content among the detected document content using two-dimensional visual fingerprints, the method comprising:

generating a document collection;

visually fingerprinting documents of the document collection;

indexing the fingerprinted documents of the document collection;

storing in memory the indexed fingerprints of the documents of the document collection;

presenting a query document to a content detection system;

fingerprinting the query document;

looking up the fingerprints of the query document in the memory storing the indexed fingerprints of the documents of the document collection;

returning a list of documents of the document collection having fingerprints matching the fingerprints of the query document;

performing a fingerprint comparison for each document of the document collection having matching fingerprints to the query document, wherein the comparison includes comparing the fingerprint locations and their corresponding layout in the documents being compared; and

automatically highlighting one of the duplicate or non-duplicate document content between the query document and the one of the documents of the document collection, wherein the method is implemented using at least an electronic processor.

12. The method according to claim 11 wherein the document collection includes at least one of electronically created documents and/or scanned versions of hardcopy documents.

13. The method according to claim 11 wherein the document collection is an abstract logical document collection that may be physically distributed in a set of sub-collections across any one of a number of folders, drives systems, networks, remote servers, web-servers, and/or cloud storage facilities.

14. The method according to claim 11 wherein the stored fingerprint lookup index is incrementally updated when documents are added to the document collection or when documents are deleted from the document collection.

15. The method according to claim 11 wherein the returning of the list of documents having matching fingerprints to the query document includes forming the list from only documents having a predefined required minimum of corresponding matching fingerprints.

16. The method according to claim 11 wherein sets of corresponding identical fingerprints in the query document and one of the documents of the document collection represent regions of matching document content, while clusters of non-corresponding fingerprints in either image document represent document content differences.

17. The method according to claim 11 wherein the indexing includes: processing a sequence of page images, computing visual fingerprints; and adding the visual finger prints and their locations to the fingerprint lookup index.

18. A computer-readable medium having computer-executable instructions for detecting duplicate document content, comprising:

receiving a query document page;

holding a plurality of documents in a document collection storage;

comparing two-dimensional fingerprints of the query document page with two-dimensional fingerprints of pages of the documents in the document collection storage using a two-dimensional visual fingerprint comparison module, wherein the visual fingerprints are at least one of affine and perspective invariant visual fingerprints, and

automatically highlighting duplicate or different document content of the query document page and at least one document page of the stored documentation collection using a highlighting module, based on the results from the two-dimensional visual fingerprint comparison module;

wherein the automatically highlighting is accomplished by calculating of a mean and standard deviation for two-dimensional locations of each cluster of non-corresponding fingerprints of the query document page and one of the document pages of the document collection.

19. The method according to claim 1 wherein the visual two-dimensional fingerprints are organized and stored in a memory in a fan-tree data structure.

Assignments (10)