METHOD OF AND APPARATUS FOR IDENTIFYING SUBSETS OF INTERRELATED IMAGE OBJECTS FROM A SET OF IMAGE OBJECTS

• US 20020097914A1
• Filed: 12/09/1998
• Published: 07/25/2002
• Est. Priority Date: 12/09/1998
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: YAUNG, ALAN TSU-I
• Granted Patent: US 6,512,850 B2
• Time in Patent Office: Days
• US Class Current: 382/225
• CPC Class Codes: G06F 16/5838 using colour
• CPC Class Codes: G06V 30/242 Division of the character sequences into groups prior to recognition; Selection of dictionaries

Abstract

In one example given, a query for identifying interrelationships between image objects of a set of image objects is received from an input device. Each of a plurality of similarity values between all image objects of the set is compared with threshold criteria from the query. Clusters of image object identifiers are generated based on comparing and are visually displayed on a visual output device.

Claims

1. An apparatus, comprising:

a similarity value generator, said similarity value generator operative to generate similarity values between image features of all image objects of a set of image objects; and

a cluster generator, said cluster generator operative to generate at least one subset of image object identifiers based on the similarity values and threshold criteria associated with a query.

2. An apparatus according to claim 1, further comprising:

memory to store the set of image objects and the image features associated therewith;

an input device, said input device to receive the query for identifying at least one subset of interrelated image objects from the set of image objects; and

a visual output device, said visual output device to visual display the at least one subset of image object identifiers corresponding to the at least one subset of interrelated image objects.

3. An apparatus according to claim 1, further comprising:

a data comparator of said cluster generator, said data comparator to receive the threshold criteria and the similarity values from said similarity value generator; and

a cluster assignor of said cluster generator, said cluster assignor operative in connection with said data comparator.

4. An apparatus according to claim 1, comprising:

an image feature comparator of said similarity value generator.

5. A method of identifying subsets of interrelated image objects from a set of image objects, comprising:

comparing image features of each image object of the set with image features of all other image objects of the set;

generating a similarity value for each comparison; and

identifying subsets of interrelated image objects of the set, where each image object identified in a subset has a similarity value that meets threshold criteria with respect to at least one other image object identified in the same subset.

6. A method according to claim 5, further comprising:

comparing the threshold criteria with each one of the similarity values.

7. A method according to claim 5, further comprising:

receiving user input associated with the threshold criteria; and

visually displaying identification of the subsets of interrelated image objects.

8. A method according to claim 7, wherein comparing image features is responsive to receiving the user input.

9. A method of clustering image object identifiers, comprising:

(a) comparing threshold criteria with each one of a first plurality of similarity values, the first plurality of similarity values being between a first image object and other image objects of a set;

(b) in response to determining that a similarity value between the first image object and a second image object meets the threshold criteria from comparing in (a), comparing the threshold criteria with each one of a second plurality of similarity values, the second plurality of similarity values being between the second image object and other image objects of the set; and

(c) in response to determining that a similarity value between the second image object and a third image object meets the threshold criteria from comparing in (b), logically grouping image object identifiers associated with the first, second, and third image objects into a cluster.

10. A method according to claim 9, further comprising:

(d) receiving a query from an input device, the query associated with the threshold criteria; and

(e) visually displaying the logical grouping of image object identifiers at a visual output device.

11. A method according to claim 10, wherein comparing in (a) is responsive to receiving the query in (d).

12. A method according to claim 10, further comprising:

generating the first and the second pluralities of similarity values responsive to receiving the query in (d).

15. A computer software product according to claim 14, further comprising:

cluster generator code, said cluster generator code executable to generate at least one cluster of image object identifiers based on the similarity values and threshold criteria.

16. A computer software product according to claim 14, further comprising:

image feature comparator code of said similarity value generator code.

17. A computer software product according to claim 15, further comprising:

data comparator code of said cluster generator code, said data comparator code executable to compare the similarity values and the threshold criteria; and

cluster assignor code of said cluster generator code, said cluster assignor code executable in connection with said data comparator code.

18. A computer software product according to claim 14, wherein said similarity value generator code comprises code executable to generate an image query for each image object against all other image objects of the set.

20. A method according to claim 19, further comprising:

if the similarity rank between the image objects I and J meets the threshold criteria;

for each similarity rank between the image object I and an image object K, where K=1 to M except where K=J or K=I;

comparing the similarity rank between the image objects I and K with the threshold criteria;

if the similarity rank between the image objects I and K meets the threshold criteria;

assigning any image object identifiers associated with a cluster K to the cluster 3; and

after assigning, setting the cluster K to an empty set.

13. Software for generating clusters of image object identifiers, comprising:

primary compare software, said primary compare software executable to compare threshold criteria with each one of a first plurality of similarity values, the first plurality of similarity values being between a first image object and other image objects of the set; and

secondary compare software, said secondary compare software executable in response to determining from said primary compare software that a similarity value between the first image object and a second image object meets the threshold criteria, said secondary compare software executable to compare the threshold criteria with each one of a second plurality of similarity values, the second plurality of similarity values being between the second image object and other image objects of the set.

14. A computer software product, comprising:

similarity value generator code, said similarity value generator code executable to generate a similarity value for each image object and all other image objects of a set of at least four image objects in response to a single request.

19. A method of clustering image object identifiers associated with a set of M image objects, comprising:

for each image object J, where J=1 to M;

assigning an image object identifier associated with an image object J to a cluster J;

for each image object J, where J=1 to M;

for each similarity rank between the image object J and an image object I, where I=J+1 to M;

comparing the similarity rank between the image objects I and J with threshold criteria;

if the similarity rank between the image objects I and J meets the threshold criteria;

assigning any image object identifiers associated with a cluster I to a cluster J; and

after assigning, setting the cluster I to an empty set.