Automated Generation of Metadata for Mining Image and Text Data

• US 20100223276A1
• Filed: 03/27/2008
• Published: 09/02/2010
• Est. Priority Date: 03/27/2007
• Status: Active Grant

• Current Assignee: Edward J. Wegman, Faleh Jassem Al-Shameri
• Original Assignee: Edward J. Wegman, Faleh Jassem Al-Shameri
• Inventors: Wegman, Edward J., Al-Shameri, Faleh Jassem
• Granted Patent: US 8,145,677 B2
• Time in Patent Office: Days
• US Class Current: 707/769
• CPC Class Codes: G06F 16/2264 Multidimensional index structures
• CPC Class Codes: G06F 16/2465 Query processing support for facilitating data mining operations in structured databases
• CPC Class Codes: G06F 16/3347 using vector based model
• CPC Class Codes: G06V 10/54 relating to texture
• CPC Class Codes: G06V 20/13 Satellite images

Abstract

A tangible computer readable medium encoded with instructions for automatically generating metadata, wherein said execution of said instructions by one or more processors causes said “one or more processors” to perform the steps comprising: a. creating at least one feature vector for each document in a dataset; b. extracting said one feature vector; c. recording said feature vector as a digital object; and d. augmenting metadata using said digital object to reduce the volume of said dataset, said augmenting capable of allowing a user to perform a search on said dataset.

Claims

1. A tangible computer readable medium encoded with instructions for automatically generating metadata, wherein said execution of said instructions by one or more processors causes said “

one or more processors”

to perform the steps comprising;

a. creating at least one feature vector for each document in a dataset;

b. extracting said one feature vector;

c. recording said feature vector as a digital object;

d. augmenting metadata using said digital object to reduce the volume of said dataset, said augmenting capable of allowing a user to perform a search on said dataset.

2. A tangible computer readable medium according to claim 1, wherein said feature vector is one of the following features:

a. a topic feature;

b. a discriminating word feature;

c. a verb feature;

d. a multigram feature; and

e. a combination thereof.

3. A tangible computer readable medium according to claim 1, wherein said feature vector reflects semantic content in each of said document for said dataset prior to extraction, said dataset comprising text data.

4. A tangible computer readable medium according to claim 3, wherein said text data is denoised to remove stopper words and stemming words.

5. A tangible computer readable medium according to claim 3, further including linking said document having similar said feature vector, said linking being based one of the following:

a. empirical patterns;

b. statistical patterns;

c. model-based patterns;

d. clustering;

e. Minimal Spanning Tree; and

f. visualization.

6. A tangible computer readable medium according to claim 1, wherein said dataset includes image data.

7. A tangible computer readable medium according to claim 6, wherein said image data is remote sensing data.

8. A tangible computer readable medium according to claim 6, wherein said feature vector extracted from image data uses a grey-level co-occurrence matrix.

9. A tangible computer readable medium according to claim 8, wherein grey-level co-occurrence matrix includes at least one measure, said measure comprising:

a. homogeneity;

b. contrast;

c. dissimilarity;

d. entropy;

e. energy;

f. angular second moment;

g. histogram-based contrast;

h. alternate vegetation index;

i. normalized difference vegetation index;

j. occurrence of linear features;

k. occurrence of circular features; and

l. a combination thereof.

10. A tangible computer readable medium according to claim 9, wherein said measure is adaptable as a new relevant feature.

11. A tangible computer readable medium according to claim 6, wherein a query language is used to search for at least one image in said image data.

12. An automated metadata generation system comprising:

a. a feature vector creator, configured for creating at least one feature vector for each document in a dataset;

b. a feature vector extractor, configured for extracting said one feature vector;

c. a digital object recorder, configured for recording said feature vector as a digital object; and

d. a metadata augmenter, configured for augmenting metadata using said digital object to reduce the volume of said dataset, said augmenting capable of allowing a user to perform a search on said dataset.

13. An automated metadata generation system according to claim 12, wherein said feature vector is one of the following features:

a. a topic feature;

b. a discriminating word feature;

c. a verb feature;

d. a multigram feature; and

e. a combination thereof.

14. An automated metadata generation system according to claim 12, wherein said feature vector reflects semantic content in each of said document for said dataset prior to extraction, said dataset comprising text data.

15. An automated metadata generation system according to claim 14, wherein said text data is denoised to remove stopper words and stemming words.

16. An automated metadata generation system according to claim 14, further including linking said document having similar said feature vector, said linking being based one of the following:

a. empirical patterns;

b. statistical patterns;

c. model-based patterns;

d. clustering;

e. Minimal Spanning Tree; and

f. visualization.

17. An automated metadata generation system according to claim 12, wherein said dataset includes image data.

18. An automated metadata generation system according to claim 17, wherein said image data is remote sensing data.

19. An automated metadata generation system according to claim 17, wherein said feature vector extracted from image data uses a grey-level co-occurrence matrix.

20. An automated metadata generation system according to claim 19, wherein grey-level co-occurrence matrix includes at least one measure, said measure comprising:

a. homogeneity;

b. contrast;

c. dissimilarity;

d. entropy;

e. energy;

f. angular second moment;

g. histogram-based contrast;

h. alternate vegetation index;

i. normalized difference vegetation index;

j. occurrence of linear features;

k. occurrence of circular features; and

l. a combination thereof.

21. An automated metadata generation system according to claim 20, wherein said measure is adaptable as a new relevant feature.

22. An automated metadata generation system according to claim 17, wherein a query language is used to search for at least one image in said image data.