Image Analysis System and Method Using Image Recognition and Text Search

• US 20120177297A1
• Filed: 01/12/2011
• Published: 07/12/2012
• Est. Priority Date: 01/12/2011
• Status: Active Grant

• Current Assignee: Yahoo AD Tech LLC
• Original Assignee: Yahoo! Inc. (Apollo Global Management, Inc.)
• Inventors: Everingham, James R.
• Granted Patent: US 9,384,408 B2
• Time in Patent Office: Days
• US Class Current: 382/229
• CPC Class Codes: G06V 20/62 Text, e.g. of license plates, overlay texts or captions on TV images

Abstract

Provided herein are systems and method for obtaining contextual information of an image published on a digital medium. The methods and systems disclosed herein generally identify and analyze the image to obtain image descriptors corresponding to the image. The methods also identify and analyze text published proximate to the image to obtain textual descriptors, which function to describe, identify, index, or name the image or content within the image. The textual descriptors are then matched to the image descriptors to provide contextual information of the published image.

Claims

1. A method of obtaining contextual information for an image published on a digital medium, the method comprising:

obtaining an image tag from an image published on a digital medium;

obtaining a textual tag from text published proximate to the image; and

matching the textual tag with the image descriptor to obtain contextual information for the image.

2. The method of claim 1, wherein the method further comprises identifying the image published on a digital medium.

3. The method of claim 1, wherein the image is analyzed with an image recognition engine.

4. The method of claim 1, wherein the method further comprises identifying the text published proximate to the image.

5. The method of claim 1, wherein the textual tag functions to describe, identify, index, or name the image or content within the image.

6. A method of obtaining contextual information for an image published on a digital medium, the method comprising:

(a) identifying an image published on a digital medium;

(b) analyzing the image with an image recognition engine to obtain an image tag corresponding to the image;

(c) identifying text published proximate to the image on the digital medium;

(d) analyzing the text from step (c) to obtain a textual tag; and

(e) matching the textual tag with the image tag to obtain contextual information of the image.

7. The method of claim 6, wherein the textual tag is a proper noun.

8. The method of claim 7, wherein the image tag is a common noun.

9. The method of claim 8, further comprising:

(f) maintaining a database of proper nouns and corresponding common nouns.

10. The method of claim 9, further comprising:

(g) conducting a match query against the database to identify matching image tags and textual tags.

11. The method of claim 6, wherein the image recognition engine creates at least one tag for the image, and wherein the tag includes the image tag and positional information corresponding to the image tag.

12. The method of claim 11, wherein the positional information includes coordinates selected from the group consisting of:

a real position coordinate, a boundary area coordinate, a coordinate that outlines content within the image, and any combination thereof.

13. The method of claim 6, wherein step (d) further comprises employing a language-independent proximity pattern matching algorithm.

14. The method of claim 6, wherein the textual tag functions to describe, identify, index, or name the image or content within the image.

15. A system for obtaining contextual information of an image published on a digital medium, the system comprising:

(a) means for identifying an image published on a digital medium;

(b) means for analyzing the image to obtain an image tag corresponding to the image;

(c) means for identifying text published proximate to the image on the digital medium;

(d) means for analyzing the text from step (c) to obtain a textual tag that functions to describe, identify, index, or name the image or content within the image; and

(e) means for matching the textual tag with the image tag.

16. A system for obtaining contextual information of an image published on a digital medium, the system comprising:

an identification module configured to identify an image published on a digital medium and text published proximate to the image; and

a processor that receives and analyzes the image and text to obtain a contextual tag by matching an image tag and textual tag corresponding to the image.

17. The system of claim 16, wherein the textual tag functions to describe, identify, index, or name the image or content within the image.

18. A method of contextual advertising, the method comprising:

(a) identifying an image published on a digital medium;

(b) analyzing the image with an image recognition engine to obtain an image tag corresponding to the image;

(c) identifying text published proximate to the image on the digital medium;

(d) analyzing the text from step (c) to obtain a textual tag;

(e) matching the textual tag with the image tag, wherein the linked textual tag and image tag serve as a contextual tag for the image; and

(f) providing an advertising creative based on the contextual tag.

19. The method of claim 18, wherein the tag includes positional information corresponding to the image tag.

20. The method of claim 19, wherein the positional information includes coordinates selected from the group consisting of:

a real position coordinate, a boundary area coordinate, a coordinate that outlines content within the image, and any combination thereof.

21. The method of claim 18, wherein step (d) further comprises employing a language-independent proximity pattern matching algorithm.

22. The method of claim 18, wherein the textual tag functions to describe, identify, index, or name the image or content within the image.

23. A computer-readable storage medium, comprising:

instructions executable by at least one processing device that, when executed, cause the processing device to

(a) identify an image published on a digital medium;

(b) analyze the image with an image recognition engine to obtain an image tag corresponding to the image;

(c) identify text published proximate to the image on the digital medium;

(d) analyze the text to obtain a textual tag that functions to describe, identify, index, or name the image or content within the image; and

(e) match the textual tag with the image tag.