Method and Systems for Processing Text Found in Images

• US 20110019869A1
• Filed: 01/21/2010
• Published: 01/27/2011
• Est. Priority Date: 09/26/2000
• Status: Active Grant

• Current Assignee: Digimarc Corporation
• Original Assignee: Digimarc Corporation
• Inventors: Conwell, William Y.
• Granted Patent: US 8,644,546 B2
• Time in Patent Office: Days
• US Class Current: 382/100
• CPC Class Codes: G06F 16/51 Indexing; Data structures therefor; Storage structures
• CPC Class Codes: G06F 16/5846 using extracted text

Abstract

An image containing text (e.g., a surveillance camera photo that includes a vehicle license plate) is analyzed to determine the text (e.g., by an OCR technique). The recognized text is then stored in a database. The image is digitally watermarked with an identifier that associates the image with the database location where the text is stored. In addition to surveillance contexts, this technology can be employed in indexing the World Wide Web. Images used in web pages can be watermarked to link to associated text or other data. When the web page is crawled by an indexer, the watermark can be decoded and the associated data repository accessed to obtain information that can augment the web index for that page.

Claims

1. A method comprising:

receiving data corresponding to an image, the image including a depiction of text;

recognizing at least some of said depicted text; and

steganographically encoding a digital watermark in said image, said steganographically encoded digital watermark comprising a visually imperceptible carrier of information rather than a conspicuous carrier of information, such as a barcode, said digital watermark serving to associate said image with said recognized text.

2. The method of claim 1 in which said recognizing includes recognizing by an automated OCR process.

3. The method of claim 1 that includes storing said recognized text in a data repository, and wherein said watermark serves as to associate said image with said stored text.

4. The method of claim 1 wherein said digital watermark serves to directly encode at least a portion of said recognized text in said image.

5. The method of claim 1 in which said encoding follows said recognizing.

6. A method comprising:

receiving an electronic document, the document comprising a graphical representation of text, but not including ASCII data corresponding thereto;

analyzing said document for text information using an OCR process; and

digitally watermarking said electronic document with a visually imperceptible marking that conveys plural bits of information;

wherein said digital watermark associates the electronic document with the text information.

7. The method of claim 6 in which the electronic document comprises FAX data.

8. The method of claim 6 in which the electronic document comprises a PDF document.

9. The method of claim 6 in which receiving an electronic document comprises scanning a paper document on a platen, and producing graphical data corresponding thereto.

10. The method of claim 6 wherein said digital watermark directly encodes the electronic document with at least some of said text information.

11. The method of claim 6 that includes storing the text information in a data repository, and wherein the digital watermark associates the electronic document with said information in the data repository.

12. A system comprising:

a scanner for producing scan data corresponding to an original document;

an OCR engine for recognizing text from said scan data; and

a watermarker that alters an output from said apparatus to steganographically encode a digital watermark therein, the watermark comprising a visually imperceptible carrier of information rather than a conspicuous carrier of information, such as a barcode, the watermark serving to associate said output with said stored text.

13. An apparatus according to claim 12 wherein the output comprises a hardcopy page, and said watermark serves as a pointer to a memory location in which said recognized text is stored.

14. An apparatus according to claim 12, wherein the output comprises a hardcopy page, and said watermark serves to directly encode at least a portion of said recognized text in said output.

15. A method of building an index to a collection of web pages by reference to text and meta tags found therein, the method including the acts:

downloading an image from one of said web pages;

decoding a steganographically encoded digital watermark from the downloaded image;

through use of said decoded digital watermark, obtaining text associated with said downloaded image; and

augmenting the index in accordance with said obtained text.

16. The method of claim 15 in which the decoded digital watermark includes a database reference, and the method includes using said reference to obtain information from an associated database record.