Method and system for character recognition

• US 8,953,886 B2
• Filed: 08/08/2013
• Issued: 02/10/2015
• Est. Priority Date: 12/03/2004
• Status: Active Grant

• Current Assignee: Kyocera Corporation
• Original Assignee: Google Inc. (Alphabet Inc.)
• Inventors: King, Martin T., Grover, Dale L., Kushler, Clifford A., Stafford-Fraser, James Quentin
• Primary Examiner(s): Seth, Manav
• Application Number:
  US13/961,934
• Publication Number:
  US 20140169675A1
• Time in Patent Office: 551 Days
• Field of Search: 382/135, 382137-140, 382181-231, 382/321, 345634-636, 345468-471
• US Class Current: 382/177
• CPC Class Codes: G06F 16/5846 using extracted text
• CPC Class Codes: G06F 16/93 Document management systems
• CPC Class Codes: G06V 20/62 Text, e.g. of license plates, overlay texts or captions on TV images
• CPC Class Codes: G06V 30/10 Character recognition
• CPC Class Codes: G06V 30/153 using recognition of characters or words
• CPC Class Codes: G06V 30/158 using character size, text spacings or pitch estimation
• CPC Class Codes: G06V 30/224 of printed characters having additional code marks or containing code marks
• CPC Class Codes: G06V 30/413 Classification of content, e.g. text, photographs or tables
• CPC Class Codes: G06V 30/414 Extracting the geometrical structure, e.g. layout tree; Block segmentation, e.g. bounding boxes for graphics or text

Abstract

Character recognition is described. In one embodiment, it may use matched sequences rather than character shape to determine a computer-legible result.

Claims

1. An article of manufacture comprising a non-transitory computer-readable medium with instructions encoded thereon, the instructions configured to cause one or more processors to perform a method comprising:

obtaining an image based on a document capture process performed on a rendered document;

identifying a portion of the image, the portion comprising a sequence of text units;

segmenting the portion of the image into a sequence of segmented sub-images, each segmented sub-image comprising a single text unit of the sequence of text units;

for each segmented sub-image of the sequence of segmented sub-images;

determining that one or more features of the segmented sub-image are classified as being similar to one or more corresponding features of a stored sub-image; and

based on determining that one or more features of the segmented sub-image are classified as being similar to one or more corresponding features of the stored sub-image, assigning to the segmented sub-image a text unit identity that is associated with the stored sub-image;

generating a representation of the portion of the image, based on the assigned text unit identities; and

identifying the sequence of segmented sub-images, based on the generated representation.

2. The article of manufacture of claim 1, wherein determining that one or more features of the segmented sub-image are classified as being similar to one or more corresponding features of the stored sub-image comprises:

identifying a likelihood that the segmented sub-image corresponds to the stored sub-image; and

determining that the likelihood meets a predetermined threshold.

3. The article of manufacture of claim 1, wherein the stored sub-image is stored by adding the stored sub-image to a template of sub-images.

4. The article of manufacture of claim 1, wherein determining that one or more features of the segmented sub-image are classified as being similar to one or more corresponding features of the stored sub-image comprises:

identifying a difference between the segmented sub-image and the stored sub-image;

identifying a pattern, based on the difference; and

determining that a size of the pattern meets a predetermined size threshold.

5. The article of manufacture of claim 1, wherein determining that one or more features of the segmented sub-image are classified as being similar to one or more corresponding features of the stored sub-image comprises:

decomposing the segmented sub-image into a first set of vectors;

decomposing the stored sub-image into a second set of vectors; and

determining that the first set of vectors and the second set of vectors meet a predetermined similarity threshold.

6. The article of manufacture of claim 1, wherein the image comprises an image of one or more words from the rendered document, each of the words comprising one or more text units.

7. The article of manufacture of claim 1, wherein segmenting a portion of the image into multiple segmented sub-images comprises identifying space between the sub-images.

8. A system, comprising:

one or more data processing apparatus; and

a computer-readable storage device including instructions executable by the data processing apparatus and upon such execution cause the data processing apparatus to perform operations comprising;

obtaining an image based on a document capture process performed on a rendered document;

identifying a portion of the image, the portion comprising a sequence of text units;

segmenting the portion of the image into a sequence of segmented sub-images, each segmented sub-image comprising a single text unit of the sequence of text units;

for each segmented sub-image of the sequence of segmented sub-images;

determining that one or more features of the segmented sub-image are classified as being similar to one or more corresponding features of a stored sub-image; and

based on determining that one or more features of the segmented sub-image are classified as being similar to one or more corresponding features of the stored sub-image, assigning to the segmented sub-image a text unit identity that is associated with the stored sub-image;

generating a representation of the portion of the image, based on the assigned text unit identities; and

identifying the sequence of segmented sub-images, based on the generated representation.

9. The system of claim 8, wherein determining that one or more features of the segmented sub-image are classified as being similar to one or more corresponding features of the stored sub-image comprises:

identifying a likelihood that the segmented sub-image corresponds to the stored sub-image; and

determining that the likelihood meets a predetermined threshold.

10. The system of claim 8, wherein the stored sub-image is stored by adding the stored sub-image to a template of sub-images.

11. The system of claim 8, wherein determining that one or more features of the segmented sub-image are classified as being similar to one or more corresponding features of the stored sub-image comprises:

identifying a difference between the segmented sub-image and the stored sub-image;

identifying a pattern, based on the difference; and

determining that a size of the pattern meets a predetermined size threshold.

12. The system of claim 8, wherein determining that one or more features of the segmented sub-image are classified as being similar to one or more corresponding features of the stored sub-image comprises:

decomposing the segmented sub-image into a first set of vectors;

decomposing the stored sub-image into a second set of vectors; and

determining that the first set of vectors and the second set of vectors meet a predetermined similarity threshold.

13. The system of claim 8, wherein the image comprises an image of one or more words from the rendered document, each of the words comprising one or more text units.

14. The system of claim 8, wherein segmenting a portion of the image into multiple segmented sub-images comprises identifying space between the sub-images.

15. A computer-implemented method, comprising:

obtaining an image based on a document capture process performed on a rendered document;

identifying a portion of the image, the portion comprising a sequence of text units;

segmenting the portion of the image into a sequence of segmented sub-images, each segmented sub-image comprising a single text unit of the sequence of text units;

for each segmented sub-image of the sequence of segmented sub-images;

determining that one or more features of the segmented sub-image are classified as being similar to one or more corresponding features of a stored sub-image; and

based on determining that one or more features of the segmented sub-image are classified as being similar to one or more corresponding features of the stored sub-image, assigning to the segmented sub-image a text unit identity that is associated with the stored sub-image;

generating a representation of the portion of the image, based on the assigned text unit identities; and

identifying the sequence of segmented sub-images, based on the generated representation.

16. The computer-implemented method of claim 15, wherein determining that one or more features of the segmented sub-image are classified as being similar to one or more corresponding features of the stored sub-image comprises:

identifying a likelihood that the segmented sub-image corresponds to the stored sub-image; and

determining that the likelihood meets a predetermined threshold.

17. The computer-implemented method of claim 15, wherein the stored sub-image is stored by adding the stored sub-image to a template of sub-images.

18. The computer-implemented method of claim 15, wherein determining that one or more features of the segmented sub-image are classified as being similar to one or more corresponding features of the stored sub-image comprises:

identifying a difference between the segmented sub-image and the stored sub-image;

identifying a pattern, based on the difference; and

determining that a size of the pattern meets a predetermined size threshold.

19. The computer-implemented method of claim 15, wherein determining that one or more features of the segmented sub-image are classified as being similar to one or more corresponding features of the stored sub-image comprises:

decomposing the segmented sub-image into a first set of vectors;

decomposing the stored sub-image into a second set of vectors; and

determining that the first set of vectors and the second set of vectors meet a predetermined similarity threshold.

20. The computer-implemented method of claim 15, wherein the image comprises an image of one or more words from the rendered document, each of the words comprising one or more text units.

21. The computer-implemented method of claim 15, wherein segmenting a portion of the image into multiple segmented sub-images comprises identifying space between the sub-images.