SCANBOX

• US 20160012287A1
• Filed: 09/22/2015
• Published: 01/14/2016
• Est. Priority Date: 09/10/2013
• Status: Active Grant

• Current Assignee: Dropbox, Inc.
• Original Assignee: Dropbox, Inc.
• Inventors: Chajed, Tej, Welinder, Peter, Babekno, Boris, Simeonov, Dimitar
• Granted Patent: US 9,558,401 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06T 2207/20061 Hough transform
• CPC Class Codes: G06V 10/243 by compensating for image skew or non-uniform image deformations
• CPC Class Codes: G06V 10/48 by mapping characteristic values of the pattern into a parameter space, e.g. Hough transformation
• CPC Class Codes: G06V 30/413 Classification of content, e.g. text, photographs or tables

Abstract

Embodiments are provided for content item classification. In some embodiments, an image for classification is received, a compact representation for the image having values indicative of pixel values within the received image is generated, a plurality of angle measurements for possible edges of at least one potential document within the received image are determined, and the image is classified using said compact representation and said plurality of angle measurements

Claims

1. A method for content item classification, comprising:

receiving a digital image;

identifying, by at least one processor, a plurality of angle measurements for possible page edges of at least one potential document within the received image, wherein identifying the plurality of angle measurements comprises;

identifying, by the at least one processor, one or more edge candidates of the at least one potential document; and

calculating the plurality of angle measurements based on a vector extending from a selected origin to a point on each of the one or more edge candidates;

determining, based on the identified plurality of angle measurements, whether the image contains a document; and

in response to determining that the image contains a document, classifying the image as a document-containing image.

2. The method as recited in claim 1, further comprising converting the plurality of angle measurements to polar coordinates.

3. The method as recited in claim 2, further comprising determining, based on the polar coordinates, a location for each page edge of the document within the received digital image.

4. The method as recited in claim 1, wherein identifying one or more edge candidates of the at least one potential document comprises calculating gradient values for a plurality of points within the received digital image.

5. The method as recited in claim 4, wherein identifying one or more edge candidates of the at least one potential document further comprises identifying a location for each of the one or more edge candidates within the received digital image based on the gradient values.

6. The method as recited in claim 4, wherein calculating the plurality of angle measurements comprises calculating polar coordinates for the point and one or more additional points on each of the one or more edge candidates based on the calculated gradient values.

7. The method as recited in claim 6, further comprising eliminating at least one of the one or more edge candidates as a page edge for the document based on the plurality of angle measurements.

8. The method as recited in claim 6, further comprising confirming at least one of the one or more edge candidates as a page edge for the document based on the plurality of angle measurements.

9. A non-transitory computer-readable medium containing instruction that, when executed by at least one processor of a computing device, cause the computing device to:

receive a digital image;

identify a plurality of angle measurements for possible page edges of at least one potential document within the received image, wherein identifying the plurality of angle measurements comprises;

identifying one or more edge candidates of the at least one potential document; and

calculating the plurality of angle measurements based on a vector extending from a selected origin to a point on each of the one or more edge candidates;

determine, based on the identified plurality of angle measurements, whether the image contains a document; and

in response to determining that the image contains a document, classify the image as a document-containing image.

10. The non-transitory computer-readable medium of claim 9, further comprising instructions that, when executed by the at least one processor of the computing device, cause the computing device to convert the plurality of angle measurements to polar coordinates.

11. The non-transitory computer-readable medium of claim 10, further comprising instructions that, when executed by the at least one processor of the computing device, cause the computing device to determine, based on the polar coordinates, a location for each page edge of the document within the received digital image.

12. The non-transitory computer-readable medium of claim 9, wherein identifying one or more edge candidates of the at least one potential document comprises calculating gradient values for a plurality of points within the received digital image.

13. The non-transitory computer-readable medium of claim 12, wherein identifying one or more edge candidates of the at least one potential document further comprises identifying a location for each of the one or more edge candidates within the received digital image based on the gradient values.

14. The non-transitory computer-readable medium of claim 12, wherein calculating the plurality of angle measurements comprises calculating polar coordinates for the point and one or more additional points on each of the one or more edge candidates based on the calculated gradient values.

15. A system for content item classification, the system comprising:

one or more processors; and

memory containing instructions thereon that, when executed, cause the one or more processors to;

receive a digital image;

identify a plurality of angle measurements for possible page edges of at least one potential document within the received image, wherein identifying the plurality of angle measurements comprises;

identifying one or more edge candidates of the at least one potential document; and

calculating the plurality of angle measurements based on a vector extending from a selected origin to a point on each of the one or more edge candidates;

determine, based on the identified plurality of angle measurements, whether the image contains a document; and

in response to determining that the image contains a document, classify the image as a document-containing image.

16. The system of claim 15, further comprising instructions that, when executed, cause the one or more processors to convert the plurality of angle measurements to polar coordinates.

17. The system of claim 16, further comprising instructions that, when executed, cause the one or more processors to determine, based on the polar coordinates, a location for each page edge of the document within the received digital image.

18. The system of claim 15, wherein identifying one or more edge candidates of the at least one potential document comprises calculating gradient values for a plurality of points within the received digital image.

19. The system of claim 18, wherein identifying one or more edge candidates of the at least one potential document further comprises identifying a location for each of the one or more edge candidates within the received digital image based on the gradient values.

20. The system of claim 18, wherein calculating the plurality of angle measurements comprises calculating polar coordinates for the point and one or more additional points on each of the one or more edge candidates based on the calculated gradient values.