TEXT DETECTION USING IMAGE REGIONS

• US 20130058575A1
• Filed: 03/06/2012
• Published: 03/07/2013
• Est. Priority Date: 09/06/2011
• Status: Active Grant

• Current Assignee: Qualcomm, Inc.
• Original Assignee: Qualcomm, Inc.
• Inventors: KOO, HYUNG-IL, You, Kisun
• Granted Patent: US 8,942,484 B2
• Time in Patent Office: Days
• US Class Current: 382/176
• CPC Class Codes: G06V 10/56 relating to colour
• CPC Class Codes: G06V 20/63 Scene text, e.g. street names
• CPC Class Codes: G06V 30/10 Character recognition
• CPC Class Codes: G06V 30/1475 Inclination or skew detection or correction of characters or of image to be recognised

Abstract

A method includes receiving an indication of a set of image regions identified in image data. The method further includes, selecting image regions from the set of image regions for text extraction at least partially based on image region stability.

Claims

1. A computer-implemented method comprising:

receiving an indication of a set of image regions identified in image data; and

selecting image regions from the set of image regions for text extraction at least partially based on image region stability.

2. The computer-implemented method of claim 1, wherein each image region of the set of image regions is a blob that corresponds to a group of connected pixels of the image data having substantially similar pixel values.

3. The computer-implemented method of claim 1, wherein the image regions are selected further based at least partially on image region curvilinearity.

4. The computer-implemented method of claim 3, wherein the image region curvilinearity is determined based on a comparison of an image region size to an image region stroke width.

5. The computer-implemented method of claim 3, wherein selecting the image regions includes:

identifying an image region of the set of image regions satisfying a stability criterion; and

determining whether the identified image region satisfies a curvilinear criterion.

6. The computer-implemented method of claim 5, further comprising determining a stability measure of each image region of the set of image regions, and wherein identifying the image region satisfying the stability criterion comprises selecting a most stable image region of the set of image regions based on a comparison of the determined stability measures.

7. The computer-implemented method of claim 6, wherein each determined stability measure indicates a change in image region size with respect to a change in a threshold.

8. The computer-implemented method of claim 5, wherein selecting the image regions further includes, after determining that the identified image region satisfies the curvilinear criterion, removing from the set of image regions all image regions that overlap the identified image region.

9. The computer-implemented method of claim 1, further comprising:

determining a set of geometric figures that correspond to the selected image regions; and

determining a projection profile of the set of geometric figures.

10. The computer-implemented method of claim 1, wherein at least one image region of the set of image regions is selected for text extraction and wherein at least another image region of the set of image regions is not selected for text extraction.

11. The computer-implemented method of claim 1, wherein at least one image region of the set of image regions overlaps another image region of the set of image regions.

12. An apparatus comprising:

an image region extractor configured to generate an indication of a set of image regions identified in image data; and

a text image region selector configured to select image regions from the set of image regions for text extraction at least partially based on image region stability.

13. The apparatus of claim 12, wherein the text image region selector is further configured to select the image regions at least partially based on image region curvilinearity.

14. The apparatus of claim 13, wherein the text image region selector includes:

a selector configured to select a stable and curvilinear image region from the set of image regions; and

an overlap region remover configured to remove image regions from the set of image regions that overlap the selected stable and curvilinear image region.

15. The apparatus of claim 12, further comprising an image region abstractor configured to determine a set of geometric figures that correspond to the selected image regions.

16. The apparatus of claim 12, further comprising a text region detector configured to determine a projection profile of the set of geometric figures.

17. An apparatus comprising:

means for identifying a set of image regions in image data; and

means for selecting image regions from the set of image regions for text extraction at least partially based on image region stability.

18. The apparatus of claim 17, wherein the image regions are selected further based at least partially on image region curvilinearity.

19. A computer readable storage medium storing program instructions that are executable by a processor, the program instructions comprising:

code for receiving an indication of a set of image regions identified in image data; and

code for selecting image regions from the set of image regions for text extraction at least partially based on image region stability.

20. The computer-readable storage medium of claim 19, wherein the program instructions further comprise code for selecting the image regions further based at least partially on image region curvilinearity.

21. A computer-implemented method comprising:

identifying a geometric figure that corresponds to an image region, the image region corresponding to at least a portion of text identified in image data; and

determining a bounding box for the text based at least partially on the geometric figure.

22. The computer-implemented method of claim 21, further comprising:

in response to a first fitting error of the geometric figure with respect to the image region exceeding a threshold, identifying a plurality of geometric figures that correspond to the image region.

23. The computer-implemented method of claim 22, wherein the first fitting error corresponds to an area of the image region located outside the geometric figure compared to a total area of the image region.

24. The computer-implemented method of claim 22, further comprising performing projection profile analysis on the plurality of geometric figures in response to a second fitting error of the plurality of geometric figures being less than the first fitting error.

25. The computer-implemented method of claim 22, wherein the plurality of geometric figures consists of two geometric figures.

26. The computer-implemented method of claim 22, wherein the geometric figure is an ellipse and wherein the plurality of geometric figures includes a plurality of ellipses.

27. The computer-implemented method of claim 21, further comprising:

determining a projection profile of the geometric figure using multiple projection lines, wherein a value of the projection profile corresponding to each particular projection line is based on a distance between intersections of the particular projection line with the geometric figure.

28. The computer-implemented method of claim 21, further comprising:

identifying a plurality of geometric figures that correspond to image regions, each of the image regions corresponding to at least a portion of the text;

determining a first projection profile of the plurality of geometric figures to estimate a skew of the text; and

determining a second projection profile of the plurality of geometric figures to estimate a tilt of the text.

29. The computer-implemented method of claim 28, wherein determining the first projection profile includes determining multiple projection profiles corresponding to angles of multiple sets of parallel projection lines intersecting the plurality of geometric figures and selecting a projection profile from the multiple projection profiles that is smaller than any other projection profile of the multiple projection profiles.

30. The computer-implemented method of claim 29, wherein the skew corresponds to the angle of the parallel projection lines corresponding to a horizontal reference.

31. The computer-implemented method of claim 28, wherein determining the second projection profile includes determining multiple projection profiles corresponding to angles of multiple sets of parallel projection lines intersecting the plurality of geometric figures and selecting a projection profile from the multiple projection profiles that has less intersections of parallel projection lines with the plurality of geometric figures than any other projection profile of the multiple projection profiles.

32. The computer-implemented method of claim 21, further comprising:

extracting image regions from a set of image regions, each of the image regions corresponding to at least a portion of the text;

partitioning the extracted image regions into multiple clusters; and

determining a projection profile of each cluster independent of each of the other clusters.

33. The computer-implemented method of claim 32, wherein partitioning the extracted image regions is based at least in part on a stroke width of the text.

34. The computer-implemented method of claim 32, wherein partitioning the extracted image regions is based at least in part on a color of the text.

35. The computer-implemented method of claim 21, further comprising:

performing a first projection profile analysis on a first set of geometric figures that correspond to image regions identified in the image data to partition the first set of geometric figures into a second set of geometric figures and a third set of geometric figures; and

performing a second projection profile analysis on the second set of geometric figures to partition the second set of geometric figures into a fourth set of geometric figures and a fifth set of geometric figures.

36. The computer-implemented method of claim 35, wherein the third set of geometric figures corresponds to a first line of the text, wherein the fourth set of geometric figures corresponds to a second line of the text, and wherein the fifth set of geometric figures corresponds to a third line of the text.

39. The apparatus of claim 27, wherein the instructions are further executable by the processor to determine a projection profile of the geometric figure using multiple projection lines, wherein a value of the projection profile corresponding to each particular projection line is based on a distance between intersections of the particular projection line with the geometric figure.

40. The apparatus of claim 27, wherein the instructions are further executable by the processor to:

identify a plurality of geometric figures that correspond to image regions, each of the image regions corresponding to at least a portion of the text;

determine a first projection profile of the plurality of geometric figures to estimate a skew of the text; and

determine a second projection profile of the plurality of geometric figures to estimate a tilt of the text.

41. The apparatus of claim 27, wherein the instructions are further executable by the processor to:

extract image regions from a set of image regions, each of the image regions corresponding to at least a portion of the text;

partition the extracted image regions into multiple clusters; and

determine a projection profile of each cluster independent of each of the other clusters.

42. The apparatus of claim 27, wherein the instructions are further executable by the processor to:

perform a first projection profile analysis on a first set of geometric figures that correspond to image regions identified in the image data to partition the first set of geometric figures into a second set of geometric figures and a third set of geometric figures; and

perform a second projection profile analysis on the second set of geometric figures to partition the second set of geometric figures into a fourth set of geometric figures and a fifth set of geometric figures.

37. An apparatus comprising:

a processor; and

a memory storing instructions executable by the processor to;

identify a geometric figure that corresponds to an image region, the image region corresponding to at least a portion of text identified in image data; and

determine a bounding box for the text based at least partially on the geometric figure.

38. The apparatus of claim 37, wherein the instructions are further executable by the processor to, in response to a first fitting error of the geometric figure with respect to the image region exceeding a threshold, identify a plurality of geometric figures that correspond to the image region.