TEXT ENHANCEMENT OF A TEXTUAL IMAGE UNDERGOING OPTICAL CHARACTER RECOGNITION

• US 20110222768A1
• Filed: 03/10/2010
• Published: 09/15/2011
• Est. Priority Date: 03/10/2010
• Status: Active Grant

• Current Assignee: Zhigu Holdings Limited
• Original Assignee: Microsoft Corporation
• Inventors: Nijemcevic, Djordje, Dresevic, Bodin, Galic, Sasa
• Granted Patent: US 8,526,732 B2
• Time in Patent Office: Days
• US Class Current: 382/170
• CPC Class Codes: G06V 2201/01 Solutions for problems related to non-uniform document background
• CPC Class Codes: G06V 30/10 Character recognition
• CPC Class Codes: G06V 30/162 Quantising the image signal
• CPC Class Codes: G06V 30/18076 by analysing connectivity, e.g. edge linking, connected component analysis or slices

Abstract

A method for enhancing a textual image for undergoing optical character recognition begins by receiving an image that includes native lines of text. A background line profile is determined which represents an average background intensity along the native lines in the image. Likewise, a foreground line profile is determined which represents an average foreground background intensity along the native lines in the image. The pixels in the image are assigned to either a background or foreground portion of the image based at least in part on the background line profile and the foreground line profile. The intensity of the pixels designated to the background portion of the image is adjusted to a maximum brightness so as to represent a portion of the image that does not include text.

Claims

1. A method for enhancing a textual image for undergoing optical character recognition, comprising:

receiving an image that includes native lines of text;

determining a background line profile that represents an average background intensity along the native lines in the image;

determining a foreground line profile that represents an average foreground background intensity along the native lines in the image;

assigning pixels in the image to either a background or foreground portion of the image based at least in part on the background line profile and the foreground line profile; and

adjusting an intensity of the pixels designated to the background portion of the image to a prescribed intensity representing a portion of the image that does not include text.

2. The method of claim 1 wherein assigning the pixels further comprises:

assigning pixels in the image having an intensity less than a selected parameter to a foreground portion of the image, wherein the selected parameter is determined at least in part from the background line profile and the foreground line profile; and

assigning pixels in the image having an intensity greater than the selected parameter to a background portion of the image.

3. The method of claim 2 wherein the selected parameter is an average of the background line profile and the foreground line profile.

4. The method of claim 2 wherein the selected parameter is adjustable.

6. The method of claim 1 further comprising representing the native lines of text in edge space.

7. The method of claim 6 further comprising detecting connected components in the native lines represented in edge space.

8. The method of claim 7 wherein determining the background and foreground line profiles includes determining a plurality of background and foreground intensity values for a plurality of localized regions along each of the native lines represented in edge space, wherein each localized region has dimensions less than or equal to predefined values.

9. The method of claim 8 wherein determining the plurality of background and foreground intensity values for the plurality of localized regions further comprises:

for each localized region, preparing a histogram of intensity values of pixels within bounding boxes that each encompass a connected component located in the respective localized region, wherein the histogram has a dominant peak that includes pixels having an intensity within a defined range; and

for each bounding box, assigning pixels having intensity values within the defined range to the background intensity values and assigning remaining pixels to the foreground intensity values.

10. The method of claim 8 wherein the localized regions have a height about equal to a height of the native lines and a width less than a predetermined multiple of the height of the native line.

5. The method of 1 further comprising estimating an average or median stroke width of connected components representing text in the image.

11. A method for enhancing a textual image for undergoing optical character recognition, comprising:

receiving an image that includes native lines of text;

determining a background line profile that represents an average background intensity along the native lines in the image;

determining a foreground line profile that represents an average foreground background intensity along the native lines in the image;

estimating an average or median stroke width of connected components representing text in the image; and

enhancing text quality if the average or median stroke width is less than a threshold value.

12. The method of claim 11 further comprising estimating an average or median stroke width of the connected components representing text in the image by determining a local stroke width for each pixel assigned to the foreground intensity values and enhancing text quality if the average or median stroke width is less than a threshold value.

13. The method of claim 11 wherein enhancing text quality includes performing image interpolation to increase image resolution.

14. The method of claim 13 wherein the image interpolation is bilinear interpolation that maps each pixel in the image to a set of at least two pixels.

15. The method of claim 11 further comprising:

assigning pixels in the image having an intensity less than a selected parameter to a foreground portion of the image, wherein the selected parameter is determined at least in part from the background line profile and the foreground line profile;

assigning pixels in the image having an intensity greater than the background line profile to the background portion of the image; and

assigning selected pixels having an intensity between the selected parameter and the background line profile to base connected components of the native lines represented in edge space by incrementally adding sets of pixels within increasingly higher intensities, wherein each set defines pixels having an intensity within one of a finite set of intensity intervals.

16. The method of claim 15 further comprising terminating assigning selected pixels to at least one of the base connected components when a distance between an edge pixel and a skeleton pixel of the base connected component is greater than half the average or median stroke width.

17. A system that enhances a textual image undergoing optical character recognition, comprising:

an input component for receiving an input image having native lines of text;

a background and foreground line profiling component for determining background and foreground line profiles which respectively represent average background and foreground intensities in the image;

an image quality detection component for determining if the quality of the image is sufficient to assign pixels to the background or foreground portion of the image;

a thresholding component 208 for assigning pixels to either a background or foreground portion of the image based on the background and foreground line profiles; and

a stroke enhancement component 210 for enhancing strokes of connected components representing text in the image if the image quality detection component determines that the quality of the image is below a threshold value.

18. The system of claim 17 wherein the image quality detection component determines the quality of the image by estimating an average or median stroke width of the connected components representing text in the image and comparing the average or median stroke width to the threshold value.

19. The system of claim 17 wherein the background and foreground line profile component is further configured to determine a plurality of background and foreground intensity values for a plurality of localized regions along each of the native lines represented in edge space, wherein each localized region has dimensions less than or equal to predefined values.

20. The system of claim 19 wherein the background and foreground line profile component is further configured to determine the plurality of background and foreground intensity values for the plurality of localized regions by:

for each localized region, preparing a histogram of intensity values of pixels within bounding boxes that each encompass a connected component located in the respective localized region, wherein the histogram has a dominant peak that includes pixels having an intensity within a defined range; and

for each bounding box, assigning pixels having intensity values within the defined range to the background intensity values and assigning remaining pixels to the foreground intensity values.