Image processing method and apparatus using self-adaptive binarization

• US 7,062,099 B2
• Filed: 07/15/2002
• Issued: 06/13/2006
• Est. Priority Date: 07/31/2001
• Status: Active Grant

• Current Assignee: Canon Kabushiki Kaisha (Canon Inc.)
• Original Assignee: Canon Kabushiki Kaisha (Canon Inc.)
• Inventors: Wang, Jie, Zhao, Hongsheng, Li, Yulin
• Primary Examiner(s): PATEL, KANJIBHAI B
• Application Number:
  US10/194,263
• Publication Number:
  US 20030031366A1
• Time in Patent Office: 1,429 Days
• Field of Search: 382/266, 382/269, 382/273, 382/274, 382/270, 382/252, 382/190, 382/197, 382/205, 382/237, 382171-173, 382/199, 358/455, 358464-466, 358/3.04, 358/1.15, 358/3.13, 358/521
• US Class Current: 382/237
• CPC Class Codes: G06T 2207/20012 Locally adaptive
• CPC Class Codes: G06T 2207/20021 Dividing image into blocks, subimages or windows
• CPC Class Codes: G06T 7/12 Edge-based segmentation
• CPC Class Codes: G06V 30/10 Character recognition
• CPC Class Codes: G06V 30/162 Quantising the image signal

Abstract

The present invention provides a unique method, apparatus, system and storage medium for image binarization process, wherein an image to be processed is divided into a plurality of sub-images and an binarization threshold for each of the sub-images is determined based on the gray-levels of the edge pixels detected in respective sub-image. The image processing method of the present invention comprises: calculating a gray-level statistical distribution of pixels of an image, detecting edge pixels in an image based on an edge criterion corresponding to the gray-level statistical distribution, dividing the image into a plurality of sub-images; determining a binarization threshold for each of the sub-images based on the gray-levels of edge pixels detected in the sub-image; and performing binarization process for each of the sub-images based on the binarization threshold of the sub-image.

Claims

1. An image processing method for processing an image with a gray-level distribution of pixels, comprising the steps of:

calculating a statistical distribution of the gray-level of the pixels of said image;

detecting edge pixels in the image in accordance with an edge criterion corresponding with the statistical distribution of gray-level;

dividing said image into a plurality of sub-images;

determining a binarizing threshold for each of the sub-images in accordance with the gray-levels of the edge pixels detected in the same sub-image; and

binarizing each of the sub-images in accordance with the binarizing threshold determined for the same sub-image.

2. The image processing method of claim 1, further comprising the step of:

determining a global minimum threshold for said image, wherein said global minimum threshold prevents the ratio of the number of pixels with gray-levels lower than said global minimum threshold to the total number of pixels in the image from becoming lower than a preset value;

wherein the step of determining a binarizing threshold for each sub-image comprises;

determining a candidate binarizing threshold of a current sub-image in accordance with the average gray-level of the edge pixels detected in the current sub-image;

comparing said global minimum threshold with a second candidate binarizing threshold determined based on said candidate binarizing threshold of the current sub-image; and

determining the binarizing threshold of the current sub-image on the basis of the result of said comparing step.

3. The image processing method of claim 2 further comprising the step of detecting a black block in the image.

4. The image processing method of claim 3, wherein, in said global minimum threshold determining step:

if a black block is detected in said black block detecting step, said global minimum threshold is determined in accordance with a reference threshold of said black block; and

if no black block is detected in said black block detecting step, said global minimum threshold is determined on the basis of said gray-level statistical distribution.

5. The image processing method of claim 4, further comprising the step of:

determining said global minimum threshold on the basis of the positions of peaks and valleys in said gray-level statistical distribution.

6. The image processing method of claim 4, wherein determining said global minimum threshold on the basis of said gray-level statistical distribution comprises the step of:

calculating the number of peaks in the gray-level statistical distribution.

7. The image processing method of claim 4, wherein said step of determining the binarizing threshold for each sub-image further comprises the step of:

using the average value of the gray-levels of the edge pixels detected in said sub-image as said candidate binarizing threshold for the current sub-image.

8. The image processing method of claim 4, wherein said step of determining the binarizing threshold for each sub-image further comprises the steps of:

determining whether the current sub-image is within a black block; and

adjusting the binarizing threshold of the current sub-image in accordance with the reference threshold of the current black block if it is determined that the current sub-image is within a black block.

9. The image processing method of claim 4, wherein said step of determining the binarizing threshold for each sub-image further comprises the steps of:

determining whether the current sub-image is a noise block; and

if it is determined that the current sub-image is a noise block, allowing the binarizing threshold for the current sub-image to become independent to said candidate binarizing threshold for the current sub-image.

10. The image processing method of claim 7, wherein said step of determining the binarizing threshold for each sub-image further comprises the steps of:

determining whether the current sub-image is a background block; and

if it is determined that the current sub-image is a background block, allowing the binarizing threshold for the current sub-image to become independent to said candidate binarizing threshold for the current sub-image.

11. The image processing method of claim 4, wherein said step of determining the binarizing threshold for each sub-image further comprises the steps of:

determining whether the current sub-image is a white block;

if it is determined that the current sub-image is a white block, determining whether the current sub-image is within a black block; and

if it is determined that the current sub-image is within a black block, adjusting the binarizing threshold of the current sub-image in accordance with the reference threshold of the black block.

12. The image processing method of claim 4, wherein said step of determining the binarizing threshold for each sub-image further comprises the steps of:

predicting whether the current sub-image will become abnormal after binarization; and

if it is predicted that the current sub-image will become abnormal after binarization, allowing the binarizing threshold for the current sub-image to become independent to said candidate binarizing threshold for the current sub-image.

13. The image processing method according to any one of claims 2 to 12, wherein the method further comprises the step of:

ensuring that the binarizing threshold of the current sub-image is equal to or larger than said global minimum threshold.

14. The image processing method according to any one of claims 1 to 12, wherein said edge pixel detecting step comprises:

calculating the magnitude of gradient of the gray-level distribution at a target pixel in the image;

determining the direction of said gradient;

comparing the calculated magnitude of gradient to said edge criterion; and

when said magnitude of gradient is larger than said edge criterion, setting a pixel near said target pixel as an edge pixel, wherein the direction from said target pixel to said pixel near the target pixel is the direction of the gradient at the target pixel.

15. The image processing method of claim 14, wherein said edge criterion is decided by the steps of:

analyzing said image in accordance with said statistical distribution of gray-level;

determining a dividing threshold for dividing said image into a foreground and a background in accordance with the result of said analyzing step; and

calculating said edge criterion based on the sum of gray-level of pixels of the foreground and the sum of gray-level of pixels of the background.

16. The image processing method of claim 15, wherein, in said analyzing step said image is classified, on the basis of said statistical distribution of gray-level, into one of the following image types:

a first image type in which the foreground and background is properly distributed;

a second image type in which the foreground is unclear; and

a third image type in which the background is unclear.

17. The image processing method according to any one of claims 3 to 12, wherein said black block detecting step comprises the steps of:

calculating the sum of gray-level of each line of pixels of the image, thereby obtaining a first sequence consisting of the sums for all the lines of pixels of the image;

calculating the sum of gray-level of each column of pixels of the image, thereby obtaining a second sequence consisting of the sums for all the columns of pixels of the image;

finding out extrema, if any, of the rate of variation in the first sequence and identifying the lines at which the extrema are located;

finding out extrema, if any, of the rate of variation in the second sequence and identifying the columns at which the extrema are located;

dividing the image with the lines and columns at which the extrema are located, thereby obtaining a plurality of divided image areas; and

determining whether the average gray-level of all the pixels in each of said divided image areas is lower than a preset black block discriminating threshold.

18. The image processing method of claim 17, further comprising the steps of:

determining whether any of the divided image areas is a broken-down area; and

when it is determined a divided image areas is a broken-down area, treating the divided image area as a non-black block.

19. The image processing method of claim 17, wherein said steps of finding out extrema of the rate of variation in the first and second sequences each comprises the steps of:

performing a convolution operation for the respective one of said sequences, thereby obtaining two new sequences;

determining whether the absolute value of each of the items in the two new sequences is greater than a predetermined value; and

recording the lines/columns at which the absolute value of the items in the new sequences being greater than said predetermined value as the lines/columns at which said extrema exist.

20. The image processing method according to any one of claims 3 to 12, further comprising the step of:

performing two-dimensional interpolating treatment for pixels in the current sub-image and the neighboring sub-images thereof with the binarizing thresholds of the current sub-image and the neighboring sub-images thereof.

21. A method for processing an image with a gray-level distribution, comprising the steps of:

calculating the magnitude of gradient of the gray-level distribution at a target pixel in the image;

determining the direction of said gradient;

comparing the calculated magnitude of gradient to a predetermined value; and

when said magnitude of gradient is larger than said predetermined value, setting a pixel near said target pixel as an edge pixel, wherein the direction from said target pixel to said pixel near the target pixel is the direction of the gradient at the target pixel,

wherein said predetermined value is decided by the steps of;

calculating a statistical distribution of gray-level of said image;

analyzing said image in accordance with said statistical distribution of gray-level;

determining a dividing threshold for dividing said image into a foreground and a background in accordance with the result of said analyzing step; and

calculating the said edge criterion based on the sum of gray-level of pixels of the foreground and the sum of gray-level of pixels of the background.

22. The method of claim 21, wherein said gradient is a discrete gradient.

23. The method of claim 21, wherein said gradient is calculated using Sobel filter.

24. The image processing method of claim 21, wherein, in said analyzing step, on the basis of said statistical distribution of gray-level, said image is classified as one of the following image types:

a first image type in which the foreground and background is properly distributed;

a second image type in which the foreground is unclear; and

a third image type in which the background is unclear.

25. An image processing apparatus for processing an image having a gray-level distribution, comprising:

a) a gray-level distribution calculating means for calculating a statistical distribution of the gray-level of the pixels of said image;

b) an edge pixel detecting means for detecting edge pixels in the image in accordance with an edge criterion corresponding with the statistical distribution of gray-level;

c) an image dividing means for dividing said image into a plurality of sub-images;

d) a sub-image binarizing threshold determining means for determining a binarizing threshold for each of the sub-images in accordance with the gray-levels of the edge pixels detected in the same sub-image; and

e) a binarizing means for binarizing each of the sub-image in accordance with the binarizing threshold determined for the same sub-image.

26. The image processing apparatus of claim 25, further comprising:

f) a global minimum threshold determining means for determining a global minimum threshold for said image, said global minimum threshold prevents the ratio of the number of pixels with gray-levels lower than said global minimum threshold to the total number of pixels in the image from becoming lower than a preset value;

wherein said sub-image binarizing threshold determining means,

i) determines a candidate binarizing threshold of a current sub-image in accordance with the average gray-level of the edge pixels detected in the current sub-image;

ii) compares said global minimum threshold with a second candidate binarizing threshold determined based on said candidate binarizing threshold of the current sub-image; and

iii) determines the binarizing threshold of the current sub-image on the basis of the result of said comparing step.

27. The image processing apparatus of claim 26, further comprising means for detecting black block in the image.

28. The image processing apparatus of claim 27, wherein, said global minimum threshold determining means determines said global minimum threshold in accordance with a reference threshold of said black block when a black block is detected by said black block detecting means and determines said global minimum threshold on the basis of said gray-level statistical distribution when no black block is detected in said black block detecting means.

29. The image processing apparatus of claim 25, wherein said edge pixel detecting means comprises:

an image classifying means for classifying said image in accordance with the calculated gray-level distribution;

a dividing threshold determining means for determining a dividing threshold for dividing said image into a foreground and a background in accordance with the result of classification by said image classifying means; and

edge criterion calculating means for calculating said edge criterion based on the sum of gray-level of pixels of the foreground and the sum of gray-level of pixels of the background.

30. A computer-readable storage medium on which are stored computer-readable program codes for an image processing method for processing an image with a gray-level distribution of pixels, said program codes comprising:

a) code for calculating a statistical distribution of the gray-level of the pixels of said image;

b) code for detecting edge pixels in the image in accordance with an edge criterion corresponding with the statistical distribution of gray-level;

c) code for dividing said image into a plurality of sub-images;

d) code for determining a binarizing threshold for each of the sub-images in accordance with the gray-levels of the edge pixels detected in the same sub-image; and

e) code for binarizing each of the sub-image in accordance with the binarizing threshold determined for the same sub-image.

31. The computer-readable storage medium of claim 30, wherein said program code further comprises:

code for determining a global minimum threshold for said image, said global minimum threshold prevents the ratio of the number of pixels with gray-levels lower than said global minimum threshold to the total number of pixels in the image from becoming lower than a preset value;

wherein the code determining a binarizing threshold for each sub-image comprises;

code for determining a candidate binarizing threshold of a current sub-image in accordance with the average gray-level of the edge pixels detected in the current sub-image;

code for comparing said global minimum threshold with a second candidate binarizing threshold determined based on said candidate binarizing threshold of the current sub-image; and

code for determining the binarizing threshold of the current sub-image on the basis of the result of said comparing step.

32. The computer-readable storage medium of claim 30, wherein said program code further comprises code for detecting black block in the image.

33. The computer-readable storage medium of claim 32, wherein the code for determining said global minimum threshold is for causing the following process:

when a black block is detected in the image by said code for detecting black block, determining said global minimum threshold in accordance with a reference threshold of said black block; and

when no black block is detected by said code for detecting black block, determining said global minimum threshold on the basis of said gray-level statistical distribution.