Document image binarization method

• US 9,367,899 B1
• Filed: 05/29/2015
• Issued: 06/14/2016
• Est. Priority Date: 05/29/2015
• Status: Active Grant

• Current Assignee: Konica Minolta Laboratory U.S.A., Inc. (Konica Minolta Inc.)
• Original Assignee: Konica Minolta Laboratory U.S.A., Inc. (Konica Minolta Inc.)
• Inventors: Fang, Gang
• Primary Examiner(s): Do, Anh
• Application Number:
  US14/725,375
• Time in Patent Office: 382 Days
• Field of Search: 382/270, 382/290, 382/254, 382/112, 382/289, 382/173, 382/219, 382/168, 382/177, 382/164, 382/176, 382/199, 382/216, 358/1.9, 358/2.1, 358/462, 358/540
• US Class Current: 1/1
• CPC Class Codes: G06T 7/11 Region-based segmentation
• CPC Class Codes: G06T 7/136 involving thresholding
• CPC Class Codes: G06V 10/28 Quantising the image, e.g. histogram thresholding for discrimination between background and foreground patterns
• CPC Class Codes: G06V 10/457 by analysing connectivity, e.g. edge linking, connected component analysis or slices
• CPC Class Codes: G06V 30/40 Document-oriented image-based pattern recognition

Abstract

A method for binarization of document image using multi-threshold process to determine an optimum global binarization threshold for the image. The optimum binarization threshold is determined by binarizing the document multiple times using different threshold values, and calculating the statistics of the useful information and noise for each threshold value to select the optimum threshold value.

Claims

1. A method implemented in a data processing system which includes a processor and a memory, for binarizing a multi-bit document image, comprising:

(a) binarizing the document image a plurality of times, each time using one of a plurality of different binarization thresholds, to generate a plurality of corresponding binary images;

for each of the binary images,

(b) applying connected component analysis to the binary image to identify connected components in the binary image;

(c) identifying all connected components in the binary image that are larger than a threshold size and have fill rates higher than a fill rate threshold and removing all connected components contained within bounding boxes of the identified connected components; and

(d) counting a first number of connected components in the binary image that have sizes equal to or larger than a first threshold size, and counting a second number of connected components in the binary image that have sizes equal to or smaller than a second threshold size;

(e) based on the first number and the second number of each binary image, selecting one of the binary images as the optimum binary image; and

(f) outputting the optimum binary image.

2. The method of claim 1, wherein in the removing step (c), the threshold size is 150 by 150 pixels and the threshold fill rate is 50%.

3. The method of claim 1, wherein in the counting step (d), the first threshold size is 6 pixels and the second threshold size is 2 pixels.

4. The method of claim 1, wherein step (e) comprises:

(f) defining a plurality of groups of binary images from the plurality of binary images, wherein the plurality of binarization thresholds form an increasing or decreasing sequence, and wherein each group includes a plurality of binary images corresponding to consecutive binarization thresholds in the sequence;

(g) for each group, computing a first average value which is an average of the first numbers and a second average value which is an average of the second numbers;

(h) based on the first average values and the second average values of the plurality of groups, selecting one of the groups as an optimum group; and

(i) within the optimum group, based on the first number and the second number of each binary image, selecting one of the binary images as the optimum binary image.

5. The method of claim 4, wherein the selecting step (h) comprises:

(h1) if the second average values of the plurality of groups have only one local minimum, selecting the group corresponding to that second average value as the optimum group; and

(h2) if the second average values of the plurality of groups have two or more local minimums,

(h3) selecting a first group and a second group corresponding respectively to a first and a second one of the local minimums;

(h4) if the second average value of the first group is significantly greater than the second average value of the second group, or if the second average value of the first group less than the second average value of the second group and the first average value of the first group is significantly less than the first average value of the second group, selecting the second group as the optimum group; and

(h5) otherwise selecting the first group as the optimum group.

6. The method of claim 5, wherein the selecting step (h4) comprises:

if the second average value of the first group is greater than 1.25 times the second average value of the second group, or if the second average value of the first group less than the second average value of the second group and the first average value of the first group is less than two thirds of the first average value of the second group, selecting the second group as the optimum group.

7. The method of claim 4, wherein the selecting step (i) comprises:

calculating a difference value of the first number and the second number for each binary image; and

selecting one of the binary images that has a largest difference value as the optimum binary image.

8. The method of claim 4, wherein in the binarizing step (a), the document image is binarized 16 times using 16 binarization thresholds, and wherein the defining step (f) defines 7 groups of binary images.

9. A computer program product comprising a computer usable non-transitory medium having a computer readable program code embedded therein for controlling a computer, the computer readable program code configured to cause the computer to execute a process for binarizing a multi-bit document image, the process comprising:

(a) binarizing the document image a plurality of times, each time using one of a plurality of different binarization thresholds, to generate a plurality of corresponding binary images;

for each of the binary images,

(b) applying connected component analysis to the binary image to identify connected components in the binary image;

(c) identifying all connected components in the binary image that are larger than a threshold size and have fill rates higher than a fill rate threshold and removing all connected components contained within bounding boxes of the identified connected components; and

(d) counting a first number of connected components in the binary image that have sizes equal to or larger than a first threshold size, and counting a second number of connected components in the binary image that have sizes equal to or smaller than a second threshold size;

(e) based on the first number and the second number of each binary image, selecting one of the binary images as the optimum binary image; and

(f) outputting the optimum binary image.

10. The computer program product of claim 9, wherein in the removing step (c), the threshold size is 150 by 150 pixels and the threshold fill rate is 50%.

11. The computer program product of claim 9, wherein in the counting step (d), the first threshold size is 6 pixels and the second threshold size is 2 pixels.

12. The computer program product of claim 9, wherein step (e) comprises:

(f) defining a plurality of groups of binary images from the plurality of binary images, wherein the plurality of binarization thresholds form an increasing or decreasing sequence, and wherein each group includes a plurality of binary images corresponding to consecutive binarization thresholds in the sequence;

(g) for each group, computing a first average value which is an average of the first numbers and a second average value which is an average of the second numbers;

(h) based on the first average values and the second average values of the plurality of groups, selecting one of the groups as an optimum group; and

(i) within the optimum group, based on the first number and the second number of each binary image, selecting one of the binary images as the optimum binary image.

13. The computer program product of claim 12, wherein the selecting step (h) comprises:

(h1) if the second average values of the plurality of groups have only one local minimum, selecting the group corresponding to that second average value as the optimum group; and

(h2) if the second average values of the plurality of groups have two or more local minimums,

(h3) selecting a first group and a second group corresponding respectively to a first and a second one of the local minimums;

(h4) if the second average value of the first group is significantly greater than the second average value of the second group, or if the second average value of the first group less than the second average value of the second group and the first average value of the first group is significantly less than the first average value of the second group, selecting the second group as the optimum group; and

(h5) otherwise selecting the first group as the optimum group.

14. The computer program product of claim 13, wherein the selecting step (h4) comprises:

if the second average value of the first group is greater than 1.25 times the second average value of the second group, or if the second average value of the first group less than the second average value of the second group and the first average value of the first group is less than two thirds of the first average value of the second group, selecting the second group as the optimum group.

15. The computer program product of claim 12, wherein the selecting step (i) comprises:

calculating a difference value of the first number and the second number for each binary image; and

selecting one of the binary images that has a largest difference value as the optimum binary image.

16. The computer program product of claim 12, wherein in the binarizing step (a), the document image is binarized 16 times using 16 binarization thresholds, and wherein the defining step (f) defines 7 groups of binary images.