Text character identification system and method thereof

• US 8,306,325 B2
• Filed: 06/01/2005
• Issued: 11/06/2012
• Est. Priority Date: 06/04/2004
• Status: Active Grant

• Current Assignee: Yoshinaga Technologies LLC (Intellectual Ventures LLC)
• Original Assignee: Yoshinaga Technologies LLC (Intellectual Ventures LLC)
• Inventors: Chang, Hao-Wei
• Primary Examiner(s): Bella, Matthew
• Assistant Examiner(s): Bekele, Mekonen
• Application Number:
  US11/142,846
• Publication Number:
  US 20050271275A1
• Time in Patent Office: 2,715 Days
• Field of Search: 382176-177, 382/164, 382/167, 382/173, 382/274, 382288-291, 358/62, 358/464, 358/462, 358/446, 358/1.9, 358/466, 358/465, 358/456, 358/458, 705/500, 705/530, 705/517, 705/518, 705/519, 705/520, 705/521
• US Class Current: 382/176
• CPC Class Codes: G06V 30/10 Character recognition
• CPC Class Codes: G06V 30/186 by deriving mathematical or geometrical properties from the whole image
• CPC Class Codes: G06V 30/413 Classification of content, e.g. text, photographs or tables

Abstract

A method for text character identification. The method acquires multiple connected components (CCs) in a binary image, and each CC has a pattern property value. The method determines at least one property limit based on the pattern property values, generates a filtering rule according to the property limit, and determines whether each of the CCs is a text character according to the filtering rule.

Claims

1. A non-transitory computer-readable storage medium having stored thereon computer-executable instructions that, if executed by a computing device, cause the computing device to perform operations comprising:

acquiring a plurality of connected components (CCs) in a binary image;

determining a property limit by acquiring at least one of a most frequently occurring CC base height and a most frequently occurring CC base width in the binary image;

generating a filtering rule according to the property limit;

determining that a first plurality of CCs are text characters according to the filtering rule, and adjacency to a CC that has previously been determined to be text characters;

assigning a first text label to the first plurality of CCs;

determining that a second plurality of CCs are not text characters according to the filtering rule, or because individual CCs are not adjacent to an individual CC that has previously been determined to be a text character;

assigning a second text label to the second plurality of CCs;

merging neighboring CCs into a plurality of text regions; and

determining that the CCs in any text region are not text characters if there are fewer than a predetermined number of CCs within the text region.

2. The non-transitory computer-readable storage medium of claim 1, wherein the operations further comprise converting a bitmap image to the binary image using a thresholding method.

3. The non-transitory computer-readable storage medium of claim 2 wherein the thresholding method comprises a minimum error thresholding method or Otsu thresholding method.

4. The non-transitory computer-readable storage medium of claim 1 wherein acquiring a plurality of CCs comprises using a region growing method.

5. The non-transitory computer-readable storage medium of claim 4 wherein the region growing method comprises merging pixels to CCs when those pixels are both adjacent to the CCs and similar in property.

6. The non-transitory computer-readable storage medium of claim 1 wherein determining a property limit further comprises determining a CC density.

7. The non-transitory computer-readable storage medium of claim 5 wherein the property limit comprises an upper height limit, an upper width limit, an upper density limit, a lower height limit, a lower width limit and a lower density limit, and wherein determining the property limit further comprises:

determining the upper and lower height limits either by multiplying a first weighted value by the base height or by adding a first weighted value to the base height;

determining the lower height limit either by dividing the base height by a second height value or by subtracting a second weighted value from the base height;

determining an upper width limit either by multiplying a third weighted value by the base width or by adding the third weighted value to the base width;

determining a lower width limit either by dividing the base width by a fourth weighted value or by subtracting the fourth weighted value from the base width; and

acquiring upper and lower density limits.

8. The non-transitory computer-readable storage medium of claim 7 wherein the filtering rule comprises:

determining that one of the CCs is not a text character if its height is lower than the lower height limit and its width is lower than the lower width limit;

determining that one of the CCs is not a text character if its height is higher than the upper height limit or its width is higher than the upper width limit; and

determining that one of the CCs is not a text character if its density is higher than the upper density limit or its density is lower than the lower density limit.

9. A method in a computing device having a memory and a processor, the method comprising:

acquiring a plurality of connected components (CCs) in a binary image;

determining with the computing device a property limit by acquiring at least one of a most frequently occurring CC base height in the binary image and a most frequently occurring CC base width in the binary image;

generating with the computing device a filtering rule according to the property limit;

determining that a first CC is a text CC if the first CC both satisfies the filtering rule, and is adjacent to a second CC that has previously been determined to be a text CC;

assigning a text label to the first CC to indicate that the first CC is a text CC;

assigning a potential-text label to the first CC if the first CC fails the filtering rule but is adjacent to a text CC;

assigning the potential-text label to the first CC if the first CC satisfies the filtering rule but is not adjacent to a text CC;

merging a plurality of the first CCs into a text region; and

assigning the text label to the first CCs with the potential text label in the text region if the quantity of first CCs with the text label and the quantity of first CCs with the potential text label is above a threshold quantity.

10. A computer system having a processor and memory for text character identification in a binary image, the system comprising:

a connected component acquisition unit configured to acquire a plurality of connected components (CCs) in a binary image;

a limit generation unit configured to determine at least one property limit by acquiring at least one of a most common CC base width in the binary image and a most common CC base height in the binary image;

a filtering rule generation unit configured to generate a filtering rule according to the property limit; and

a text determination unit configured to—

determine that a CC is a text character according to the filtering rule, and because the CC is adjacent to another CC that has previously been determined to be a text character;

assign a first text label to the CC to indicate that the CC is a text character;

assign a second text label to the CC if the CC is determined not to be a text character according to the filtering rule;

merge neighboring CCs into a plurality of text regions; and

determine that the CCs in any text region are not text characters if there are fewer than a predetermined number of CCs in the text region.

11. The system of claim 10, further comprising a binary image generation unit configured to convert a bitmap image to the binary image using a thresholding method.

12. The system of claim 11 wherein the binary image generation unit is configured to convert the bitmap image using a minimum error thresholding method or an Otsu thresholding method.

13. The system of claim 10 wherein the connected component acquisition unit is configured to acquire CCs using a region growing method.

14. The system of claim 13 wherein the connected component acquisition unit is configured to merge pixels to CCs when those pixels are both adjacent to the CCs and similar in property.

15. The system of claim 10 wherein the limit generation unit is further configured to determine a CC density.

16. The system of claim 14 wherein the property limit comprises an upper height limit, an upper width limit, a lower height limit, and a lower width limit.

17. The system of claim 16 wherein the limit generation unit is further configured to:

determine the upper height limit by multiplying a first weighted value by the base height or adding the first weighted value to the base height;

determine the lower height limit by dividing the base height by a second weighted value or subtracting the second weighted value from the base height;

determine the upper width limit by multiplying a third weighted value by the base width or adding the third weighted value to the base width; and

determine the lower width limit by dividing the base width by a fourth weighted value or subtracting the fourth weighted value from the base width.

18. The system of claim 17 wherein the filtering rule comprises:

one of the CCs is not a text character if its height is lower than the lower height limit and its width is lower than the lower width limit; and

one of the CCs is not a text character if its height is higher than the upper height limit or its width is higher than the upper width limit.

19. The system of claim 10 wherein the text determination unit is further configured to merge neighboring CCs into a plurality of text regions, and determine that the CCs in any text region are not text characters if the number of CCs is lower than a count threshold.

20. A system for text character identification, the system comprising:

means for acquiring a plurality of connected components (CCs) in a binary image;

means for determining a property limit of the CCs by acquiring either a most frequently occurring CC base height or by acquiring a most frequently occurring CC base width;

means for generating a filtering rule according to the property limit;

means for determining that the CCs satisfy the filtering rule;

means for determining that the CCs are adjacent to at least one other CC that has previously been determined to be a text character;

means for assigning a first text label to the CCs if the CCs satisfy the filtering rule and are adjacent to the other CCs that has previously been determined to be a text character;

means for assigning a second text label to each of the CCs that are determined not to satisfy the filtering rule but are adjacent to the other CC that has previously been determined to be a text character;

means for assigning the second text label to each of the CCs that are determined to satisfy the filtering rule but are not adjacent to the other CC that has previously been determined to be a text character;

means for counting the first text labels and the second text labels in a text region; and

means for assigning the first text label to all CCs within the text region if the text region contains more than a threshold number of first text labels and second text labels.

21. A computer system having a processor and memory, the computer system comprising:

a filtering component configured to analyze a plurality of connected components in a binary image to determine whether the connected components satisfy a first rule and a second rule;

a text character labelling component configured to apply an affirmative-text character label to connected components that satisfy both the first and second rules, and to apply a potential-text character label to connected components that satisfy either the first rule or the second rule, but not both the first and second rules;

a text region labelling component configured to apply a text region label to a first and second connected component in the plurality of connected components if a distance between the first connected component and the second connected component is less than a predetermined threshold value; and

a text character counting component configured to count connected components having the text region label and determine that a region of the binary image defined by the connected components having the text region label contains text characters if the connected components number more than a predetermined threshold value, wherein at least one of the filtering component, the text character labelling component, and the text region labelling component is configured as instructions stored in the memory for execution by the processor.

22. The computer system of claim 21 wherein the first rule comprises at least one of a height limit, a width limit, or a density limit, and wherein the second rule comprises at least one of a height limit, a width limit, or a density limit, and further wherein the first rule is different from the second rule.

23. The computer system of claim 21 wherein the text region labeling component further comprises a text region removal unit configured to determine a quantity of connected components in the text region and to remove the text region label from the text region if the quantity is less than a predetermined threshold value.

24. The computer system of claim 21 wherein the first rule comprises a filtering rule based upon at least one of height or width of the connected components, and wherein the second rule comprises a rule based on adjacency of connected components.