Method and apparatus for detecting an orientation of characters in a document image

• US 7,580,571 B2
• Filed: 07/19/2005
• Issued: 08/25/2009
• Est. Priority Date: 07/20/2004
• Status: Expired due to Fees

• Current Assignee: Ricoh Company Limited
• Original Assignee: Ricoh Company Limited
• Inventors: Ohguro, Yoshihisa
• Primary Examiner(s): Ahmed; Samir A.
• Assistant Examiner(s): Koziol; Stephen R
• Application Number:
  US11/183,852
• Publication Number:
  US 20060018544A1
• Time in Patent Office: 1,498 Days
• Field of Search: 382/181, 382/182, 382/185, 382/276, 704/9
• US Class Current: 382/181
• CPC Class Codes: G06V 30/10 Character recognition
• CPC Class Codes: G06V 30/1463 Orientation detection or correction, e.g. rotation of multiples of 90 degrees

Abstract

An apparatus and method for extracting a predetermined number of consecutive characters from a document image; converting the consecutive characters into a set of symbols using layout information obtained from the consecutive characters; and detecting an orientation of the consecutive characters using occurrence probabilities of the set of symbols.

Claims

1. An image processing method, comprising:

using a processor to perform the steps of;

extracting a predetermined number of consecutive characters from a document image;

converting the consecutive characters into a set of symbols using layout information obtained from the consecutive characters; and

detecting an orientation of the consecutive characters based on occurrence probabilities of the set of symbols obtained using a plurality of reference images, each reference image corresponding to one of the predetermined orientations,

wherein the plurality of reference images includes four types of reference images, each type corresponding to one of four orientations, the four types of reference images being prepared for at least one of a vertical character line and a horizontal character line.

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

determining an orientation of the document image based on the detected orientation of the consecutive characters.

3. The method of claim 1, wherein the converting step comprises the steps of:

forming one or more circumscribed rectangles for each of the consecutive characters; and

extracting one or more parameters from each of the circumscribed rectangles, the parameters being used as the layout information.

4. The method of claim 3, wherein the parameters are selected from a group consisting of:

a height of the circumscribed rectangle starting from a baseline of the circumscribed rectangles;

a height of the circumscribed rectangle;

a width of the circumscribed rectangle;

a distance of the circumscribed rectangle with respect to one of the circumscribed rectangles arranged adjacent to the circumscribed rectangle; and

a number of black pixels contained in the circumscribed rectangle.

5. The method of claim 3, wherein the converting step further comprises the steps of:

forming a character line based on a plurality of circumscribed rectangles of the one or more circumscribed rectangles, the plurality of circumscribed rectangles being located closely to each other; and

normalizing each of the parameters with respect to the character line using a line parameter obtained from the character line, the normalized parameters being used as the layout information.

6. An image processing method, comprising:

using a processor to perform the steps of;

extracting a predetermined number of consecutive characters from a document image;

converting the consecutive characters into a set of symbols using layout information obtained from the consecutive characters; and

detecting an orientation of the consecutive characters based on occurrence probabilities of the set of symbols obtained using a plurality of reference images, each reference image corresponding to one of the predetermined orientations,

wherein the converting step comprises the steps of;

quantizing the layout information into a set of integers; and

assigning each of the integers with the corresponding symbol to generate the set of symbols.

7. The method of claim 6, wherein the quantizing step uses a vector quantization method.

8. An image processing method comprising:

using a processor to perform the steps of;

extracting a predetermined number of consecutive characters from a document image;

converting the consecutive characters into a set of symbols using layout information obtained from the consecutive characters; and

detecting an orientation of the consecutive characters based on occurrence probabilities of the set of symbols obtained using a plurality of reference images, each reference image corresponding to one of the predetermined orientations,

wherein the set of symbols are expressed as 8-bit data.

9. An image processing method comprising:

using a processor to perform the steps of;

extracting a predetermined number of consecutive characters from a document image;

converting the consecutive characters into a set of symbols using layout information obtained from the consecutive characters; and

detecting an orientation of the consecutive characters based on occurrence probabilities of the set of symbols obtained using a plurality of reference images, each reference image corresponding to one of the predetermined orientations,

wherein the detecting step comprises the steps of;

obtaining a plurality of n-gram models each trained on one of the plurality of reference images;

calculating the occurrence probabilities of the set of symbols for each of the predetermined orientations using the n-gram models; and

selecting one of the predetermined orientations corresponding to the occurrence probability having a largest value.

10. The method of claim 9, wherein the selected orientation is used as the detected orientation of the consecutive characters.

11. The method of claim 9, wherein the plurality of reference images are each generated based on a specific language.

12. The method of claim 9, wherein the detecting step further comprises the step of:

detecting a language type of the character line using the set of symbols,

wherein the plurality of reference images correspond to the detected language type.

13. The method of claim 9, wherein the detecting step further comprises the step of:

determining whether the largest value of the occurrence probability is larger by a predetermined amount than a second largest value of the occurrence probability to generate a determination result,

wherein the selected orientation is used as the detected orientation of the consecutive characters based on the determination result.

14. An image processing method, comprising:

using a processor to perform the steps of;

forming one or more character lines based on a plurality of consecutive characters extracted from a document image, each of the character lines having a plurality of consecutive characters;

converting the plurality of consecutive characters in the character lines into a set of symbols using layout information obtained from the plurality of consecutive characters in the character line;

detecting an orientation of one or more of the character lines based on occurrence probabilities of the set of symbols obtained using a plurality of reference images, each reference image corresponding to one of the predetermined orientations; and

determining an orientation of the document image based on the detected orientations of the character lines, wherein the determining step comprises the steps of;

calculating, for each of the detected orientations, a number of character lines corresponding to the detected orientations; and

selecting one of the detected orientations having a largest number of character lines.

15. The method of claim 14, wherein the selected detected orientation is used as the determined orientation of the document image.

16. The method of claim 14, wherein the determining step further comprises the step of:

determining whether the largest number of the selected orientation is larger by a predetermined value than a second largest number of the detected orientation to generate a determination result,

wherein the selected orientation is used as the determined orientation of the document image based on the determination result.

17. An image processing method, comprising:

using a processor to perform the steps of;

forming one or more character lines based on a plurality of consecutive characters extracted from a document image, each of the character lines having a plurality of consecutive characters;

converting the plurality of consecutive characters in the character lines into a set of symbols using layout information obtained from the plurality of consecutive characters in the character line;

detecting an orientation of one or more of the character lines based on occurrence probabilities of the set of symbols obtained using a plurality of reference images, each reference image corresponding to one of the predetermined orientations; and

determining an orientation of the document image based on the detected orientations of the character lines, wherein the determining step comprises the steps of;

obtaining, for each of the one or more character lines, the number of symbols contained in the character line; and

selecting one of the detected orientations based on the obtained number of symbols.

18. An image processing apparatus, comprising:

means for extracting a predetermined number of consecutive characters from a document image;

means for converting the predetermined number of consecutive characters into a set of symbols using layout information obtained from the predetermined number of consecutive characters; and

means for detecting an orientation of the predetermined number of consecutive characters based on occurrence probabilities of the set of symbols obtained using a plurality of reference images, each reference image corresponding to one of the predetermined orientations,

wherein the plurality of reference images includes four types of reference images, each type corresponding to one of four orientations, the four types of reference images being prepared for at least one of a vertical character line and a horizontal character line.

19. The image processing apparatus of claim 18, further comprising:

means for determining an orientation of the document image based on the detected orientation of the predetermined number of consecutive characters.

20. The image processing apparatus of claim 19, further comprising:

means for inputting the document image.

21. The image processing apparatus of claim 20, further comprising:

means for correcting the orientation of the document image based on the determined orientation of the document image.

22. The image processing apparatus of claim 21, further comprising:

means for outputting the document image having the orientation being corrected.

23. An image processing apparatus comprising:

means for extracting a predetermined number of consecutive characters from a document image;

means for converting the consecutive characters into a set of symbols using layout information obtained from the consecutive characters;

means for detecting an orientation of the consecutive characters based on occurrence probabilities of the set of symbols obtained using a plurality of reference images, each reference image corresponding to one of predetermined orientations; and

means for storing a plurality of n-gram models each trained on one of a plurality of reference document images,

wherein at least one of the plurality of n-gram models is used by the detecting means to detect the orientation of the consecutive characters.

24. An image processing system, comprising:

a processor; and

a storage device configured to store a plurality of instructions which, when activated by the processor, cause the processor to perform an image processing operation comprising;

extracting a predetermined number of consecutive characters from a document image;

converting the predetermined number of consecutive characters into a set of symbols using layout information obtained from the predetermined number of consecutive characters; and

detecting an orientation of the predetermined number of consecutive characters based on occurrence probabilities of the set of symbols obtained using a plurality of reference images, each reference image corresponding to one of the predetermined orientations,

wherein the plurality of reference images includes four types of reference images, each type corresponding to one of four orientations, the four types of reference images being prepared for at least one of a vertical character line and a horizontal character line.

25. The image processing system of claim 24, wherein the image processing operation further comprises:

determining an orientation of the document image based on the detected orientation of the predetermined number of consecutive characters.

26. A computer readable medium storing computer instructions for performing an image processing operation comprising:

extracting a predetermined number of consecutive characters from a document image;

converting the predetermined number of consecutive characters into a set of symbols using layout information obtained from the predetermined number of consecutive characters; and

detecting an orientation of the predetermined number of consecutive characters based on occurrence probabilities of the set of symbols obtained using a plurality of reference images, each reference image corresponding to one of the predetermined orientations,

wherein the plurality of reference images includes four types of reference images, each type corresponding to one of four orientations, the four types of reference images being prepared for at least one of a vertical character line and a horizontal character line.

27. The medium of claim 26, wherein the image processing operation further comprises:

determining an orientation of the document image based on the detected orientation of the predetermined number of consecutive characters.

28. An image processing method comprising:

using a processor to perform the steps of;

extracting a predetermined number of consecutive characters from a document image;

converting the consecutive characters into a set of symbols using layout information obtained from the consecutive characters; and

detecting an orientation of the consecutive characters based on occurrence probabilities of the set of symbols obtained using a plurality of reference images, each reference image corresponding to one of the predetermined orientations,

wherein the predetermined number of consecutive characters is three for trigrams.

29. An image processing apparatus comprising:

means for extracting a predetermined number of consecutive characters from a document image;

means for converting the consecutive characters into a set of symbols using layout information obtained from the consecutive characters; and

means for detecting an orientation of the consecutive characters based on occurrence probabilities of the set of symbols obtained using a plurality of reference images, each reference image corresponding to one of predetermined orientations,

wherein the predetermined number of consecutive characters is three for trigrams.

30. An image processing apparatus comprising:

means for extracting a predetermined number of consecutive characters from a document image;

means for converting the consecutive characters into a set of symbols using layout information obtained from the consecutive characters; and

means for detecting an orientation of the consecutive characters based on occurrence probabilities of the set of symbols obtained using a plurality of reference images, each reference image corresponding to one of predetermined orientations,

wherein the plurality of reference images includes four types of reference images, each type corresponding to one of four orientations, the four types of reference images being prepared for at least one of a vertical character line and a horizontal character line.

31. An image processing system comprising:

a processor; and

a storage device configured to store a plurality of instructions, which when activated by the processor, cause the processor to perform an image processing operation comprising;

extracting a predetermined number of consecutive characters from a document image;

converting the predetermined number of consecutive characters into a set of symbols using layout information obtained from the predetermined number of consecutive characters; and

detecting an orientation of the predetermined number of consecutive characters based on occurrence probabilities of the set of symbols obtained using a plurality of reference images, each reference image corresponding to one of predetermined orientations, wherein the predetermined number of consecutive characters is three for trigrams.

32. An image processing system comprising:

a processor; and

a storage device configured to store a plurality of instructions, which when activiated by the processor, cause the processor to perform an image processing operation comprising;

extracting a predetermined number of consecutive characters from a document image;

converting the predetermined number of consecutive characters into a set of symbols using layout information obtained from the predetermined number of consecutive characters; and

detecting an orientation of the predetermined number of consecutive characters based on occurrence probabilities of the set of symbols obtained using a plurality of reference images, each reference image corresponding to one of predetermined orientations, wherein the plurality of reference images includes four types of reference images, each type corresponding to one of four orientations, the four types of reference images being prepared for at least one of a vertical character line and a horizontal character line.

33. A computer readable medium storing computer instructions for performing an image processing operation comprising:

extracting a predetermined number of consecutive characters from a document image;

converting the consecutive characters into a set of symbols using layout information obtained from the consecutive characters; and

detecting an orientation of the consecutive characters based on occurrence probabilities of the set of symbols obtained using a plurality of reference images, each reference image corresponding to one of predetermined orientations,

wherein the predetermined number of consecutive characters is three for trigrams.

34. A computer readable medium storing computer instructions for performing an image processing operation comprising:

extracting a predetermined number of consecutive characters from a document image;

converting the consecutive characters into a set of symbols using layout information obtained from the consecutive characters; and

detecting an orientation of the consecutive characters based on occurrence probabilities of the set of symbols obtained using a plurality of reference images, each reference image corresponding to one of predetermined orientations,

wherein the plurality of reference images includes four types of reference images, each type corresponding to one of four orientations, the four types of reference images being prepared for at least one of a vertical character line and a horizontal character line.