Page analysis system

• US 6,512,848 B2
• Filed: 11/18/1996
• Issued: 01/28/2003
• Est. Priority Date: 11/18/1996
• Status: Expired due to Fees

• Current Assignee: Canon Kabushiki Kaisha (Canon Inc.)
• Original Assignee: Canon Kabushiki Kaisha (Canon Inc.)
• Inventors: Niki, Toru, Wang, Shin-Ywan
• Primary Examiner(s): AHMED, SAMIR ANWAR
• Application Number:
  US08/751,677
• Publication Number:
  US 20010012400A1
• Time in Patent Office: 2,262 Days
• Field of Search: 382/173, 382/174, 382/175, 382/176, 382/177, 382/180, 382/190, 382/179, 382/202, 382/204, 382/228, 382/232, 382/245, 382/289, 382/290, 382/291, 382/317, 358/261.1, 358/462, 358/464, 358/453
• US Class Current: 382/176
• CPC Class Codes: G06V 30/413 Classification of content, e.g. text, photographs or tables

Abstract

A method for increasing the accuracy of image data classification in a page analysis system for analyzing image data of a document page. The method includes inputting image data of a document page as pixel data, analyzing the pixel data in order to locate all connected pixels, rectangularizing connected pixel data into blocks, analyzing each of the blocks of pixel data in order to determine the type of image data contained in the block, outputting an attribute corresponding to the type of image data determined in the analyzing step, and performing optical character recognition to attempt to recognize a character of the block of image data in the case that the analyzing step cannot determine the type of image data contained in the block.

Claims

1. In a page analysis system for analyzing image data of a document page, a method comprising the steps of:

inputting image data of a document page as pixel data;

a first analyzing step for analyzing the pixel data in order to locate connected pixels;

rectangularizing the located connected pixels into blocks;

a second analyzing step for analyzing a block of pixel data in order to determine a type of image data contained in the block;

outputting an attribute corresponding to the type of image data within the block determined in the second analyzing step in a case that the second analyzing step determines that the type of image data in the block is not unknown; and

performing optical character recognition so as to recognize image data in the block in a case that the second analyzing step determines that the type of image data contained in the block is unknown,

wherein, if the image data in the block is recognized, the type of image data is determined to be unknown.

2. A method according to claim 1, wherein, in the second analyzing step, the pixel data is analyzed for text data or non-text data and wherein, in the outputting step, in the case that the block of image data is determined to be text data, a text data attribute is output, or, if the data is determined to be non-text data, a non-text data attribute is output.

22. A method according to claim 1, wherein, if the image data in the block is not recognized, the type of image data contained in the block is determined to be picture data.

3. In a page analysis system for analyzing image data of a document page, a method comprising the steps of:

dividing the image data into blocks using a block selection technique, each block including one or more connected components and having an associated type classification;

comparing a preset threshold size range to each connected component of a block classified as text;

performing optical character recognition on connected components in the text block when it is determined that most connected components in the block have a size that is outside the preset threshold size; and

reclassifying the block as other than text when it is determined as a result of performing the optical character recognition that most connected components in the block are unrecognizable.

4. A method according to claim 3, wherein in reclassifying the block as other than text, the block is reclassified as picture data when it is determined as a result of performing the optical character recognition that most connected components in the block are unrecognizable.

5. A method according to claim 3, wherein the optical character recognition is performed in a case that most connected components in the block have a size that is greater than the preset size threshold.

6. A method according to claim 3, wherein the optical character recognition is performed in a case that most connected components in the block have a size that is less than the preset size threshold.

7. A method according to claim 3, wherein the size threshold is based on an average size of connected components in the image data of the document page.

8. Computer-executable process steps stored in a computer-readable medium, the process steps for use in a page analysis system for analyzing image data of a document page, the process steps comprising:

an inputting step to input image data of a document page as pixel data;

a first analyzing step to analyze the pixel data in order to locate connected pixels;

a rectangularizing step to rectangularize the located connected pixels into blocks;

a second analyzing step to analyze a block of pixel data in order to determine a type of image data contained in the block;

an outputting step to output an attribute corresponding to the type of image data within the block determined in the second analyzing step in a case that the type of image data within the block is not determined to be unknown in the second analyzing step; and

a performing step to perform optical character recognition so as to recognize image data of the block in a case that the type of image data contained in the block is determined to be unknown in the second analyzing step,

wherein, if the image data in the block is recognized, the type of image data is determined to be unknown.

9. Computer-executable process steps according to claim 8, wherein, in the second analyzing step, the pixel data is analyzed for text data or non-text data and wherein, in the outputting step, in the case that the block of image data is determined to be text data, a text data attribute is output, or, if the data is determined to be non-text data, a non-text data attribute is output.

23. Computer-executable process steps according to claim 8, further comprising an outputting step to output a picture attribute in a case that the image data of the block is not recognized in said performing step.

10. Computer-executable process steps for analyzing image data of a document page, the steps comprising:

a dividing step to divide the image data into blocks using a block selection technique, each block including one or more connected components and having an associated type classification;

a comparing step to compare a preset threshold size range to each connected component of a block classified as text;

a performing step to perform optical character recognition on the connected components in the text block when it is determined that most connected components in the block have a size that is outside the preset threshold size; and

a reclassifying step to reclassify the block as other than text when it is determined as a result of performing the optical character recognition that most connected components in the block are unrecognizable.

11. Computer-executable process steps according to claim 10, wherein the reclassifying step reclassifies the block as picture data when it is determined as a result of performing the optical character recognition that most connected components in the block are unrecognizable.

12. Computer-executable process steps according to claim 10, wherein the optical character recognition is performed in a case that most connected components in the block have a size that is greater than the preset size threshold.

13. Computer-executable process steps according to claim 10, wherein the optical character recognition is performed in a case that most connected components in the block have a size that is less than the preset size threshold.

14. Computer-executable process steps according to claim 10, wherein the size threshold is based on an average size of connected components in the image data of the document page.

15. An apparatus for performing page analysis of a document page, the apparatus comprising:

a memory which stores page analysis process steps executable by a processor and an image of a document page; and

a processor which executes the page analysis process steps stored in the memory (1) to input image data of a document page as pixel data, (2) to analyze the pixel data in order to locate connected pixels, (3) to rectangularize the located connected pixels into blocks, (4) to analyze a block of pixel data in order to determine a type of image data contained in the block, (5) to output an attribute corresponding to the type of image data within the block in a case that the type of image data within the block is not determined to be unknown, and (6) to perform optical character recognition to attempt to recognize a character of the block of image data in a case that the type of image data contained in the block is determined to be unknown,

wherein, if the image data in the block is recognized, the type of the image data is determined to be unknown.

16. An apparatus according to claim 15, wherein the processor analyzes each block of pixel data in order to determine a type of image data contained in each block by analyzing the pixel data for text data or non-text data;

and

wherein the processor outputs (1) a text data attribute in a case that the block of image data is determined to be text data, or (2) a non-text data attribute in a case that the data is determined to be non-text data.

24. An apparatus according to claim 15, wherein the processor executes the page analysis process steps stored in the memory to output a picture attribute in a case that a character of the block of image data is not recognized.

17. An apparatus for analyzing image data of a document page, the apparatus comprising:

a memory which stores page analysis process steps executable by a processor and an image of a document page; and

a processor which executes the page analysis process steps stored in the memory (1) to divide the image data into blocks using a block selection technique, each block including one or more connected components and having an associated type classification, (2) to compare a preset threshold size range to each connected component of a block classified as text, (3) to perform optical character recognition on connected components in the text block when it is determined that most connected components in the block have a size that is outside the preset threshold size, and (4) to reclassify the block as other than text when it is determined as a result of performing the optical character recognition that most connected components in the block are unrecognizable.

18. An apparatus according to claim 17, wherein the processor executes process steps stored in the memory to reclassify the block as picture data when it is determined as a result of performing the optical character recognition that most connected components in the block are unrecognizable.

19. An apparatus according to claim 17, wherein the processor performs the optical character recognition in a case that most connected components in the block have a size that is greater than the preset size threshold.

20. An apparatus according to claim 17, wherein the processor performs the optical character recognition in a case that most connected components in the block have a size that is less than the preset size threshold.

21. An apparatus according to claim 17, wherein the size threshold is based on an average size of connected components in the image data of the document page.