Image processing apparatus and pattern extraction apparatus

• US 6,771,813 B1
• Filed: 07/21/1999
• Issued: 08/03/2004
• Est. Priority Date: 12/09/1998
• Status: Expired due to Fees

• Current Assignee: Fujitsu Limited
• Original Assignee: Fujitsu Limited
• Inventors: Katsuyama, Yutaka
• Primary Examiner(s): Au, Amelia M.
• Assistant Examiner(s): BHATNAGAR, ANAND P
• Application Number:
  US09/358,380
• Time in Patent Office: 1,840 Days
• Field of Search: 382/162, 382/165, 382/193, 382/194, 382/270-273, 345/589, 345/593, 345/597, 345/601, 345/602, 348/453
• US Class Current: 382/165
• CPC Class Codes: G06T 2207/10008 from scanner, fax or copier
• CPC Class Codes: G06T 2207/10024 Color image
• CPC Class Codes: G06T 2207/30176 Document
• CPC Class Codes: G06T 7/11 Region-based segmentation
• CPC Class Codes: G06T 7/136 involving thresholding
• CPC Class Codes: G06V 10/143 Sensing or illuminating at different wavelengths
• CPC Class Codes: G06V 30/414 Extracting the geometrical structure, e.g. layout tree; Block segmentation, e.g. bounding boxes for graphics or text

Abstract

A control unit selects a first labeling for a color image in a small number of colors so that the labeling is performed by clustering a color palette, and selects a second labeling for a color image in a large number of colors so that the labeling is performed by an adjacency expanding method.

Claims

1. An image processing apparatus, comprising:

a read information obtaining unit to obtain read information about an input image, including

a local area extraction unit to extract a local area from the input image; and

a color difference information extraction unit to extract color difference information about the input image from the local area;

a unicolor range setting unit to set a unicolor range of the input image according to the read information about the input image, including a threshold setting unit to set a labeling threshold for the input image according to the color difference information; and

a labeling unit to label the input image by assigning a same label to a connection picture element in the unicolor range set by said unicolor range setting unit, including

an average color computation unit to compute an average color of connection areas assigned the same label; and

a color difference computation unit to compute a color difference between the average color of the connection area and a color of a picture element adiacent to the connection area, said labeling unit assigning the same label of the connection area to the picture element when the color difference is not greater than the labeling threshold.

2. The apparatus according to claim 1, wherein said read information about the input image comprises at least one of a number of colors of the input image, read resolution, luminance value, color saturation, color difference between adjacent picture elements, and color variance.

3. The apparatus according to claim 1, wherein said labeling unit changes a labeling method according to the read information about the input image.

4. The apparatus according to claim 3, wherein said labeling unit performs a labeling process by at least one of an area expanding method and clustering color distribution of an image.

5. The apparatus according to claim 1, wherein said labeling unit further comprises a clustering unit to classify similar colors of a color image into a same cluster, and assigns the same label to picture elements connected by a color belonging to the same cluster.

6. The apparatus according to claim 1,

wherein said local area extraction unit comprises a mesh area extraction unit to extract a mesh area whose color variance is within a predetermined range among mesh areas obtained by dividing the input image;

wherein said color difference information extraction unit comprises a standard deviation computation unit to compute a standard deviation of a color in the extracted mesh area; and

wherein said threshold setting unit sets the labeling threshold based on a mean value of the standard deviations obtained for each mesh area.

7. The apparatus according to claim 1, wherein said unicolor range setting unit comprises a color conversion unit to convert a color of a color signal such that a color difference resolution by naked eyes of a first color with low color saturation matches the color difference resolution by naked eyes of a second color with high color saturation by reducing the color difference between the first color and a third color around the first color and enlarging the color difference between the second color and a fourth color around the second color.

8. The apparatus according to claim 1,

wherein said read information obtaining unit comprises a resolution computation unit to compute read resolution of a color image for each color element;

wherein said unicolor range setting unit sets the labeling threshold based on the read resolution of each color element.

9. The apparatus according to claim 1, further comprising a grouping unit to group label patterns generated by said labeling unit.

10. The apparatus according to claim 9, wherein said grouping unit comprises a pattern setting unit to set a pattern of a specific shape, and removes the pattern of the specific shape from patterns to be grouped.

11. The apparatus according to claim 10, wherein said specific shape is an L shape or a shape.

12. The apparatus according to claim 9, wherein said grouping unit comprises a search range setting unit to set a search range of patterns to be grouped.

13. The apparatus according to claim 9, wherein said grouping unit groups patterns in a unicolor range set by said unicolor range setting unit into a same group.

14. The apparatus according to claim 13, wherein a color of the patterns has an average value of colors of picture elements assigned the same label.

15. The apparatus according to claim 9, wherein said grouping unit comprises an enclosing rectangle generation unit to generate an enclosing rectangle of patterns assigned the same label, and groups the patterns based on a size, shape, or position of the enclosing rectangle.

16. The apparatus according to claim 9, wherein said grouping unit comprises a thickness computation unit to compute a thickness of the patterns assigned the same label, and groups patterns based on the thickness of the patterns.

17. The apparatus according to claim 16, wherein said thickness computation unit comprises:

a scanning unit to scan an image in a predetermined direction; and

an outline length computation unit to compute a length of an outline of a pattern in the image.

18. The apparatus according to claim 9,

further comprising a group information computation unit to compute image information about a group according to image information about patterns classified into a same group, and

wherein said grouping unit groups patterns according to the image information about the group.

19. The apparatus according to claim 18, wherein said group information computation unit comprises a color information computation unit to compute color information about the group according to the color information about patterns classified into the same group.

20. The apparatus according to claim 19, wherein said color information about the group is an average value of colors of the patterns belonging to the same group.

21. The apparatus according to claim 19, wherein said group information computation unit computes a grouping threshold according to the color information about the pattern classified into the same group.

22. The apparatus according to claim 19, wherein said grouping unit groups patterns based on a result of comparing the color information about the group with color information about patterns to be grouped.

23. The apparatus according to claim 19, wherein said grouping unit integrates groups based on a result of comparing color information about the group.

24. The apparatus according to claim 19, wherein said unicolor range setting unit computes the labeling threshold according to color information of patterns classified into the same group.

25. The apparatus according to claim 19, wherein said labeling unit re-extracts patterns belonging to the group based on a result of comparing color information of the group with color information about picture elements in an area of enclosing rectangles of the group.

26. The apparatus according to claim 18,

wherein said group information computation unit comprises:

a group rectangle generation unit to generate an enclosing rectangle of an entire pattern belonging to the same group; and

a geometric information computation unit to compute a shape, size, or position of the enclosing rectangle of the group, and

wherein said grouping unit integrates the groups based on the shape, size, or position of the enclosing rectangle of the group.

27. The apparatus according to claim 18,

wherein said group information computation unit comprises:

a frequency distribution generation unit to generate a frequency distribution of thicknesses of patterns in the same group; and

a same thickness setting unit to set a same pattern thickness range based on the frequency distribution of the thicknesses, and

wherein said grouping unit re-classifies patterns belonging to the same group based on the thickness of a pattern set by said same thickness setting unit.

28. The apparatus according to claim 18,

wherein said group information computation unit comprises a character recognition unit to recognize a character of a pattern belonging to the same group;

and

wherein said grouping unit re-classifies patterns belonging to the same group based on a result of character recognition.

29. An image processing apparatus, comprising:

a color difference table for storing a maximum value of a color difference between adjacent picture elements based on a color luminance value and read resolution as variables;

a resolution obtaining unit to obtain a read resolution at which a color difference between adjacent picture elements obtained from an input image matches a color difference stored in said color difference table individually for each of three primary colors;

a resolution computation unit to compute the read resolution of the input image based on the read resolution of the three primary colors;

a color difference obtaining unit to obtain from said color difference table a color difference corresponding to the read resolution of the input image;

a threshold setting unit to set a labeling threshold based on the color difference obtained from said color difference table; and

a labeling unit to label the input image based on the threshold.

30. The apparatus according to claim 29, wherein said color difference table contains a maximum value of a color difference between adjacent picture elements for all luminance values of colors of an image.