Method And Apparatus For Detecting Objects In An Image

• US 20080170784A1
• Filed: 11/08/2007
• Published: 07/17/2008
• Est. Priority Date: 01/12/2007
• Status: Active Grant

• Current Assignee: Seiko Epson Corporation (Seiko Group)
• Original Assignee: Seiko Epson Corporation (Seiko Group)
• Inventors: Zhou, Hui, Guerzhoy, Michael
• Granted Patent: US 8,098,936 B2
• Time in Patent Office: Days
• US Class Current: 382/173
• CPC Class Codes: G06T 2207/10008 from scanner, fax or copier
• CPC Class Codes: G06T 2207/30176 Document
• CPC Class Codes: G06T 7/12 Edge-based segmentation
• CPC Class Codes: G06T 7/143 involving probabilistic approaches, e.g. Markov random field [MRF] modelling
• CPC Class Codes: G06T 7/181 involving edge growing; involving edge linking

Abstract

A method of detecting generally rectangular objects in an image comprises determining candidate rectangles in the image based on detected corners in the image, ranking the candidate rectangles on the basis of a set of differentiating factors and detecting objects in the images based on the ranked candidate rectangles.

Claims

1. A method of detecting generally rectangular objects in an image comprising:

determining candidate rectangles in the image based on detected corners in the image;

ranking the candidate rectangles on the basis of a set of differentiating factors; and

detecting objects in said images based on the ranked candidate rectangles.

2. The method of claim 1 further comprising, prior to said ranking, discarding selected candidate rectangles.

3. The method of claim 2 wherein selected candidate rectangles are discarded when they overlap with higher ranked candidate rectangles by more than a threshold amount.

4. The method of claim 3 further comprising, prior to said ranking, merging proximate candidate rectangles.

5. The method of claim 4 wherein said set of differentiating factors is based at least on edge and color information associated with pixels proximate the candidate rectangles.

6. The method of claim 4 further comprising, prior to said ranking, adjusting boundaries of the candidate rectangles.

7. The method of claim 6 wherein during said adjusting, the positions of the corners defining the candidate rectangles are adjusted within neighborhoods surrounding the candidate rectangles.

8. The method of claim 1 wherein said determining comprises:

estimating a background color of said image;

detecting edges delimiting the objects on the basis of the estimated background color;

examining the edges to detect intersecting edges representing corners; and

grouping corners that define vertices of a rectangle thereby to determine the candidate rectangles.

9. The method of claim 8 further comprising, prior to said ranking, adjusting boundaries of the candidate rectangles.

10. The method of claim 9 further comprising, prior to said ranking, discarding selected candidate rectangles.

11. The method of claim 10 wherein selected candidate rectangles are discarded when they overlap with higher ranked candidate rectangles by more than a threshold amount.

12. The method of claim 11 further comprising, prior to said ranking merging proximate candidate rectangles.

17. A computer readable medium embodying a computer program comprising program code for carrying out the method of claim 1.

13. A method of estimating background color in an image comprising:

segmenting each scan line of the image into like-color line segments;

grouping the like-color line segments by color; and

estimating the background color on the basis of edge pixels along the line segments of said groups.

14. The method of claim 13 wherein during said grouping, only like-color line segments exceeding a threshold length are grouped.

15. The method of claim 14 wherein said estimating is made on the basis of the number and strength of edge pixels.

16. The method of claim 15 wherein said estimating comprises:

determining the groups of like-color line segments having a threshold number of line segments;

designating the average color of each group of like-color line segments as a candidate background color;

for each candidate background color, detecting edge pixels along the line segments of the group representing that candidate background color and calculating the strength of edges defined by the edge pixels; and

selecting one of the candidate background colors as the background color estimate based on edge pixel count and edge strength.

18. A computer readable medium embodying a computer program comprising program code for carrying out the method of claim 13.

19. An apparatus for detecting generally rectangular objects in an image comprising:

memory for storing the image; and

processing structure communicating with said memory, said processing structure processing the image to determine candidate rectangles based on detected corners in the image, rank the candidate rectangles on the basis of a set of differentiating factors, and determine objects in the images from the ranked candidate rectangles.

20. An apparatus according to claim 19 wherein said processing structure, prior to said ranking, further processes the image to at least one of discard selected candidate rectangles, merge proximate candidate rectangles and adjust candidate rectangles.

21. An apparatus according to claim 20 wherein said processing structure processes the image to discard selected candidate rectangles, merge proximate candidate rectangles and adjust candidate rectangles.

22. An apparatus for estimating background color in an image comprising:

memory for storing said image; and

processing structure communicating with the memory, said processing structure processing the image to segment each scan line of the image into like-color line segments, group the like-color line segments by color and estimate the background color on the basis of edge pixels along the line segments of the groups.