Systems and methods for segmenation by removal of monochromatic background with limitied intensity variations

• US 8,411,986 B2
• Filed: 04/13/2010
• Issued: 04/02/2013
• Est. Priority Date: 04/13/2009
• Status: Expired due to Fees

• Current Assignee: Flashfoto, Inc.
• Original Assignee: Flashfoto, Inc.
• Inventors: King, Paul, Ip, Katharine, Piramuthu, Robinson
• Primary Examiner(s): Chen, Wenpeng
• Assistant Examiner(s): Niu, Feng
• Application Number:
  US12/798,917
• Publication Number:
  US 20100316288A1
• Time in Patent Office: 1,085 Days
• Field of Search: 382/199, 382/203, 382/190, 382/195, 382/173, 382/177, 382/181, 382/266, 382/273
• US Class Current: 382/266
• CPC Class Codes: G06V 10/248 by interactive preprocessing or interactive shape modelling, e.g. feature points assigned by a user
• CPC Class Codes: G06V 10/28 Quantising the image, e.g. histogram thresholding for discrimination between background and foreground patterns

Abstract

A method for distinguishing a foreground object from a monochromatic background in a digital image depicting a foreground object in front of said monochromatic background. A outer edge intensity map is created of the foreground object to which a tracing process is conducted. A continuous edge map is created by utilizing the traced edges in the outer edge intensity map. By selecting a path on the continuous edge map based on average edge intensity, a final edge map is created. The pixels comprised within the final edge map are distinguished as the foreground object. Optionally, the final edge map may be refined utilizing certain techniques.

Claims

1. A method of identifying a foreground object of an image from a background object of said image, said method comprising:

generating an outer edge intensity map associated with said foreground object, wherein information associated with said image is stored in a memory component;

generating a plurality of paths on a continuous edge map from said outer edge intensity map, wherein said plurality of paths on said continuous edge map is created by connecting pixels in said outer edge intensity map; and

forming an edge mask by selecting one path on said continuous edge map from said plurality of paths on said continuous edge map, wherein said selecting is based on a highest average edge intensity.

2. The method as described by claim 1, wherein Sobel filters are used to generate said outer edge intensity map.

3. The method as described by claim 1, wherein said outer edge intensity map comprises a horizontal and a vertical edge map, wherein magnitude of said horizontal and said vertical edge maps provide initial edge intensity map.

4. The method as described by claim 1, wherein said outer edge intensity map is generated for each color channel of red, green and blue (RGB) channels.

5. The method as described by claim 4, wherein outer edge intensity map for each color channel is combined to create a binary edge map.

6. The method as described by claim 1, wherein orientation edge map is used to generate said outer edge intensity map.

7. The method as described by claim 1, wherein said generating said plurality of paths on said continuous edge map comprises:

searching for a first pixel at a row of said outer edge intensity map that has an intensity greater than a threshold;

identifying a next pixel from a plurality of pixels, wherein said plurality of pixels is within a certain distance from said first pixel, and wherein said next pixel has an intensity value greater than other pixels within said plurality of pixels;

increasing said certain distance to create a new distance if said next pixel is unidentifiable and repeating said identifying using said new distance;

selecting said next pixel as a pixel in said continuous edge map; and

repeating said searching, said identifying, said increasing, and said selecting to form said plurality of paths on said continuous edge map.

8. The method as described by claim 7, wherein said intensity value is pixel intensity weighted by an orientation of pixel associated therewith.

9. The method as described by claim 1 further comprising:

prior to said generating said outer edge intensity map, locating a reference point associated with said foreground object;

determining a scope of said background object; and

cropping and centering said image based on said reference point and further based on said scope of said background object.

10. The method as described by claim 9, wherein said locating said reference point is in response to a user selection, and wherein said scope of said background object is determined based on said reference point or based on a user selection.

11. The method as described by claim 1, wherein a color associated with said foreground object is similar to a color of said background object.

12. A non-transitory computer readable storage medium having computer-readable program code stored thereon for causing a processor to implement a method comprising:

generating an outer edge intensity map associated with a foreground object of an image;

generating a plurality of paths on a continuous edge map from said outer edge intensity map, wherein said plurality of paths on said continuous edge map is created by connecting pixels in said outer edge intensity map; and

forming an edge mask by selecting one path on said continuous edge map from said plurality of paths on said continuous edge map, wherein said selecting is based on a highest average edge intensity.

13. The non-transitory computer readable storage medium as described by claim 12, wherein said generating said plurality of paths on said continuous edge map comprises:

searching for a first pixel at a row of said outer edge intensity map that has an intensity greater than a threshold;

identifying a next pixel from a plurality of pixels, wherein said plurality of pixels is within a certain distance from said first pixel, and wherein said next pixel has an intensity value greater than other pixels within said plurality of pixels;

increasing said certain distance to create a new distance if said next pixel is unidentifiable and repeating said identifying using said new distance;

selecting said next pixel as a pixel in said continuous edge map; and

repeating said searching, said identifying, said increasing, and said selecting to form said plurality of paths on said continuous edge map.

14. The non-transitory computer readable storage medium as described by claim 13, wherein said intensity value is pixel intensity weighted by an orientation of pixel associated therewith.

15. The non-transitory computer readable storage medium as described by claim 12, wherein said method further comprises:

prior to said generating said outer edge intensity map, locating a reference point associated with said foreground object;

determining a scope of said background object; and

cropping and centering said image based on said reference point and further based on said scope of said background object.

16. The non-transitory computer readable storage medium as described by claim 12, wherein Sobel filters are used to generate said outer edge intensity map.

17. The non-transitory computer readable storage medium as described by claim 12, wherein orientation edge map is used to generate said outer edge intensity map.

18. A computer system comprising:

a processor;

a display; and

a memory coupled to said processor and said display, wherein said memory comprises computer-readable program code stored thereon for causing said processor to implement a method comprising;

generating an outer edge intensity map associated with a foreground object of an image;

generating a plurality of paths on a continuous edge map from said outer edge intensity map, wherein said plurality of paths on said continuous edge map is created by connecting pixels in said outer edge intensity map; and

forming an edge mask by selecting one path on said continuous edge map from said plurality of paths on said continuous edge map, wherein said selecting is based on a highest average edge intensity.

19. The computer system as described by claim 18, wherein said generating said plurality of paths on said continuous edge map comprises:

searching for a first pixel at a row of said outer edge intensity map that has an intensity greater than a threshold;

identifying a next pixel from a plurality of pixels, wherein said plurality of pixels is within a certain distance from said first pixel, and wherein said next pixel has an intensity value greater than other pixels within said plurality of pixels;

increasing said certain distance to create a new distance if said next pixel is unidentifiable and repeating said identifying using said new distance;

selecting said next pixel as a pixel in said continuous edge map; and

repeating said searching, said identifying, said increasing, and said selecting to form said plurality of paths on said continuous edge map.

20. The computer system as described by claim 18, wherein Sobel filters are used to generate said outer edge intensity map or an orientation edge map is used to generate said outer edge intensity map.