METHODS AND SYSTEMS FOR OBJECT SEGMENTATION IN DIGITAL IMAGES

• US 20110085028A1
• Filed: 10/14/2009
• Published: 04/14/2011
• Est. Priority Date: 10/14/2009
• Status: Active Grant

• Current Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Original Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Inventors: Malzbender, Thomas G., Samadani, Ramin
• Granted Patent: US 8,199,165 B2
• Time in Patent Office: Days
• US Class Current: 348/51
• CPC Class Codes: G06T 2207/10016 Video; Image sequence
• CPC Class Codes: G06T 2207/10024 Color image
• CPC Class Codes: G06T 2207/20012 Locally adaptive
• CPC Class Codes: G06T 2207/20068 Projection on vertical or horizontal image axis
• CPC Class Codes: G06T 2207/30201 Face
• CPC Class Codes: G06T 7/11 Region-based segmentation
• CPC Class Codes: G06V 10/507 Summing image-intensity values; Histogram projection analysis
• CPC Class Codes: G06V 40/162 using pixel segmentation or colour matching

Abstract

Various embodiments of the present invention are directed to object segmentation of digital video streams and digital images. In one aspect, a method for segmenting an object in a digitally-encoded image includes designing a non-linear local function that generates function values associated with identifying the object of interest and a combination function that combines the function values, the object of interest encoded in digital image data. The method includes forming orthogonal projections of the digital image data based on the function values and the combination function. In addition, the orthogonal projections can be used to determining boundaries segmenting the object in the digital image data. The method also includes extracting an image of the object that lies within the boundaries.

Claims

1. A method for segmenting an object of interest in a digitally-encoded image using a computing device, the method comprising:

designing a non-linear local function that generates function values associated with identifying the object of interest encoded in digital image data and a combination function that combines the function values;

forming orthogonal projections of the digital image data using the computing device, the orthogonal projections based on the function values and the combination function;

based on the orthogonal projections, determining boundaries segmenting the object in the digital image data using the computing device; and

extracting an image of the object that lies within the boundaries using the computing device.

2. The method of claim 1 wherein the non-linear local function further comprises a mapping of pixel values associated with a neighborhood of a pixel to a single value for the pixel.

3. The method of claim 1 wherein the non-linear local function further comprises a non-linear filter function.

4. The method of claim 1 wherein forming orthogonal projections of the digital image data further comprises filtering pixels values based on luminance.

5. The method of claim 1 wherein forming orthogonal projections of the digital image data further comprises filtering pixels values based on depth.

6. The method of claim 1 wherein forming orthogonal projections of the digital image data further comprises filtering pixels values based on color.

7. The method of claim 1 wherein the combination function combines the output of the local functions along one direction of the image data.

8. The method of claim 1 wherein forming the orthogonal projections of the digital image data further comprises:

forming a first projection of the digital image data in a first direction using the computing device, the first projection based on the filter function and the combination function; and

forming a second projection of the digital image data in a second direction using the computing device, the second projection based on the filter function and the combination function, and the second direction orthogonal to the first direction.

9. The method of claim 1 wherein determining boundaries of the object in the image data further comprise determining where each orthogonal projection equals zero.

10. The method of claim 1 wherein extracting the image of the object that lies within the boundaries further comprises cropping the image data associated with the object by discarding image data located outside the boundaries.

11. A computer-readable medium having instructions encoded thereon for enabling a processor to perform the operations of:

designing a non-linear local function that generates function values associated with identifying the object of interest encoded in digital image data and a combination function that combines the function values;

forming orthogonal projections of the digital image data, the orthogonal projections based on the function values and the combination function;

based on the orthogonal projections, determining boundaries segmenting the object in the image data; and

extracting an image of the object that lies within the boundaries.

12. The medium of claim 11 wherein the non-linear local function further comprises a mapping of pixel values associated with a neighborhood of a pixel to a single value for the pixel.

13. The medium of claim 11 wherein the non-linear local function further comprises a non-linear filter function.

14. The medium of claim 11 wherein forming orthogonal projections of the digital image data further comprises filtering pixels values based on luminance.

15. The medium of claim 11 wherein forming orthogonal projections of the digital image data further comprises filtering pixels values based on depth.

16. The medium of claim 11 wherein forming orthogonal projections of the digital image data further comprises filtering pixels values based on color.

17. The medium of claim 11 wherein the combination function combines the output of the local functions along one direction of the image data.

18. The medium of claim 11 wherein forming the orthogonal projections of the digital image data further comprises:

forming a first projection of the digital image data in a first direction using the computing device, the first projection based on the filter function and the combination function; and

forming a second projection of the digital image data in a second direction using the computing device, the second projection based on the filter function and the combination function, and second direction orthogonal to the first direction.

19. The medium of claim 11 wherein determining boundaries of the object in the image data further comprise determining where each orthogonal projection equals zero.

20. The medium of claim 11 wherein extracting the image of the object that lies within the boundaries further comprises cropping the image data associated with the object by discarding image data located outside the boundaries.