Interactive segmentation of images with single scribbles

• US 8,594,428 B2
• Filed: 07/25/2007
• Issued: 11/26/2013
• Est. Priority Date: 07/25/2006
• Status: Active Grant

• Current Assignee: HumanEyes Technologies Limited
• Original Assignee: HumanEyes Technologies Limited
• Inventors: Aharoni, Nadav, Zomet, Assaf, Gonen, Yoav
• Primary Examiner(s): Alavi, Amir
• Application Number:
  US12/309,532
• Publication Number:
  US 20090304280A1
• Time in Patent Office: 2,316 Days
• Field of Search: 382/173, 382/190, 382/284, 345/179, 345/326, 345/348, 345/589, 345/592, 345/629, 715/236, 715/863
• US Class Current: 382/180
• CPC Class Codes: G06T 2207/10016 Video; Image sequence
• CPC Class Codes: G06T 2207/20072 Graph-based image processing
• CPC Class Codes: G06T 2207/20096 Interactive definition of curve of interest
• CPC Class Codes: G06T 7/11 Region-based segmentation
• CPC Class Codes: G06T 7/162 involving graph-based methods
• CPC Class Codes: G06T 7/194 involving foreground-background segmentation

Abstract

A computer-implemented method for assigning attributes to an image. The method comprises allowing a user to mark a plurality of pixels of an image as a single marked area by a single scribble, the single marked area defines, for a current iteration, a target attribute, determining according to the target attribute an optimal function that defines at least one respective attribute of pixels in the image, assigning the at least one respective attribute to pixels in the image according to the optimal function, and displaying the at least one attribute of the pixels.

Claims

1. A computer-implemented method for assigning attributes to an image, the method comprising:

displaying an unsegmented image;

marking by a user a plurality of pixels in a region of the unsegmented image as a single marked area by a single scribble covering the plurality of pixels, the single marked area defines, for a current iteration, at least one target attribute for the region;

determining according to the target attribute an optimal function that defines at least one respective attribute of pixels in the region;

assigning the at least one respective attributes to pixels in the region according to the optimal function; and

segmenting the unsegmented image in accordance with the at least one respective attribute.

2. The method according to claim 1, further including inputting a sensitivity measure to determine the size of the region that is influenced by the marked area.

3. The method according to claim 1, further including constraining pixels in the marked area to have said target attribute.

4. The method according to claim 1, further including applying an additional input constraint to pixels in the unsegmented image.

5. The method according to claim 1, wherein an additional input constraint is a given assignment of attributes to pixels in the unsegmented image.

6. The method according to claim 5, wherein the given assignment is the result of a previous iteration.

7. The method according to claim 1, wherein said image is an unsegmented image.

8. The method according to claim 1, wherein said single marked area having at least two sub areas each having a different of a plurality of colors.

9. A method for assigning properties to an image or a video sequence in a video space-time volume, the method comprising:

displaying an unsegmented frame of a video sequence video space-time volume;

marking by a user to use a computer selection tool during successive iterations for marking a respective single area in a region of the unsegmented frame by a single scribble covering a plurality of pixels, so as to assign at least one property to the pixels in said area;

computing at each iteration respective properties of pixels in the video sequence, given the properties of the pixels in the marked area and given the computed result at a previous iteration; and

displaying the video sequence so as to highlight at least one of the computed properties.

10. The method according to claim 9, wherein computing properties preserves properties assigned to the marked area and some of the properties of pixels in non-marked areas computed in a previous iteration.

11. The method according to claim 9, wherein the computed properties being used for segmenting a frame of the video sequence to a plurality of regions, each assigned property being respectively a classification of each image pixel to one of plurality of segments or a classification of each frame pixel at each video frame to one of plurality of segments.

12. The method according to claim 9, wherein the computed properties being used for matting of an image or a video, wherein the assigned property for each pixel is respectively the portions of which it belongs to each of plurality of layers or the portions of which it belongs to each of plurality of layers.

13. The method according to claim 9, wherein the computed properties being used for a member of a group consisting of augmenting color to an image or a video sequence, changing the colors of an image or a video sequence, and removing red-eye effects from videos and images.

14. A system (30) for assigning properties to an image or a video sequence, the system comprising:

a marking tool (33) to allow a user to mark by a single scribble during successive iterations a respective single area in a region of an unsegmented image or in an unsegmented video sequence in a a video space-time volume spanning more than a single pixel, so as to assign properties to the pixels in said area;

a computational unit (31) responsive to the properties of the marked area and to an additional input constraint for computing at each iteration respective properties of pixels in the unsegmented image or the unsegmented video and segmenting the unsegmented image or an unsegmented frame of the unsegmented video sequence in accordance with the respective properties; and

a display unit (36) coupled to the computational unit for displaying the image or video.

15. The system according to claim 14, wherein the additional input constraint is a given assignment of attributes to pixels in the image.

16. The system according to claim 15, wherein the given assignment is the result of a previous iteration.

17. The system according to claim 14, wherein the given assignment is used as a boundary constraint.

18. The system according to claim 14, wherein said additional input constraint is derived as an output from an external computer program.

19. The system according to claim 14, wherein the computational unit is adapted to compute said properties such that properties assigned to the marked area are preserved and some of the properties of pixels in non-marked areas computed in the previous iteration are preserved.

20. The system according to claim 14, further including a sensitivity selector (35) coupled to the computational unit for adjusting the sensitivity of the distance at which properties propagate away from marking made to said area.