Method and apparatus for bilayer image segmentation
View Patent ↗A method and an apparatus for bilayer image segmentation are described. A set of segmentation seeds for the image is generated by analyzing a depth histogram of the image. Then a segmentation map is generated by minimizing an objective function, which models a directed flow from the foreground segmentation seeds towards the background segmentation seeds.
1. A method for bilayer segmentation of an image from a sequence of images comprising:
generating a set of segmentation seeds for the image by analyzing a temporal stack of depth histograms of the sequence of images; and
generating a segmentation map by minimizing an objective function, which models a directed flow from the foreground segmentation seeds towards the background segmentation seeds.
2. The method according to claim 1 , further comprising computing the objective function taking into account color data and depth data.
3. The method according to claim 2 , further comprising analyzing distances in color space and in depth space for the objective function.
4. The method according to claim 1 , further comprising minimizing the objective function using a graph-cut approach.
5. The method according to claim 1 , further comprising extracting an optimal bisection from the stack of depth histograms using a dynamic programming approach.
6. The method according to claim 5 , wherein at each frame the bisecting depth value is used to detect segmentation seeds.
7. An apparatus for bilayer segmentation of an image from a sequence of images comprising:
a depth histogram analyzer configured to analyze a temporal stack of depth histograms of the sequence of images;
a segmentation seed generator configured to generate a set of segmentation seeds for the image using the result output by the depth histogram analyzer;
a function minimizer configured to minimize an objective function, which models a directed flow from the foreground segmentation seeds towards the background segmentation seeds; and
a segmentation map generator configured to generate a segmentation map using the result output by the function minimizer.
8. The apparatus according to claim 7 , wherein the function minimizer is further configured to compute the objective function taking into account color data and depth data.
9. The apparatus according to claim 8 , wherein the function minimizer is further configured to analyze distances in color space and in depth space for the objective function.
10. The apparatus according to claim 7 , wherein the function minimizer is further configured to minimize the objective function using a graph-cut approach.
11. The apparatus according to claim 7 , wherein the segmentation seed generator is further configured to extract an optimal bisection from the stack of depth histograms using a dynamic programming approach.
12. The apparatus according to claim 11 , wherein the segmentation seed generator is further configured to use the bisecting depth value at each frame to detect segmentation seeds.
13. A computer non transitory readable storage medium having stored therein instructions enabling bilayer segmentation of an image from a sequence of images, which when executed by a computer, cause the computer to:
generate a set of segmentation seeds for the image by analyzing a temporal stack of depth histograms of the sequence of images; and
generate a segmentation map by minimizing an objective function, which models a directed flow from the foreground segmentation seeds towards the background segmentation seeds.
14. The computer non transitory readable storage medium according to claim 13 , wherein the instructions further cause the computer to compute the objective function taking into account color data and depth data.
15. The computer non transitory readable storage medium according to claim 14 , wherein the instructions further cause the computer to analyze distances in color space and in depth space for the objective function.
16. The computer non transitory readable storage medium according to claim 13 , wherein the instructions further cause the computer to minimize the objective function using a graph-cut approach.
17. The computer non transitory readable storage medium according to claim 13 , wherein the instructions further cause the computer to extract an optimal bisection from the stack of depth histograms using a dynamic programming approach.
18. The computer non transitory readable storage medium according to claim 17 , wherein the instructions further cause the computer to use the bisecting depth value at each frame to detect segmentation seeds.
19. An apparatus for bilayer segmentation of an image from a sequence of images comprising:
means for generating a set of segmentation seeds for the image by analyzing a temporal stack of depth histograms of the sequence of images; and
means for generating a segmentation map by minimizing an objective function, which models a directed flow from the foreground segmentation seeds towards the background segmentation seeds.
20. The apparatus according to claim 19 , further comprising means for computing the objective function taking into account color data and depth data.
21. The apparatus according to claim 20 , further comprising means for analyzing distances in color space and in depth space for the objective function.
22. The apparatus according to claim 19 , further comprising means for minimizing the objective function using a graph-cut approach.
23. The apparatus according to claim 19 , further comprising means for extracting an optimal bisection from the stack of depth histograms using a dynamic programming approach.