Method and apparatus for generating superpixel clusters

Patent metadata

Field	Value
Publication number	US-2016104294-A1
Application number	US-201514877791-A
Country	US
Kind code	A1
Filing date	Oct 7, 2015
Priority date	Oct 8, 2014
Publication date	Apr 14, 2016
Grant date	—

Abstract

A method and an apparatus for generating a superpixel cluster for an image or a sequence of images. A primary clustering unit generates a primary superpixel cluster using a fixed reference superpixel, whereas a secondary clustering unit generates two or more secondary superpixel clusters using a propagating reference superpixel. A combining unit then combines intersections between the primary superpixel cluster and the two or more secondary superpixel clusters to generate a final superpixel cluster.

First claim

1 . A method for generating a superpixel cluster for an image or a sequence of images, the method comprising:
generating a primary superpixel cluster using a fixed reference superpixel; generating two or more secondary superpixel clusters using a propagating reference superpixel; and combining intersections between the primary superpixel cluster and the two or more secondary superpixel clusters to generate a final superpixel cluster.

2 . The method according to claim 1 , wherein the primary superpixel cluster and the two or more secondary superpixel clusters are generated by analyzing distances between properties of a superpixel and the respective reference superpixel.

3 . The method according to claim 1 , wherein the primary superpixel cluster is generated based on a first property of the superpixels and the two or more secondary superpixel clusters are generated based on a second property of the superpixels.

4 . The method according to claim 3 , wherein the first property is color information and the second property is depth information.

5 . The method according to claim 1 , wherein a superpixel is marked as tentatively accepted if a distance between a selected property of the superpixel relative to the respective reference superpixel does not exceed a respective threshold.

6 . The method according to claim 5 , wherein a superpixel marked as tentatively accepted is only affiliated to the final superpixel cluster if the superpixel is marked as tentatively accepted both in the primary superpixel cluster and in at least one of the two or more secondary superpixel clusters.

7 . The method according to claim 1 , wherein the primary superpixel cluster and the two or more secondary superpixel clusters are generated from superpixels associated to the same image.

8 . The method according to claim 1 , wherein the primary superpixel cluster and the two or more secondary superpixel clusters are generated from superpixels associated to the same image and superpixels associated to different images of the sequence of images.

9 . An apparatus configured to generate a superpixel cluster for an image or a sequence of images, the apparatus comprising:
a primary clustering unit configured to generate a primary superpixel cluster using a fixed reference superpixel; a secondary clustering unit configured to generate two or more secondary superpixel clusters using a propagating reference superpixel; and a combining unit configured to combine intersections between the primary superpixel cluster and the two or more secondary superpixel clusters to generate a final superpixel cluster.

10 . A computer readable non-transitory storage medium having stored therein instructions enabling generating a superpixel cluster for an image or a sequence of images, which when executed by a computer, cause the computer to:
generate a primary superpixel cluster using a fixed reference superpixel; generate two or more secondary superpixel clusters using a propagating reference superpixel; and combine intersections between the primary superpixel cluster and the two or more secondary superpixel clusters to generate a final superpixel cluster.

11 . The apparatus according to claim 9 , wherein the primary clustering unit is configured to generate the primary superpixel cluster by analyzing distances between properties of a superpixel and the respective reference superpixel and the secondary clustering unit is configured to generate the two or more secondary superpixel clusters by analyzing distances between properties of a superpixel and the respective reference superpixel.

12 . The apparatus according to claim 9 , wherein the primary clustering unit is configured to generate the primary superpixel cluster based on a first property of the superpixels and the secondary clustering unit is configured to generate the two or more secondary superpixel clusters based on a second property of the superpixels.

13 . The apparatus according to claim 12 , wherein the first property is color information and the second property is depth information.

14 . The apparatus according to claim 9 , wherein the apparatus is configured to mark a superpixel as tentatively accepted if a distance between a selected property of the superpixel relative to the respective reference superpixel does not exceed a respective threshold.

15 . The apparatus according to claim 14 , wherein the apparatus is configured to affiliate a superpixel marked as tentatively accepted to the final superpixel cluster only if the superpixel is marked as tentatively accepted both in the primary superpixel cluster and in at least one of the two or more secondary superpixel clusters.

16 . The apparatus according to claim 9 , wherein the primary clustering unit and the secondary clustering unite are configured to generate the primary superpixel cluster and the two or more secondary superpixel clusters, respectively, from superpixels associated to the same image.

17 . The apparatus according to claim 9 , wherein the primary clustering unit and the secondary clustering unite are configured to generate the primary superpixel cluster and the two or more secondary superpixel clusters, respectively, from superpixels associated to the same image and superpixels associated to different images of the sequence of images.

18 . The computer readable non-transitory storage medium according to claim 10 , wherein the instructions cause the computer to generate the primary superpixel cluster and the two or more secondary superpixel clusters by analyzing distances between properties of a superpixel and the respective reference superpixel.

19 . The computer readable non-transitory storage medium according to claim 10 , wherein the instructions cause the computer to generate the primary superpixel cluster based on a first property of the superpixels and to generate the two or more secondary superpixel clusters based on a second property of the superpixels.

20 . The computer readable non-transitory storage medium according to claim 19 , wherein the first property is color information and the second property is depth information.

21 . The computer readable non-transitory storage medium according to claim 10 , wherein the instructions cause the computer to mark a superpixel as tentatively accepted if a distance between a selected property of the superpixel relative to the respective reference superpixel does not exceed a respective threshold.

22 . The computer readable non-transitory storage medium according to claim 21 , wherein the instructions cause the computer to affiliate a superpixel marked as tentatively accepted to the final superpixel cluster only if the superpixel is marked as tentatively accepted both in the primary superpixel cluster and in at least one of the two or more secondary superpixel clusters.

23 . The computer readable non-transitory storage medium according to claim 10 , wherein the instructions cause the computer to generate the primary superpixel cluster and the two or more secondary superpixel clusters from superpixels associated to the same image.

24 . The computer readable non-transitory storage medium according to claim 10 , wherein the instructions cause the computer to generate the primary superpixel cluster and the two or more secondary superpixel clusters from superpixels associated to the same image and superpixels associated to different images of the sequence of images.

Assignees

• Thomson Licensing

Inventors

• Gandolph Dirk
• Putzke-Roeming Wolfram

Classifications

• G06T2207/10024

  Color image · CPC title

• G06T2207/10028

  Range image; Depth image; 3D point clouds · CPC title

• G06T7/11

  Region-based segmentation · CPC title

• G06T7/187 Primary

  involving region growing; involving region merging; involving connected component labelling · CPC title

• G06T2207/20101

  Interactive definition of point of interest, landmark or seed · CPC title