Method and device for continuous figure-ground segregation in images from dynamic visual scenes

• US 8,126,269 B2
• Filed: 11/14/2008
• Issued: 02/28/2012
• Est. Priority Date: 11/16/2007
• Status: Active Grant

• Current Assignee: Honda Research Institute Europe GmbH (Honda Motor Company)
• Original Assignee: Honda Research Institute Europe GmbH (Honda Motor Company)
• Inventors: Eggert, Julian, Weiler, Daniel
• Primary Examiner(s): Ahmed, Samir
• Assistant Examiner(s): Bayat, Ali
• Application Number:
  US12/271,696
• Publication Number:
  US 20090129675A1
• Time in Patent Office: 1,201 Days
• Field of Search: 382/103, 382/107, 382/164, 382/173, 382/176, 382/190, 382/236, 382/243, 382/302, 358/453, 348/169, 235/411, 375/240.16
• US Class Current: 382/173
• CPC Class Codes: G06T 7/215 Motion-based segmentation

Abstract

A method for segregating a figure region from a background region in image sequences from dynamic visual scenes, comprising the steps of: (a) acquiring an image; (b) determining local motion estimations and confidences for each position of the image; (c) modifying a level-set function by moving and distorting the level-set function with the local motion estimations and smearing it based on the local motion confidences, to generate a predicted level set function that is geometrically in correspondence with the image and diffused at positions where the confidence of the motion estimation is low; (d) obtaining input features of the image by using a series of cues; (e) calculating a mask segregating the figure region from the background region of the image using the modified level-set function and the obtained input features; (f) extracting the figure region from the image; and repeating steps (a)-(f) until a termination criterion is satisfied.

Claims

1. A method for segregating a figure region from a background region in image sequences from dynamic visual scenes, comprising the steps of:

(a) acquiring an image;

(b) determining local motion estimations and confidences for each position of the image;

(c) modifying a level-set function by moving and distorting the level-set function with the local motion estimations and smearing it based on the local motion confidences, to generate a predicted level set function that is geometrically in correspondence with the image and diffused at positions where the confidence of the motion estimation is low;

(d) obtaining input features of the image by using a series of cues;

(e) calculating a mask segregating the figure region from the background region of the image using the modified level-set function and the obtained input features;

(f) extracting the figure region from the image; and

repeating steps (a)-(f) until a termination criterion is satisfied.

2. The method of to claim 1, further comprising the step of:

calculating a global motion pattern for the figure region;

wherein the results from the local motion estimation as well as the global motion pattern calculation are used in the modifying the level-set function step.

3. The method of claim 2, wherein the cues comprise motion information determined in the determining local motion step or the calculating a global motion step.

4. The method of claim 3, wherein the cues further comprise cues from an infrared camera system.

5. The method of claim 1, wherein the image acquisition occurs via cameras mounted on a moving platform such as a robot or a car.

6. The method of claim 1, wherein at least one of the input images or the motion estimation include mechanisms that compensate for an expected motion of the objects in the input image and of the level-set function and the object contour from the segmentation.

7. The method of claim 1, wherein the figure-ground segregation step incorporates information and constraints about an objects expected characteristic form or about its contour properties.

8. The method of claim 1 for controlling a robot or a car.

9. A computer program product embodied on a non-transitory computer readable medium which when executed performs the method steps of claim 1.

10. A device for segregating a figure region from a background region in image sequences from dynamic visual scenes, comprising:

acquiring means for acquiring an image;

determining means for determining local motion estimations and confidences for each position of the image;

modifying means for modifying a level-set function by moving and distorting it with the local motion estimations and smearing it based on the local motion confidences, resulting in a predicted level set function that is geometrically in correspondence with the image and diffused at positions where the confidence of the motion estimation is low;

features means for obtaining input features of the image by using a series of cues;

mask means for calculating a mask segregating the figure region from the background region of the image using the modified level-set function and the obtained input features;

extraction means for extracting the figure region from the image; and

termination means for checking whether a termination criterion is met.

11. The device of claim 10, wherein the device is one of a robot or a car.