Video object cut and paste
First Claim
Patent Images
1. A method, comprising:
- dividing frames of a video sequence into regions prior to applying a 3-D graph cut segmentation for designating an outline of a video object in the video sequence;
constructing a 3-dimensional graph including embedding temporal coherence in the 3-dimensional graph by forming associations between corresponding regions in adjacent video frames;
applying the 3-D graph cut segmentation to the 3-dimensional graph according to a global color model to derive a binary segmentation representing the outline of the video object; and
applying a 2-D graph cut segmentation to at least some of the binary segmentation according to a local color model to obtain a refined outline of the video object.
2 Assignments
0 Petitions
Accused Products
Abstract
Video object cutting and pasting is described. In one implementation, pre-segmentation of video frames into regions is performed prior to a 3-D graph cut segmentation. The 3-D graph cut segmentation uses temporal coherence and a global color model to achieve accuracy of video object boundaries. A 2-D local graph cut segmentation can then be used to refine the boundaries. The boundaries can be tracked within a user-selected sequence of windows and refined using a local color model.
85 Citations
20 Claims
-
1. A method, comprising:
-
dividing frames of a video sequence into regions prior to applying a 3-D graph cut segmentation for designating an outline of a video object in the video sequence;
constructing a 3-dimensional graph including embedding temporal coherence in the 3-dimensional graph by forming associations between corresponding regions in adjacent video frames;
applying the 3-D graph cut segmentation to the 3-dimensional graph according to a global color model to derive a binary segmentation representing the outline of the video object; and
applying a 2-D graph cut segmentation to at least some of the binary segmentation according to a local color model to obtain a refined outline of the video object. - View Dependent Claims (2, 3, 4, 5)
-
-
6. A method, comprising:
-
pre-segmenting frames of a video sequence into regions;
selecting two model frames of the video sequence, wherein each of the two model frames has a foreground representing a video object, and a background;
constructing a 3-dimensional (3-D) graph from a 3-D volume of the frames temporally bounded by the two model frames, including associating regions on a single frame with adjacent regions on the same frame and associating the regions on the single frame with candidate corresponding regions on adjacent frames; and
segmenting the 3-D graph into associated foreground regions and associated background regions according to a global color model, wherein the associated foreground regions represent the video object in the frames of the video sequence. - View Dependent Claims (7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19)
-
-
20. A system, comprising:
-
means for determining visual regions that endure from frame to frame within a video sequence;
means for building a 3-dimensional graph from the regions of the video sequence;
means for embedding temporal coherence in the 3-dimensional graph by including associations between corresponding regions in adjacent frames of the video sequence;
means for applying a 3-dimensional graph cut segmentation to the 3-dimensional graph, based on global colors of the video sequence, in order to obtain segmentation results;
means for designating a local part of the segmentation results; and
means for applying a 2-dimensional graph cut segmentation to the local part based on local colors of the local part.
-
Specification