Motion detection using normal optical flow
First Claim
1. A method of detecting motion comprising:
- obtaining a plurality of images of an object over a predetermined period of time;
generating a motion region image of the object from the plurality of images;
tracking regions of the motion region image from a current region to an older region; and
identifying movement based on the tracking regions of the motion region image, wherein generating the motion region image comprises;
isolating the object in the plurality of images from a background to form a plurality of silhouettes;
overlapping the plurality of silhouettes; and
setting all pixels of the overlapped silhouettes to an equal value.
1 Assignment
0 Petitions
Accused Products
Abstract
A system and method allow for the detection, segmentation and analysis of regions of motion. The system and method use a motion history image generated by observing an object over a given period of time. The motion history image contains numerous regions captured at different time intervals. In one method, the most recent region(s) are located and adjacent older regions are labeled with a unique identifier to perform a down fill operation. Each down fill operation can identify a single motion contained in the motion history image. If a down filled region of the motion history image contains more than one movement, separate up fill operations can be performed to help separate the movements. Each up fill operation begins at a set of oldest or lowest value regions and progresses toward the most recent region. The down fill and up fill operations can be combined to provide an image mask(s) to separate the multiple motions from the motion history image.
73 Citations
12 Claims
-
1. A method of detecting motion comprising:
-
obtaining a plurality of images of an object over a predetermined period of time;
generating a motion region image of the object from the plurality of images;
tracking regions of the motion region image from a current region to an older region; and
identifying movement based on the tracking regions of the motion region image, wherein generating the motion region image comprises;
isolating the object in the plurality of images from a background to form a plurality of silhouettes;
overlapping the plurality of silhouettes; and
setting all pixels of the overlapped silhouettes to an equal value. - View Dependent Claims (2, 3, 4, 5, 6)
labeling contents of each region of the motion region image with a unique identifier, wherein the unique identifier is selected based upon an age of the region being labeled.
-
-
3. The method of claim 2 wherein the movement is identified by a placement of the unique identifiers.
-
4. The method of claim 1 further comprising tracking regions of the motion region image from the older region to the current region.
-
5. The method of claim 4 further comprising generating a motion region image mask that can be used in combination with the plurality of images to identify regions containing the detected motion.
-
6. A computer readable medium comprising instructions to instruct a processor to perform the method of claim 1.
-
7. A method of detecting motion comprising:
-
obtaining a plurality of images of an object using a camera, the plurality of images are obtained over a predetermined sliding window of time;
generating a motion region image of the object from the plurality of images by isolating the object in the plurality of images from a background;
performing a down fill operation on the motion region image, wherein the down fill operation tracks regions of the motion region image from a current region to an older region, and labels each region of the motion region image with a first unique identifier, wherein the first unique identifier is selected based upon an age of the region being labeled;
performing at least one up fill operation on the motion region image, wherein the up fill operation tracks regions of the motion region image from the older region to the current region, and labels contents of each region of the motion region image with a second unique identifier; and
identifying movement based on the down fill operation and the at least one up fill operation. - View Dependent Claims (8, 9, 10, 11, 12)
combining results of the down fill operation and the at least one up fill operation to generate a motion region image mask that can be used in combination with the plurality of images to identify regions containing the detected motion.
-
-
9. The method of claim 7 wherein the down fill operation comprises:
-
labeling pixels of a first region with a first identifier;
scanning a boundary of the first region to identify pixels in a second region, located outside of the boundary of the first region, captured one time period prior to the pixels of the first region; and
labeling pixels of the second region with a second identifier.
-
-
10. The method of claim 7 wherein the object is a person.
-
11. The method of claim 7 further comprising:
combining results of the down fill operation and the at least one up fill operation to generate multiple motion region image masks that can be individually used in combination with the plurality of images to identify multiple regions containing motion.
-
12. A computer readable medium comprising instructions to instruct a processor to perform the method of claim 7.
Specification