Video Surveillance and Activity Monitoring Motion Compensation
Approach:	Our approach is based on multigrid matching of the features points extracted on the reference and target frame. We first match, using a correlation of the gray levels, a set of feature points on both images at the lowest resolution, recover a set of affine parameters by solving a linear set of equations and iterate by increasing image resolution.
Publications:	Detection and Tracking of Objects in Airborne Video Imagery, Interpretation of Visual Motion Workshop, CVPR98, Santa Barbara, June 1998. Detecting and Tracking Objects in Video Surveillance, Proc. of the IEEE Computer Vision and Pattern Recognition 99, Fort Collins, June 1999.
Results: VSAM Project	Original image sequence Stabilized image sequence Mosaic Original image sequence Stabilized image sequence Mosaic
Last modified: November 2006 USC Computer Vision