A Tensor Based on Optical Flow for Global Description of Motion in Videos.
Virgínia Fernandes MotaEder de Almeida PerezMarcelo Bernardes VieiraLuiz Maurilio MacielFrédéric PreciosoPhilippe Henri GosselinPublished in: SIBGRAPI (2012)
Keyphrases
- optical flow
- motion features
- moving camera
- motion field
- motion model
- image sequences
- global motion
- camera motion
- optical flow estimation
- motion estimation
- independently moving objects
- structure tensor
- motion segmentation
- motion parameters
- image motion
- motion analysis
- flow field
- moving objects
- video sequences
- consecutive frames
- successive frames
- optical flow computation
- optical flow fields
- dynamic textures
- motion patterns
- dynamic scenes
- video surveillance
- velocity field
- video frames
- d scene
- visual motion
- ego motion
- vector field
- computation of optical flow
- video content
- video stabilization
- key frames
- high order
- periodic motion
- feature tracking
- high frame rate
- optical flow algorithm
- motion discontinuities
- motion trajectories
- dense optical flow
- space time
- static images
- adjacent frames
- recognition of human actions
- scene flow
- motion capture data
- action recognition
- visual features
- activity recognition
- visual odometry
- motion boundaries
- motion estimates
- motion cues
- rigid motion
- smoothness constraint
- computer vision
- video data
- motion vectors