A novel framework for motion segmentation and tracking by clustering incomplete trajectories.
Vasileios KaravasilisKonstantinos BlekasChristophoros NikouPublished in: Comput. Vis. Image Underst. (2012)
Keyphrases
- motion segmentation
- moving objects
- spectral clustering
- missing data
- feature trajectories
- image sequences
- motion model
- motion estimation
- multibody
- clustering algorithm
- dynamic scenes
- model selection
- multiple frames
- epipolar geometry estimation
- point trajectories
- appearance model
- optical flow
- motion analysis
- motion field
- mean shift
- clustering method
- unsupervised learning
- motion sequences
- depth ordering
- optical flow algorithm
- semi supervised
- k means