Motion Segmentation by Spatiotemporal Smoothness Using 5D Tensor Voting

@article{Min2006MotionSB,
  title={Motion Segmentation by Spatiotemporal Smoothness Using 5D Tensor Voting},
  author={Changki Min and G{\'e}rard G. Medioni},
  journal={2006 Conference on Computer Vision and Pattern Recognition Workshop (CVPRW'06)},
  year={2006},
  pages={199-199}
}
Our goal is to recover temporal trajectories of all pixels in a reference image for the given image sequence, and segment the image based on motion similarities. These trajectories can be visualized by observing the 3D (x, y, t) spatiotemporal volume. The mathematical formalism describing the evolution of pixels in time is that of fiber bundles, but it is difficult to implement directly. Instead, we express the problem in a higher dimensional 5D space, in which pixels with coherent apparent… CONTINUE READING