Occlusion Tracking Using Logical Models

1
Occlusion Tracking Using Logical Models

• Summary

2
A Variational Partial Differential Equations
based model is used for tracking objects under
occlusions and in complex backgrounds.

• Two assumptions are needed to successfully detect
  and segment the object
• Availability of prior shaped
• Affine movement.
• The two possible logic models(EnergyImage/Shape)
  to segment the object with different instances
  are
• And logic model - Intersection ( )
• Or logic model- Union ( )

3
The segmentation energy is of the form
4
The correct logical model combination depends
on the occlusion's intensity relative to the
object.

• (and) similar intensity

• (or)dissimilar intensity

The correct logic model is the one
whichminimizes the quantity.
5
The general algorithm

• Active contour ( ), affine transformation (g)
  and current frame (f) are used to generate an
  initial guess for the current frame.
• Compute and g via gradient descent on the
  AND/OR energy with initial data.
• Choose the minimum, using.

6
The limitations are

• No explicit motion model,
• Sensitive to local minima when the quality of the
  image is low.
• Does not work with
• total occlusions.

• Computationally expensive.
• If motion is too great
• the sequential segmentation does not work.