The Analysis Loop

Given a person model and a scene model, we can now acquire a new image, interpret it, and update the scene and person models. To accomplish this there are several steps:

1. First, we predict the appearance of the user in the new image using the current state of our model. This is accomplished using a set of Kalman filters with simple Newtonian dynamics that operate on each blob's spatial statistics.
2. Next, for each image pixel we must measure the likelihood that it is a member of each of the blob models and the scene model. Self-shadowing and cast shadows are a particular difficulty in measuring this likelihood.
3. Resolve these pixel-by-pixel likelihoods into a support map, indicating for each pixel whether it is part of one of the blobs or of the background scene. Spatial priors and connectivity constraints are used to accomplish this resolution.
4. Update the statistical models for each blob and for the background scene; also update the dynamic models of the blobs.

Each of these steps will now be described in more detail.

 
• Predict Model Parameters
• Measure Likelihoods For Each Class
  • Shadowing.
• Determine Support Map
  • Morphology.
• Update Models
  • Blending Measurements and Priors.