An Innovative SIFT-Based Method for Rigid Video Object Recognition

Figure 2

The Relationship between frames and feature models. Each frame is transformed into feature view; similar object views contained in different frames are clustered into one or more feature models.