with Prof. R. Frezza and Prof. B. S. Manjunath
(2000-2001) In this project we implemented an head tracking system able to recover the trajectory of a human head in the space (i.e. for each frame the position and orientation of the head). The algorithm uses as only input the video stream obtained by one single camera. The algorithm is markerless, it is able to track also fairly complex movements of the head, it is quite robust under varying light conditions and it is easy to initialize.
Experimental Results: Synthesized Sequences
In order to have a quantitative evaluation of the performance of the head tracking algorithm, synthetic sequences of a moving head have been rendered using Open GL. All the six parameters that univocally identify the pose of the head in the space (given a fixed reference frame) are known. Here is an example of such sequences, where the 3D model of the head has been superimposed to the synthesized movie frame by frame. Note that the superposition of the model has been done using the pose parameters of the head retrieved by the tracking algorithm.
 |
 |
 |
 |
Original sequence |
Ellipsoidal model |
Superellipsoidal model |
Detailed head mode |
Experimental Results: Real Sequences
Here there are two head tracking examples using real sequences and different head models. Note that the head movements are quite large.
 |
 |
Ellipsoidal model |
Detailed head mode |
Here there are two head tracking examples using real sequences under varying light conditions.
 |
 |
Ellipsoidal model |
Detailed head mode |
Related Publications
Click here for my laurea degree thesis describing in detail this project.