Page 9 of 11

Conclusion

CAMSHIFT is a simple, computationally efficient face and colored object tracker. While acknowledging the limitation imposed by its simplicity, we can still see that CAMSHIFT tracks virtually as well as more expensive tethered trackers (Polhemus) or much more sophisticated, computationally expensive vision systems [17], and it tracks well in noisy environments. Thus, as we have shown, even though CAMSHIFT was conceived as a simple part of a larger tracking system, it has many uses right now in game and 3D graphics' control.

Adding perceptual interfaces can make computers more natural to use, more fun for games and graphics, and a better medium of communication. These new features consume more MIPs and so will take advantage of more MIPs available with future Intel® CPUs.

In this project, we designed a highly efficient face tracking algorithm rather than a more complex, higher MIPs usage algorithm. This was done because we want to be able to demonstrate compelling applications and interfaces on today’s systems in order to prepare the way for the future use of computer vision on PCs. CAMSHIFT is usable as a visual interface now, yet designed to be part of a more robust, larger tracking system in the future. CAMSHIFT will be incorporated into larger, more complex, higher MIPs-demanding modules that provide more robust tracking, posture understanding, gesture and face recognition, and object understanding. In this way, the functionality of the computer vision interface will increase with increasing Intel CPU speeds. A user will thus be able to upgrade their computer vision interface by upgrading to higher speed Intel CPUs in the future.


    Page 9 of 11