Realistic Implementation of the Facial Tracking and the Scenario Interaction on Android Robot

摘要

The aim of facial tracking is to follow the selected moving people through a sequence of images detected from the artificial vision. In this paper, the artificial vision will be integrated with an intelligent platform of the android robot. However, existing vision modules with artificial intelligence are still a trade-off issue of how to reduce the steady-state error of the image identification with faster tracking speed. This paper proposes an inexpensive, however, robust methodology with PC-based fuzzy-logic controller to facilitate the signal transmission, image acquisition, and fast identification. The image detecting, gripping, recording, and code comparing will be proceeded with the speed of 0.1 sec/frame for face discrimination and 0.2 sec/frame for data processing. The proposed tracking system can grab moving target by a set of coupled-lens CCD cameras. Finally, computer simulations and experiments are accomplished and can successfully justify the advantages of the proposed methodology.