Exploration of VLIW space processor for adaboost-based applications

The increase in the levels integrations of the circuits and the request pressing for reduction execution times of the applications led to the new methodologies of implementation. This implies the design a strongly architectures skeletal and reconfigurable. This article presents a study of the space of VLIW processor for a real time system of vision based-Adaboost, making it possible to locate faces in video sequences. With an aim of comparing the performances with other existing implementations, the work consists in the identification of optimal processor parameters representing the difference between the performances and processor complexity.