Circuito integrado para extração de fundo não homogêneo de imagens dinâmicas em tempo real visando baixo custo.

Abstract

This thesis describes the implementation of part of a system, aimed at making the extraction of objects from images with static and non-homogeneous backgrounds in real time, capable of tolerating slight global and local brightness variations. This system proposes a re-configurable hardware implementation of a background subtraction algorithm, which was adapted in order to allow a low cost hardware implementation, and real time operation, to render it suitable for applications in the entertainment industry. Simulation results are presented, which validate the algorithm functionality for such applications. A GPEP-01 (General Purpose Image Processor) platform is also presented. This platform has been designed and implemented to achieve the minimum requirements of a prototyping embedded system for real time image processing applications, needed for development of the proposed system.