Um estudo sobre consumo energético na indexação de conteúdo visual com síntese em alto nível usando hardware reconfigurável.

Abstract

Nowadays, the biggest impact on the environment caused by computers is the power consumption. This work studies an alternative technology to the high energy consumption caused by desktop computers and servers, in the execution of a certain task, for example, visual content indexing. The main goal of this dissertation was to evaluate the energy efficiency using a method of indexing based on wavelets of a CBIR (Content Based Image Retrieval) system implemented on a FPGA- plataform. We use energy efficiency as metrics to compare the two platforms. Results show that FPGAs offer 758 times better energy efficiency than the system that we implemented previously on PC platform.