Approximate Processing and Incremental Refinement Concepts

Approximate processing and incremental reenement concepts are n e e ded for applications where i t i s d e-sirable to provide a systematic tradeoo between the quality of signal processing results and the availability of resources, such as time, bandwidth, memory, and power. We examine the impact of these concepts for three distinct application areas: 1 low-power frequency-selective FIR ltering, 2 real-time time-frequency analysis of signals and 3 DCT-based image encodinggdecoding. Results from approximate processing of signal data illustrate the practical utility of these types of systems.