Evolving Novel Coefficient Sets for Optimized Reconstruction of Quantized One-Dimensional (1-D) and Two-Dimensional (2-D) Signals. 2004 Visiting Faculty Research Program (VFRP) In-House Work at AFRL/IFTA

摘要

This paper describes a genetic algorithm that evolves optimized sets of coefficients for signal reconstruction under lossy conditions due to quantization. Beginning with a population of mutated copies of the set of coefficients describing a wavelet-based inverse transform, our genetic algorithm systemically evolves a new set of coefficients that significantly reduces mean squared error (relative to the performance of the selected wavelet) for various classes of one- and two-dimensional signals.