Optimal Design of Wavelets via Lifting Based on Genetic Algorithm and Ant System

Lifting scheme wavelet design based on genetic algorithm (GA) and ant system (AS) is studied. First, a multi-population genetic algorithm and lifting are used to evolve wavelets. The global fitness and the local fitness are introduced in GA and they are homologous to the global elicitation information and the local one utilized by ants in AS. So the AS algorithm is then used to find the exact wavelet that is best adapted to the given application of the wavelet transform, and the resulting stochastic optimization method with elicitation is good both in time efficiency and with accurate solution. The wavelets designed are applied to wavelet-based image coders for fingerprint compression, and the experiment results are presented to validate the advantage of the design method.