Optimal Design of Binary Phase-Only Filters Using Genetic Algorithms
Abstract
The genetic algorithm is a mathematical optimization technique, which has generally been applied to one-dimensional problems. In this work, the genetic algorithm was applied to a two-dimensional problem--the construction of binary phase-only spatial filters for optical pattern recognition. Spatial filters that are invariant to range and aspect changes are required for robust pattern recognition. Construction of invariant filters is an optimization problem where the correlation is the objective function for the genetic algorithm. Results are presented for correlation of a genetic algorithm- constructed filter with a multiple aspect angle target set. Filters using a hill-climber algorithm were also constructed and tested.
Authors: Kalyanmoy Deb
Published in: Defense Technical Information Center (DTIC) (1993)