Evaluating Mutation Operators for Evolved Image Reconstruction Transforms

Various military systems require image and signal processing, often in noisy or bandwidth-limited situations. In this research, we employ genetic algorithms (GAs) to evolve forward and inverse transforms that reduce quantization error in reconstructed signals and images. The resulting transforms produce higher quality images than current waveletbased transforms at a given compression ratio and thus allow transmission of compressed data at a lower bandwidth. We expand on previous research by evaluating several mutation strategies for evolving reconstruction filters. Our results indicate that GAs employing Gaussian mutation applied with shrinking standard deviations evolve transforms superior to transforms evolved by GAs employing other tested mutation operators.