A comparison between representations for evolving images

Abstract

Evolving images using genetic programming is a complex task and the representation of the solutions has an important impact on the performance of the system. In this paper, we present two novel representations for evolving images with genetic programming. Both these representations are based on the idea of recursively partitioning the space of an image. This idea distinguishes these representations from the ones that are currently most used in the literature. The first representation that we introduce partitions the space using rectangles, while the second one partitions using triangles. These two representations are compared to one of the most well known and frequently used expression-based representations, on five different test cases. The presented results clearly indicate the appropriateness of the proposed representations for evolving images. Also, we give experimental evidence of the fact that the proposed representations have a higher locality compared to the compared expression-based representation.