Cartesian Genetic Programming for Image Processing

Abstract

Combining domain knowledge about both imagingprocessing and machine learning techniques can expandthe abilities of Genetic Programming when used forimage processing. We successfully demonstrate our newapproach on several different problem domains. We showthat the approach is fast, scalable and robust. Inaddition, by virtue of using off-the-shelf imageprocessing libraries we can generate human readableprograms that incorporate sophisticated domainknowledge.