The Evolution of Stylized Images using Genetic Algorithms

by Steven Bergen (sb04qv AT brocku . ca)
Supervisor: Brian Ross

Vector graphics images are composed of lists of discrete geometric shapes, such as circles, squares, and lines. Vector graphics is popular in illustration and graphic design. The generation of vector images by evo- lutionary computation techniques, however, has been given little attention. This research uses genetic algorithms to evolve vector images. By restricting the numbers of primitives and colour schemes used, stylized inter- pretations of target images are produced. Automatic evolution involves measuring the pixel-by-pixel colour distance between a candidate and target image. The JNetic evolutionary vector graphics system is described. JNetic supports automatic and user-guided evolution, chromosome editing, and high-detail masks. The user can paint masks over areas of the target image, which will be used to reproduce the high-detail features within those areas. Example images from the system are viewable below.



Generated images copyright (c) 2009 Steve Bergen and Brian J. Ross.
JNetic system copyright (c) 2009 Steve Bergen.

Back up: http://www.cosc.brocku.ca/~bross/