Image clustering using Genetic Algorithms

Hi all; I am trying to cluster images into a known number of segments using a genetic algorithm. I am looking for the best way to represent my data. So far I''ve thought of two ways: 1- Each Chromosome is made up of all the pixels in the image arranged as a MXN matrix of cluster values. This representation has the problem that I cannot maintain cluster heterogeniety across individual solutions (i.e. cluster X might represent a flower in solution A and a rabbit in solution B. crossover will likely destroy rather than enhance this clustering) 2- A chromosome is arranged by clusters, each cluster referring to the individual pixels that form it. This representation has the problem that a normal crossover operation is likely to duplicate pixels in different clusters and erase pixels in others. Any suggestions? ( K-means is out of the question. It has to be genetic) Peace Out!