Image segmentation : A survey of methods based on evolutionary computation

Image segmentation is mainly used as a preprocessing step in problems of image processing and computer vision. Its performance has a great influence on subsequent tasks. Evolutionary Computation (EC) techniques have been introduced to the area of image segmentation due to their high search capacity. However, there are rarely comprehensive surveys on EC based image segmentation methods, which can enable researchers to get a quick understanding of this area and compare the existing methods. Therefore, this paper provides an overview of EC based image segmentation methods, and discusses the remaining issues in this area. It is observed that among all EC techniques, four of them (genetic algorithms, genetic programming, differential equation and partial swarm optimization) are more frequently used and GAs are the most popular technique. It is noted that low generalization capacity and computational complexity are two common problems in EC techniques applied to image segmentation.