Abstract
Many structures in the nature are formed by local nonlinear interactions among components that have a simple structure. Such a process of forming structure is called "emergence" in the research field of artificial life. The advantage of emergent structure formation is that a complex structure beyond our imagination can be obtained from a simple structure. Meanwhile, its disadvantage is that we cannot predict what structure is eventually obtained easily. In this study we propose a method for emergent formation of structures that enables us to form a variety of novel structures similar to a desired structure somewhat. In addition, we concretely realize a method for forming images according to the idea of the proposed method, which generates a variety of novel images that are somewhat similar to a given image by applying a genetic clustering method to a portion of pixel data of the given image.
