Artificial Life And Pattern Formation

In this thesis two fields related to artificial life are studied, one is pattern formation, another is vision model of virtual creatures.The endless array of patterns in nature has long been a source of joy and wonder to people. Some patterns can be seen, like beautiful shapes of snowflakes, spiral waves in the B-Z reaction, morphologies throughout the biological world, others can not, like perception, cognition, imagination. Underlying these disparate patterns is there a set of overarching principles that can lend a unified perspective to this field of study?As a step towards such a perspective, Pattern formation in generalized cellular automata(GCA) is studied in this paper, different dynamic behaviors are classified, a definition of edge of chaos is proposed, and experiments are described where self-making patterns emerge which reminiscent of simple living systems. The influence of remembrance to dynamic behavior of GCA is also studied.Intelligent virtual environment is a test-bed for theories in artificial life, cognitive creatures arises from the evolutionary computation embedded in the artificial life simulation, they are intelligent agents that perceives the virtual environment through a simulated vision system. In this thesis a vision model consist of Cellular neural network as feature detector and BP network as classifier is proposed, and experiments are described where genetic algorithm is used to evolve an optimal feature detector in order to aid a classifier to generalize across different conditions. Application of Cellular neural network to image filtering is also studied.