Abstract

In computer graphics, realistic 3D modeling has been becoming a challenging research area. The terrain synthesis, one of the 3D models, is being developed with two approaches: real geographical data based and fractal terrain. Evolutionary Computation (EC) mimics Darwin's principles of natural evolution processes by representing a chromosome and then applying crossover and mutation. In this thesis, we are interested in how the EC applies to terrain modeling and generates a self-adaptive dynamic terrain system, which can adapt to the personal mannerism of different users. The thesis reviews the background and related works about terrain synthesis and evolutionary computation including related theories and practices. This thesis presents a dynamic adaptive terrain system based on Evolutionary Computation - a specific genetic algorithm. The EC algorithm has been designed and implemented in the EC module; the graphic terrain module shows the result of terrain. The system requires a user's input - mouse click event, drive the interaction between the EC module and the dynamic graphic terrain module. Based on the results that the system has given, the thesis gives out some analysis and conclusions - strengths and weaknesses. The thesis also proposes future works so that researchers picking up this work in future have the benefit of the ideas that thesis generated.