Automatic rule generation for procedural modeling of a sketched tree using genetic programming and particle swarm optimization

Advantages: • Memory efficiency. • Dynamism (i.e. growth process). • Scalability (i.e. growth level). • Portability. • Generating identical trees while preserving the general aspects. Procedural Tree Modeling Proposed Approach Tree modeling is the process of generating realistic 3D models of trees within virtual environments. These models are widely employed in computer animation, game design, and botanology. To address this demand, several approaches such as image-based modeling and procedural modeling have been widely explored. L-system is the most exploited procedural modeling paradigm in this area. It employs a few rules to model the growth of a tree. It has been shown that using high order L-systems leads to realistic 3D tree models. In many cases, only a sketch of a tree is drawn and the rules are not available. Defining a set of rules to reproduce the drawn tree is a tedious task. In this research, a combination of genetic programming and particle swarm optimization is proposed to automatically generate a set of rules from a sketch to model the corresponding tree.