Modeling of Virtual Pagoda Based on Neural Network

This paper presents an efficient method to modeling and simulation system. A hybrid of graph-based and neural network modeling system is developed. To implement the pagoda modeling system, we have built a component model library of pagoda by L system and a corresponding feature library. Feature points have been detected by improved feature detection algorithm. After the pagoda nodes graph (PNG) is transformed into representation vectors, and these vectors are presented to the neural network which classifies them into feature classes. A new BP algorithm based on the enlarging error is presented. Once the feature information is detected, it is easy to calculate the ratios between features. XML parser and verifier ensure the coherent architecture styles of ancient pagodas. Because of the advantages (accurate modeling and rapid generating) of this modeling system, it can be applied to the preserving of the digital heritage in China ancient architecture.