An Automatic Selection Method for the Visual Scene LOD Model and Viewpoint based on Genetic Algorithm and Image Entropy

摘要

The development of 3D real-time visual scenes is usually a time-consuming manual process due to its massive models and viewpoint layout. As a result, an automatic selection method for the visual scene level of detail (LOD) model and viewpoint based on genetic algorithm (GA) and image entropy is proposed in this paper. Image entropy is an objective index for evaluation, which makes automatic selection possible. The method comprises three steps: obtaining the model set and the viewpoint set according to the analysis of scene elements; searching for an optimal combination scheme of the scene through GA; rendering the optimal scene. The experiment result shows that the method proposed above effectively determines the model level and viewpoint, which can be used to the visual scene development.