Animated mesh simplification based on motion features in visual sensor networks

Abstract

The three-dimensional animated model is widely used in scientific visualization, entertainment, and virtual applications, especially in visual sensor networks. The main purpose of simplification is to capture the shape sequence of an object with very few elements while preserving the overall shape. As three-dimensional animated mesh is time-varying in all frames, the trade-off between the temporal coherence and geometric distortion must be considered to develop simplification algorithm. In this article, a novel three-dimensional animated mesh simplification algorithm based on motion features is presented. Here, motion features are the connection areas of the relative movement consisted of vertices and edges. Motion feature extraction is to find a subgraph that has movement property. Dihedral angle of the edge through all frames is used to determine whether an edge is connected or not to the movement parts. Then, a rotation connected graph is defined to extract motion features. Traveling this graph, all motion features can be extracted. Based on the motion features, animated quadric error metric is created and quadric error matrix is built through all frames. Compared with the other methods, the important advantages of this method are high-efficiency simplification process and smoother simplification effects. It is suitable to be used in real-time applications. Experiment results show that the 3D animated mesh simplification effects by our method are satisfactory.