Adaptive Mesh Simplification Using Vertex Clustering with Topology Preserving

An improved adaptive mesh simplification method based on vertex clustering is presented in this paper. This algorithm can preserve the model topology and geometric features better than traditional vertex-clustering methods. Using the adjacency relationship of all points, the unit normal corresponding to each vertex can be calculated firstly. Then the algorithm adopts the octree structure to subdivide the mesh model adaptively with the guidance of these normal vectors. This subdivision will continue until the angles between the normal vectors in one cell satisfy the predefined threshold. The vertices in this cell are then replaced by a unique vertex, which is a weighted sum of all inside points. Therein, the weight is the normalized cosine value of the angle between the normal and their mean. Experimental results have demonstrated the presented algorithm can not only simplify 3D meshes effectively but also preserve the geometric characters vividly.