A New Triangle Mesh Simplification Method with Sharp Feature

Abstract

In this paper, a mesh simplification method with Sharp Feature based on a reverse interpolation loop subdivision (RILSP) is proposed. Combined with the treatment of extraordinary vertex in the improved butterfly subdivision, the loop subdivision mask is expanded, thereby improving the traditional loop subdivision algorithm into interpolation subdivision. The reverse operation of the interpolation loop subdivision is used to simplify the complex 3D mesh; a progressive mesh is generated by an initial mesh and a series of vertex offsets. The algorithm reduces the regular point relative reverse butterfly subdivision of compensation operation and greatly reduces simplification and reconstruction. Likewise, the reverse butterfly subdivision algorithm reduces the regular point compensation operation and greatly reduces simplification and reconstruction compared with the existing reverse loop subdivision by considering more control vertices. Furthermore, the edge point with respect to the center point is compensated by sacrificing a small amount of time to calculate the smaller vertex offset, this method gives high transmission speed and low offset. RILSP performs the loss-less and reversible simplification and reconstruction for the vertex compensation in reconstruction process. Compared to Luo's Reverse Butterfly Subdivision Process (RBSP), RILSP cancels the offsets compensation of even vertex in the update process of vertexes, and significantly increased speed of mesh simplification. Compared to Luo's Reverse Loop Subdivision Process (RLSP), RILSP's vertex offset error is obviously less. It faithfully resembles the shape of the original mesh well due to it considers more control vertexes (its mask of regular vertex concludes 14 vertexes). The experiments show that RILSP with sharp feature detection and implementation gets the better mesh quality and low offset compensation compared to above others simplification methods. In addition, in future, it can be applied to the large-scale point cloud model simplification and other fields.