Local Feature Entropy Based Non-Uniform Simplification Algorithm

摘要

Aiming to solve the issue that the accuracy and efficiency after the simplification of 3D model is difficult to be balanced, a new method of simplified algorithm based on half-edge collapse nonhomogeneous mesh method of local characteristic entropy is proposed. Detect clustering local area twice. Firstly, detect edge clustering local area where there is 3D data point to obtain normal vector in the area; secondly, detect the normal vector of secondary regional clustering area by the constraints of the center of gravity of the region near 3D data points. According to the definition of information entropy, take the local area characteristic entropy constructed by angle information between the two normal vectors from the two detection method as the half edge collapse cost. The bigger the local area characteristic entropy is, the flatter the region tends to be, and the priority of simplification shall be given to this, otherwise it shall be retented. Lastly, retain the triangle regularity in the simplified mesh judged by the interior angles to reduce the deformation caused by the error. The experimental results show that the algorithm can achieve a better balance in the accuracy and time efficiency of the local details.