An improved quadric error metrics based on feature matrix

Abstract

The research on the base of Quadric Error Metrics (QEM) has provided some excellent eclectic methods of model approximation and time cost for mesh simplification. However, there are still some deficiencies in these methods to be settled, such as ignoring some important features and excessive simplification in some parts of the model, etc. To preserve the important geometric features, the paper has presented an improved method based on Garland¿s QEM by integrating a feature matrix into the approximating error with quadrics of the vertex. The new matrix based on the geometric importance and the area attribution of a vertex are considered together and used for changing the order of edge collapse in the simplification. It is shown by the experimental results that the proposed algorithm can not only be highly efficient in terms of both space and time cost, but also get highly quality approximation of the original model with the important features well preserved.