A model simplification algorithm based on visual features preservation

Aiming at the problem that the traditional mesh simplification algorithms can not preserve the geometric features that is sensitive to the human visual system, a model simplification algorithm is put forward based on visual feature preservation. Firstly, the degree of bend (DOB) and the curvature of the vertex to describe the geometric features that are more sensitive to human visual system are the imported; Secondly, they are used to improve the traditional quadratic error metric (QEM) algorithm, and then a simplified algorithm that can well preserve visual features is proposed,. In this algorithm, the model surfaces are classified into the visual sensitive areas and non-sensitive areas according to the given threshold value ground on computing the DOB and the curvature of the vertex at first, then the non-sensitive areas are simplified by using the improved QEM while the sensitive areas, are just preserved as they are. Hence the visual features on the model surfaces that are the sensitive to human visual system are well retained. Experiments show that the new model simplification algorithm can maintain the high curvature areas and highlight the visual features of the simplified model.