Mesh simplification is a key research area in scientific visualizationand virtual reality. The paper presents a new method of polyhedral model simplification based on vision characteristic and edge collapse. This method determines model’s vision characteristic by computing the maximum product of angle between normal of vertex and adjacent triangles with area of these triangles. It can achieve a high decimation rate, while keeping the details of vision. This algorithm can also be extended to triangle decimation criterion easily. Keywords Mesh Simplification Vision Characteristic Edge Collapse Triangle Decimation