Abstract: A vision model of user avatar with nine freedoms based-on simple convergent binoc-ular stereo vision is proposed and used to imaging and transformation in virtual scene. The formulas of imaging in left and right virtual camera about the virtual scene points are educed with its world coordinate values and the nine freedoms variables of the user avatar vision model. The equations are deduced about the relationship between the parallax of the imaging points in left and right images about the scene point projected by virtual binocular camera and depth in-formation of the scene point relative to the original points of the observation coordinate system. The results of the simulation experiment prove the feasibility of the model and the reliability of the algorithmic method in this paper. Simultaneously, it is proved that the vision model of user avatar with nine freedoms is more effective, more practical and more vicinal to human beings visional habit.