Abstract: Based on intrinsic geometric metric,a new method of selecting canonical viewpoints is proposed for observing 3D objects.First,a large quantity of points are sampled on the surface uniformly,and the centroid of 3D object is calculated.Second,a histogram is derived by the distances between sampling points and the centroid.Finally the Shannon entropy is computed and regarded as the measurement of the canonical viewpoint.According to recent research of cognitive psychology,the canonical viewpoint always exists as a fact and is stable.So the viewpoints can be positioned on the enclosed sphere of 3D object.The experimental results show that,by comparison with other methods,more functional structures and distinct features could be observed at the canonical viewpoint of this method,which share more common with the sensory choice of human beings.