Abstract: How to reduce the time and increase the efficiency of retrieval are the key problems in 3D model retrieval.Osada's D3 shape distribution doed not describe the content of model enough and the calculated amount is comparatively large. A novel method for 3D models retrieval was proposed. Computing the area between the center of mass of model and two barycenters of random triangles, using the statistical data to compute area distribution histograms. Then the histogram is used to match the model. Experimental results show the proposed method received a better performance.

Key words: 3D models retrieval,Feature extracting,Shape distribution,Area distribution histogram