Skip to Main Content
This paper proposes the feature descriptor for 3D model similarity search using the distribution of normal directions on the simplified surface. Feature descriptor of 3D model should be invariant to translation, rotation and scale for its model. So this paper normalizes all the model using PCA and preprocesses surface mesh simplification to robust against noise. The normal is sampled in proportion to each polygon's area and then it is calculated by weight average method via angles and interpolated. We implemented the 3D model retrieval system and performed the similarity search test with the shape bench mark data provided by the Princeton University. Experimental results show the performance improvement of proposed algorithm from 24.7% to 32.2% in comparison with conventional methods by ANMRR.