Abstract :
If two 3d models are similar, they look similar from all angles of view. In this paper, a method for 3d polygon model retrieval based on multiple characteristic views is proposed. First, a model called NCM (New Canonical Model) is obtained by preprocess to an original model, and the range images under 21 different angles of view based on NCM are obtained. Secondly, the features of 3d model called phFT are computed by using Fourier Transform based on the corresponding phase-encoded range image of model, which are composed of a set of histograms and reflect the distribution of the normal orientation and area of facets of 3d modes. Last, Similarity Measurement is calculated by using Image correlation technique. Experiment results show that our method is robust to the translation, rotation and scaling of 3d models, and provides better performance than some competing approaches in 3d models similarity matching.
