Researches on mold-based feature extraction and retrieval technology of 3D models

As the key problem of content-based three-dimensional model retrieval, feature extraction methods do not achieve ideal performance at present. Most of past feature extraction methods just utilize the attributes of the model itself. Therefore, it is a promising direction to utilize the categorization information in the process of feature extraction. According to the fact that models in the same category have many similarities and little differences, we represent the contents of models in one category by elastic mold. We also introduce the feature extraction and the selecting method of the optimal mold. Finally, we state an effective retrieval mechanism based on mold. The experimental result shows that our method performs better than the traditional ones.