A Similarity Evaluation Method for 3D Models by Using HLAC Mask Patterns

Higher order local autocorrelation (HLAC) has been used as feature descriptors for various pattern recognition applications. We have extended typical 2D HLAC mask patterns to 3D HLAC mask patterns, which enable the masks to handle extraction of shape features from 3D polygonal models. 3D HLAC features are useful for evaluating similarities of 3D models and can be applied to 3D model similarity search engines. We have examined the similarities of 3D HLAC features by applying 3D HLAC masks patterns to each 3D model in a database. An experimental search engine has been implemented, and preliminary tests have shown that the system is able to find similar shaped models efficiently from a database. Key-Words: 3D model, Mask patterns, Similarity retrieval, Search engine