Shape Retrieval for Khon 3D Model.

We proposed a method for shape retrieval on collection of Khon 3D model. Khon is Thai traditional mask dance. Khon 3D model is important for not only the archival of dance but also for dance technology or digital content. Furthermore, Khon 3D model are complex and unique especially their structure and surface. Moreover, Khon 3D model compose of numerous components and costume with Thai patterns, demonstrating the aesthetic of Thai culture. Therefore our method proposes using Reeb graph to represent structural properties and decomposition into subparts to represent complex Khon 3D model. Moreover, the surface of each subpart is described by a Pose Oblivious Shape Signature. We measure similarity by comparing structure of Khon 3D model based on the Approximate Maximum Common Sub graph which preserves topology. From average precision, we successfully to retrieval on Khon 3D model. The average precision of our method is 0.61 while previous work is 0.51.