Details of Research Outputs

The additive manufacturing (AM) has met a big challenge on account of customized shapes, especially complex and articulated models, designed by traditional approaches. For realizing AM processing, the reuse, matching, and modifications of the existing 3D digitized models are prerequisite to the applications which also avoid designing from start. In this study, a 3D model matching framework for articulated models serves as a solution for the initial digital design and model processing steps of AM. A discriminative shape descriptor is proposed, which is based on a slice-based model representation and combined with the shape feature distribution in the histogram form. Also a multi-scale histogram approach, integrating an improved Earth Mover's Distance (iEMD), is developed for feature matching which overcomes noise and scale perturbations. The experiments show better performances (average up 7.92 %∼15.80 %) than the classical competitors according to the retrieval performances evaluated by five typical measurements on the McGill database. © 2021 Elsevier B.V.