Hoang, Nguyen VuGouet-Brunet, ValérieBenjamin Bustos and Hedi Tabia and Jean-Philippe Vandeborre and Remco Veltkamp2014-12-152014-12-152014978-3-905674-58-31997-0463https://doi.org/10.2312/3dor.20141048https://diglib.eg.org/handle/10.2312/3dor.20141048.037-044This paper presents an approach for 3D object retrieval, dedicated to partial shape retrieval in large datasets. A word manipulation, i.e. quantized descriptors, as in the Bag-of-Words representation is employed, based on the extraction of 3D Harris points and on a local description involving local manifold harmonic transform. By adding Δ-TSR, a triangular spatial information between words, the richness and robustness of this representation is reinforced. The approach is invariant to different geometrical transformations of 3D shape such as translation, rotation, scale and robust to shape resolution changes. We have evaluated it in terms of quality of retrieval, facing several state-of-the-art methods and on different public 3D benchmarks involving different contents and degrees of complexity.Modeling [Computer Graphics]3D Shape MatchingRepresentationsdata structurestransforms