Modeling of 2D Parts Applied to Database Query
Guillaume-Alexandre Bilodeau, Robert Bergevin · PolyPublie (École Polytechnique de Montréal) · 2000
This paper presents the latest developments in a project aimed at the design of an image database query engine, where the images are searched at the 3D object-level. This is a novelty since the majority of existing image database query engines search images by comparing the colors, the textures and the 2D shape of regions in the images. This paper specifically discusses a new method to hypothesize volumetric primitives from 2D parts (2D regions corresponding to projections of volumetric primitives). Our new hybrid approach combines two existing approaches to benefit from the advantages of both. It combines a model-fitting approach and a rule-based approach. Using fuzzy logic, this new approach can produce multiple hypotheses to attain the robustness necessary for processing 2D parts originating from real 2D images. A detailed description of the approach is presented along with preliminary results.