Multi-resolution mesh based 3D object recognition

We proposed an improved 3D object representation first, which computes the local signatures of a given basic polygon on the surface mesh, and converts the signatures to a 2D array called the distance-angle (DA) images. This representation is adaptive to free-form objects, resistant to occlusion and clutter. Furthermore, it has distinct meaning, and adaptation to different resolution and irregular triangle mesh. Secondly, based on the representation, a novel 3D object recognition algorithm is presented, which is a multi-resolution mesh based, coarse-to-fine recognition. By matching the DA image of a polygon in the scene surface mesh with the DA images of models in low resolution, a model candidate set is obtained. The set is filtered in the neighborhood of the matched polygons in high-resolution mesh and verified by the model candidate sets of other polygons. Experiments show that this algorithm needs less computation, and is very accurate and robust.