Coarse Visual Registration from Closed-Contour Neighborhood Descriptor

This article introduces an innovative visual coarse-registration process suitable for textureless objects. Because our framework is industrial, the process is designed for metallic, complex objects containing multiple bores and repetitive patterns. This technique is based on a local shape descriptor, invariant under affine transform, which characterizes the neighborhood of a closed contour. The affine invariance is exploited in the learning stage to produce a lightweight model: for an automobile cylinder head, a learning view-sphere with twelve viewpoints is sufficient. Moreover, during the learning stage, this descriptor is combined to a 2D/3D pattern, concept likewise presented in this article. Once associated, the 2D/3D information wealth of this descriptor allows a pose estimation from a single match between two descriptors. This ability is exploited to obtain efficiently a great number of coarse pose hypothesis. A pose hypothesis classification method is proposed to select the best-ones. An evaluation on a cylinder head and a binding beam confirms both the robustness and the precision of the process