Method for Automatic Surface Inspection using Models-Based 3D Descriptor. Madrigal, C., Branch, J., Restrepo, A., & Mery, D. Sensors, 17(10):2262, 2017.
Method for Automatic Surface Inspection using Models-Based 3D Descriptor [link]Paper  doi  abstract   bibtex   
Automatic visual inspection allows identifying surface defects in manufactured parts. Nevertheless, when defects are on a sub-millimeter scale, its detecting and recognizing is a challenge. In particular, if the defect generates topological deformations that are not shown as a strong contrast in the 2D image. In this paper, we presented a method to recognize surface defects on 3d point clouds. First, we propose a novel 3d local descriptor called MPFH (Model Point Feature Histogram) for defect detection. Our descriptor is inspired from earlier one such as PFH (Point Feature Histogram). To construct the MPFH descriptor, the models that best fit the local surface and their normal vectors are estimated. For each surface model, its contribution weight to the formation of the surface region is calculated and from the relative difference between models of the same region a histogram is generated representing the underlying surface changes. Second, through a classification stage, the points on the surface are labeled in 5 types of primitives and the defect is detected. Third, the connected components of primitives are projected to a plane, forming a 2D image. Finally, 2D geometrical features are extracted and by a support vector machine, the defects are recognized. The database used is composed of 3D simulated surfaces, 3D reconstructions of defects in welding, artificial teeth, indentations in materials, ceramics and 3D models of defects. The quantitative and qualitative results showed that the proposed method of description is robust to noise and to the scale factor and sufficiently discriminative to detect some surface defects. The performance evaluation of the proposed method was performed for a classification task of 3D point cloud in primitives, reporting an accuracy of 95% and higher than other state-of-art descriptors. The rate of recognition of defects was close to 94%.

Downloads: 0