Energy-based reconstruction of 3D curves for quality control

摘要: In the area of quality control by vision, reconstruction 3D curves is a convenient tool to detect and quantify possible anomalies. Whereas other methods exist that allow us describe surface elements, contour approach will prove be useful reconstruct object close discontinuities, such as holes or edges. We present an algorithm for parametric curves, based on fixed complexity model, embedded in iterative framework point insertion. The successive increase degrees freedom provides good precision while avoiding over-parameterize model. curve reconstructed adapting projections NURBS snake observed multi-view setting. optimization performed with respect points using gradient-based energy minimization method, whereas insertion procedure relies computation distance from image edges.