How would you approach an edge detection problem in a 3D scanning context for Augmentus intelligent 3D scanning?