Optimizing 3D chamfer masks with norm constraints
Résumé
Chamfer distances are widely used in image analysis. One of their major interest is to approximate the Euclidean distance with integers. Optimizing approximations, in the 3D case, is done in the literature but without worrying if the computed masks actually induce a norm. In that paper, we propose a construction of chamfer masks in 3D, based on Farey triangulations, which gives constraints on the weightings; by scanning the whole space of solutions, we compute for each mask, an exhaustive list of optimal weightings.