Specularity elimination in range sensing for accurate 3D modeling of specular objects

We present a novel range sensing method that is capable of constructing accurate 3D models of specular objects. Our method utilizes a new range imaging concept called multipeak range imaging, which accounts for the effects of mutual reflections. False measurements generated by mutual reflections are then eliminated by applying a series of constraint tests based on local smoothness, global coordinate consistency and visibility consistency. We show the usefulness of our method by applying the method to three real objects with specular surfaces. The ground truth data for those three objects were also acquired in order to evaluate the elimination of false measurements and to justify the selection of the parameters in the constraint tests. Experimental results indicate that our method significantly improves upon the traditional methods for constructing reliable 3D models of specular objects with complex shapes.