USING PHOTOMETRIC INVARIANTS FOR 3D OBJECT RECOGNITION

In this paper we describe a new efficient algorithm for recognizing 3D objects by combining photometric and geometric invariants. We derive some photometric properties that are invariant to the changes of illum ination and to relative object motion with respect to the camera and/o r the lighting source in 3D space. We show that recognition does not r equire a full constancy of colors; rather, it only needs something tha t remains unchanged under the varying light conditions and poses of th e objects. Combining the derived color invariants and spatial constrai nts on the object surfaces, we identify corresponding positions in the model and the data space coordinates, using centroid invariance of co rresponding groups of feature positions. Tests are given to show the s tability and efficiency of our approach to 3D object recognition. (C) 1998 Academic Press.