Cross correlation of sampled boundary distances-an approach to object recognition

An approach to object recognition using cross correlation (CC) is presented. The CC utilizes samples of distances defined from a major axis to points located on the boundary of the object image (sampled boundary distances, or SBDs). With this approach (CCSBD), the recognition of an object under translation, rotation, and uniform scaling can be achieved. The recognition of an object under stretching may also be achieved under the condition that the major axis of the image is not changed after transformation. The generalization of this approach is discussed. In particular, one of the generalizations is the use of axis of symmetry for symmetric images to measure SBDs. This can achieve recognition invariant under translation, rotation, uniform scaling, and stretching along the axis of symmetry and/or the direction perpendicular to it. The experimental results of using CCSBD to recognize some sampling objects are presented. The capability of recognizing objects under a certain type of nonaffine transformation using CCSBD is also investigated