Fast shape discrimination using one-dimensional moments

A simple approach for the fast shape discrimination using one-dimensional moments is presented. For each shape, the authors determine the centroid and calculate the distances from the centroid to the boundary, taking into account the holes in the shape, if any exist. The four one-dimensional moments of these latter distances, which uniquely characterize any two-dimensional shape, are calculated. The proposed approach is simple, involves only simple calculations, and is invariant under rotation, translation, scale, and starting point