Data algorithm: a numerical method to extract shape information from gray scale images
Renzo Davoli, Fabio Tamburini · 1994
This paper presents a numerical algorithm which realizes a translation, rotation and scale invariant transform for single object gray scale images. This kind of algorithm is applicable to all those applications which need to deal with shape information, independent of the orientation, distance and position of an object. The algorithm we present takes its mathematical foundation from the log-polar transform (Casasent-Psaltis 76). These analytical results cannot be mapped directly into an algorithm: they involve generalized integrals that are hard to compute and the numerical implementation has to take account of discretization error which may propagate and lead to meaningless results. The aim of this paper is to present an algorithm which overcomes these problems.