Research has been performed investigating the use of moments for patte
rn recognition in recent years. The basic problem of the influence of
discretization and noise on moment accuracy as object descriptors, has
been barely investigated. In this paper, the detailed error analysis
involved in the moment method is discussed. Several new techniques to
increase the accuracy and efficiency of moment descriptor are proposed
. We utilize later these results for the problem of the image reconstr
uction from the orthogonal Legendre moments computed from discrete and
noisy data is. The automatic selection of an optimal number of moment
s is also discussed.