Some of the earliest attempts in machine vision utilized
moment invariants to reduce the dimensionality of the machine vision
problem. For vision systems which operate under controlled
environments, moment invariance can provide translation, rotation, and
scale independent feature vectors which can be rapidly classified. A
simple machine vision system is described, based on system of
invariant moments for two-dimensional intensity images developed by
Hu[
3] and Dudani[
2]. Results are presented for a
classifier trained to detect a pair of scissors, a cleaver, and a knife.
