The moment invariant method of shape retrieval was first proposed in English literature

by Khotanzad and Hong in 1990. They introduced a set of moment invariants that could be used to represent the shape of an object in a rotation and translation invariant manner. These moment invariants are calculated from the image moments, which are mathematical descriptors of the shape of an object. The moment invariants are then used to compare the shapes of different objects and retrieve similar shapes from a database. This method has since been widely used in pattern recognition and computer vision applications