New Set of Rotationally Legendre Moment Invariants
Khalid Mohamed Hosny · 2010
Abstract—Orthogonal Legendre moments are used in several pattern recognition and image processing applications. Translation and scale Legendre moment invariants are achieved directly by using Legendre polynomials. Unfortunately, this method can not convert into a rotational invariant form. This is a big disadvantage and limits the use of Legendre moment invariants in many vital applications. In this paper, an indirect method is used in computing invariant Legendre moments as a linear combination of translation, scaling and rotation geometric moment invariants. Geometric moment invariants are computed exactly for symmetric as well as non-symmetric binary and gray-scale images. According to the tremendous reduction of the computational complexity, the proposed set of Legendre descriptors is suitable for large size images. The performance of these descriptors is evaluated by using a set of standard images. Keywords—Legendre moment invariants, geometric moments, symmetric images. I.