Rotational invariants for Tchebichef moments

Abstract

Image moments that are invariants to distortions such as translation, scale and rotation are an important tool in pattern recognition. In this paper, derivation of invariants for Tchebichef moments with respect to rotation will be presented. The rotational invariants are achieved neither by tempering with the image nor transforming the coordinates from rectangular to polar. They are derived using moment normalization method, which attempts to map the distorted moments with the undistorted ones. Experimental results show that the derivation is correct and it poses as a viable solution to test whether one image is a rotationally distorted version of another.