Rotation angle estimation of an image using least square method

摘要

In this paper, we have developed a novel scheme to detect and estimate the rotation angle between an image and its noisy rotated version using a statistical approach. Radon Transform is applied on to the image intensity to obtain projections on various lines specified by the distance of the line from the origin, p, and angle made by its unit normal with x axis, θ. Further Fourier Series Coefficients of Radon Transform are evaluated with respect to the angle θ for various distances of the line, p. The angle of rotation of the image is estimated from the Fourier Series Coefficients using the Least Square Estimation Technique. The proposed scheme can efficiently detect the traces of Copy-Rotate-Move forgery in an image.