A Hough transform technique for the detection of rotational symmetry under parallel projection

The authors generalize the detection techniques of the Hough transform from analytically defined shapes to rotational symmetry and their parallel projected counterparts. A detection algorithm for rotational symmetry and their parallel projected counterparts is proposed. This algorithm is useful for extracting global and local features of rotational symmetry. It can handle rotational symmetric patterns in the presence of noise and occlusion. This method can be used to determine: (i) the gradient vector (p,q), (ii) the center of rotation O and the order K, and (iii) the relevant image points and the missing data due to noise or occlusion of the rotational symmetry