For the shortcomings of slow convergence of mean shift algorithm, based on the study of accelerated mean shift
method, the overrelaxed methods coming from optimization algorithm are extended to generalized overrelaxed methods, and
the method for determining the parameters in the diagonal matrix is given. Then the convergence criteria is amended. For the
above analysis, mean shift algorithm based on generalized overrelaxed methods is proposed. A large number of experiments
on cluster analysis and image segmentation show the effectiveness of the proposed algorithm.