Video dejittering by bake and shake

摘要

Video jittering occurs when the horizontal lines of video image frames are randomly displaced due to the corruption of synchronization signals or electromagnetic interference during video transmission. Inspired by the recent Bayesian/variational dejittering model of Shen (SIAM J. Appl. Math., vol. 64, pp. 1691-1708, 2004), in the current paper we propose a novel dejittering approach nicknamed 'bake-and-shake.' The bake step is to apply Perona-Malik type nonlinear diffusions to 'melt away' or heat up the jittered video frames, based upon which the shake step is able to optimally estimate the individual line jitters and renormalize the jittered images. Numerical implementation of the bake-and-shake algorithm as well as several computational results are presented.