MCMC for Joint Noise Reduction and Missing Data Treatment in Degraded Video

Abstract

Image sequence restoration has been steadily gaining importance with the increasing prevalence of visual digital media. Automated treatment of archived video material typically involves dealing with replacement noise in the form of "blotches" that have varying intensity levels and "grain" noise. In the case of replacement noise, the problem is essentially one of missing data that must be detected and then reconstructed based on surrounding spatio-temporal information, whereas the additive noise can be treated as a noise-reduction problem. It is typical to treat these problems as separate issues; however, it is clear that the presence of noise has an effect on the ability to detect missing data and vice versa. This paper therefore introduces a fully Bayesian specification for the problem that allows an algorithm to be designed that acknowledges and exploits the influences from each of the subprocesses, causing the observed degradation. Markov chain Monte Carlo (MCMC) methodology is applied to the joint detection and removal of both replacement and additive noise components. It can be seen that many of the previous processes presented for noise reduction and missing data treatment are special cases of the framework presented here.