Removal of Line Artefacts for

Line scratches are a common form of degradation in archived motion picture lm. The automatic restoration of such material has become of increasing interest in the last few years with the rise of consumer digital video applications and the need to supply more programming material of an acceptable quality in a multi-media context. This paper introduces mechanisms for removing line artefacts from degraded images. It also proposes schemes which solve for both the image data and the hidden model parameters simultaneously. The algorithms introduced are based on the use of the 2D-AR process as an underlying image model and employ a Bayesian approach. The Gibbs sampler is used to draw samples from the underlying image texture and so create a convincing interpolation in the region previously obscured by the line feature. Line scratches are a common problem in archived lm. The problem is transferred to video during the telecine transfer process. The artefact is easily visible as a vertical line of bright or dark intensity, oriented more or less vertically over much of the image. It may be caused when material from some particle is smeared vertically over the lm material in the projector or by the abrasion of the lm as it passes over some particle caught in the mechanism. As digital video broadcasters and those wishing to release material on DVD are faced with the prospect of restoring previously released, but degraded material, it has become more attractive to seek automatic methods for restoration. The EU project AURORA, which terminates at the end of 1998 was set up in 1995 to address the problem of creating real time hardware for achieving good quality automatic restoration. Line scratch removal was one of the problems addressed, although the solution developed will not appear in hardware just yet. The general idea has been to split the line removal process into two stages, an initial detection step followed by an interpolation or reconstruction step. Previous work 1, 3] has introduced deterministic and stochastic methods for the detection of this artefact. This paper concentrates on the interpolation problem. The removal of the defect is treated as a missing data problem. The following material explores stochastic varieties of reconstruction and presents a fast scheme for model based interpolation. In particular the introduced ideas are presented based on the requirement to solve for all the parameters jointly. The algorithms presented here therefore mark a distinct …