Abstract
Region-based coding schemes for video sequences have recently received much attention owing to their potential to enable several interesting multimedia applications. In this paper, we focus on the optimisation of the coding of uncovered regions. We find that the widely-used tools for coding the texture within arbitrarily shaped regions of the intra image are not appropriate for failure regions. We propose a new method for doing so which consists in applying a predictive coding step to the newly visible portions based on already transmitted spatial data.