A cascaded MAP-based linear prediction (CMAP-LP) error concealment technique for consecutive block losses

Linear Prediction has successfully been employed to recover the isolated random losses corrupting images. It has also been used for image coding in some applications. The idea of modelling the image using Markov random fields (MRF) and then replacing any corrupted pixels with the corresponding maximum a posteriori (MAP) estimates has proven to be very successful in the recovery of the isolated corrupted blocks in the image. However, both techniques fail to achieve satisfactory results when used separately to recover the consecutive block losses caused by ATM cell losses. In this paper, we propose a novel error concealment technique that utilises both techniques in the concealment of such consecutive blocks losses