Comparison of error concealment techniques for an MPEG-2 video decoder in terrestrial TV-broadcasting1

摘要

In terrestrial broadcasting the received power fluctuation might be in the order of 10 dB, where a non-hierarchical TV-transmission scheme may fail at bad reception conditions. A two layer hierarchical transmission system, also specified by the European digital terrestrial TV-broadcasting standard, may help to have better reception conditions especially for fixed receivers. However, the portable/mobile receiver that only needs the base layer of this hierarchical signal may risk a total loss of picture. Since no time interleaving is foreseen in the standard, the situation is more critical in the case of shadowing, high Doppler (mobile reception), or impulsive noise, whereas even for fixed receivers with two hierarchical levels a good quality of service cannot be guaranteed. Therefore, in order to assume a continuous service with higher quality, some post-processing techniques such as error concealment at the receiver side will be desirable. These techniques, for instance even in the case of hierarchical transmission, may offer better results by concealing the errors instead of using only the most robust stream. The aim of this article is to study different error concealment techniques for the European digital terrestrial TV-broadcasting standard where the source coding is based on MPEG-2. These techniques are based on temporal, spatial and frequency error concealment techniques and on the combination of them. Simulations showed that motion compensated temporal error concealment provides best results for P- and B-pictures, whereas a combination of temporal error concealment and re-synchronization or the use of shorter slices provides best results for I-pictures.