Video Error Concealment Using Spatio-Temporal Boundary Matching

Transmission of videos in error prone environments may lead to video corruption or loss. Therefore error concealment at the decoder side has to be applied. Commonly error concealment techniques make use of the surrounding correctly received image data or motion information for concealment. In this paper, a novel spatio-temporal boundary matching algorithm (STBMA) by exploiting both spatial and temporal information to reconstruct the lost motion vectors (MV) is proposed, and also introduce a new side smoothness measurement. By using the motion vector that is found by the proposed algorithm, the lost macro block (MB) can be recovered. Compared with the well known Boundary Matching Algorithm (BMA), the proposed algorithm is able to achieve higher PSNR as well as better visual quality.