Concealment for video communications over error-prone channels

Transmission of compressed video over error-prone channels requires techniques which are robust against errors and yet efficient in compressing the video. In this work, we study the use of concealment techniques for robust coding of video. We find that for sequences with irregular motion and changing scenes, a concealment technique which exploits both spatial information from the current frame and temporal information from the previous frame performs better than techniques which exploit only one of the sources of information