Enhanced error concealment with mode selection

Delay sensitive video transmission over error prone networks can suffer from packet erasures when channel conditions are not favorable. Use of error concealment (EC) at the video decoder is necessary in such cases to prevent error induced artefacts making the affected video frames visibly intolerable. This paper proposes an EC method that incorporates enhanced temporal and spatial concealment elements, the use of which is controlled by a mode selection (MS) algorithm well matched to the characteristics of the temporal concealment approach. The performance of the individual enhancements and of the MS algorithm are compared with the respective features of the method employed in the H.264 joint model (JM) decoder and with other state of the art methods. The overall performance of the proposed method is shown to offer significant gains (up to 9 dB) compared to that of the JM decoder for a wide range of natural and animation image sequences without any considerable increase in complexity.