During the last 10 - 15 years there has been increased research activity in error resilient video transmission. In general, all these approaches involve a trade-off between error resilience and coding efficiency. The increased robustness in case of errors can only be achieved when a lower picture quality for the error-free case is accepted. For a variable-error-rate channel, however, this results in the problem of selecting the right amount of error resilience. A worst case design would result in an overly conservative system. Instead, the overhead required for error resilience should only be spent when necessary. This leads to channel-adaptive source coding.Main Results:
1) The concept of distortion-distortion functions (DDFs) is introduced as a formal tool for the evaluation of video quality in the presence of errors. The use of DDFs and their theoretical analysis allows a better understanding of the interactions and trade-offs between system components involved in video transmission.
2) A low-complexity Error Tracking algorithm is introduced that allows to reconstruct spatio-temporal error propagation based on feedback information from the decoder. The result of this algorithm is an estimate of the error distribution in the current frame of the video decoder, which can then be used in the coding control of the encoder to terminate interframe error propagation. This algorithm forms the basis for the practical implementation of feedback-based error control.
3) An analysis of interframe error propagation is presented that describes the additional mean squared error (MSE) that is introduced at the video decoder after transmission errors. The derived model includes the effects of INTRA coding and spatial loop filtering and corresponds to simulation results very accurately. It can also be used to explain the performance gain of feedback-based error control - both in terms of DDFs and overall MSE.