1. Introduction
Transmitting video data over error prone networks can be very unreliable due to packet-loss, and still present a number of challenges to streaming video applications. In a video streaming system, a server pre-stores encoded video streams and transmits them to client terminals for decoding and playback. There are several existing video coding techniques developed to compress video sequences into bitstreams to reduce the data sizes. These video encoding techniques exploit spatial and temporal redundancy to achieve a high compression ratio, while making the compressed data very sensitive to transmission error. This packet-loss problem may lead to serious video quality degradation, which not only affects the quality of current frame, but also leads to error propagation to subsequent frames due to the motion-compensated prediction technique used in standard video codecs. Furthermore, the heterogeneity of client networks also makes the encoder very difficult to adapt the video contents to a wide degree of different client channel conditions, especially for wireless client terminals. In order to achieve error robustness for transmitting video over wireless networks, the server must be able to adapt or transcode the non-error-resilient compressed video streams into error-resilience-capable streams at the intermediate network node. To serve this purpose, a video transcoder [1]–[3] can be placed in a network node connected to a high-loss network to insert error-resilience features into the video bitstream to achieve robust video transmission over wireless channels [4]–[7]. In our previous works [8], a novel two-pass error-resilient transcoding scheme by using prioritized intra-refresh was proposed. Like as adaptive intra refresh (AIR) in the MPEG4 standard, it does not need to make any change for standard video decoders, which is important in terms of cost and convenience for many practical applications.