Academic Press Library in Signal Processing. http://dx.295 296 CHAPTER 9 Video Error Concealment operation that takes place after the decoding of the current frame and prior to its display, or in the case of whole frame loss, after a missing frame has been detected and prior to decoding the next frame.Encoded video bitstreams, because of the compression methods employed, are generally highly sensitive to channel errors. Due to the widespread use of prediction methods (e.g., for motion compensation, for differential DC transform coefficient coding, and for motion vector coding) and variable length coding (e.g., in entropy coding for motion vectors and residual data), a single uncorrected channel error can propagate to corrupt adjacent spatiotemporal regions of a video signal. The video transmission chain therefore usually includes forward error correction coding in order to protect video packets prior to sending them over a channel to the receiver.When the number of errors exceeds the correction capabilities of the error correction code employed, many communication systems will invoke an auto-repeat request (ARQ) process, whereby the missing packets are retransmitted. ARQ methods are not however applicable in multicast transmission, where clients experience different channel conditions. Even in unicast transmission, the number of ARQs is often limited because of real-time rendering or buffering constraints. When these limits are exceeded, packets are not received correctly and are usually not propagated to the application layer leading to missing blocks of pixels in the decoded frame or whole frame losses. The extent of loss will depend on the packetization applied at the transmitter and the severity of the channel errors [1]. Missing packets can also occur when delays exceed the limits imposed by the delay-sensitive nature of video applications, whereby video packets have to be displayed at certain time intervals or dropped otherwise.To avoid a large drop in received video quality due to intra-and inter-frame propagation of such errors, incorporation of error resilience features at the encoder and the use of error concealment at the decoder are necessary [2][3][4][5]. Error resilience is the process of adding redundancy to the video stream during encoding for the purpose of preventing severe video quality degradation at the decoder in the presence of errors. A typical example is the use of multiple slices within a frame. Slices are independently decodable units that consist of a number of coding units or macroblocks 1 (e.g., a row). They add redundancy to the coded stream in the form of headers but also by setting allowable limits for intra-frame prediction of pixel values and side information. Their use prevents intra-frame error propagation. Without them a missing packet would result in a whole frame loss and create very difficult (if not impossible) conditions for the error concealment module at the decoder. Other such error resilience tools include redundant slices (data repeated at a lower fidelity), data pa...