Packet interpolation in two-layer ATM video coding

Transmission of moving images through asynchronous transfer mode (ATM) networks can suffer long delays which may lead to packet loss. The authors have used a method which employs H.261, the CCITT standard video codec, at the base layer and its residual quantization distortions at the second layer. It is assumed that the base layer has a constant bit rate output at a selectable rate up to 1920 kb/s with a step size of 64 kb/s. This makes the base layer coder compatible with the existing circuit switch H.261 codecs. The output bit rate of the second layer is made variable. The base layer data are transmitted with a higher priority than the second one, such that the latter will suffer loss first if the network is congested. The second layer data mostly carry information about picture details or edges, hence their loss can be seen as loss of picture resolution in these areas. The authors introduce a packet interpolation technique to enhance the effect of lost packets from the second layer and is only applicable to the coding techniques where the motion vectors can be safety delivered to the destination