Transmission of two-tone images over noisy channels with memory

Extended abstract only given. We consider an alternate approach to coding information bearing data for the reliable transmission of two-tone images over noisy communication channels with memory. This consists of jointly designing the source and channel codes (a technique referred to as joint source-channel coding). Source and channel coding are two problems that have traditionally been implemented separately, forming what is known as a tandem source-channel coding system. The separation of channel and source coding is only optimal in an asymptotic sense, i.e., when no constraints exist on the coding block lengths (delay) and on the complexity of the encoder/decoder. Joint source-channel coding, however, has received increased attention. It has previously been shown that if delay and complexity are constrained, performance can be increased if the source and channel codes are jointly designed, as opposed to being treated independently. We propose joint source-channel coding schemes for the reliable transmission of two-tone images over a binary channel with additive Markov noise. Applications of this work are in the transmission of facsimile documents over land mobile radio channels