A stack algorithm for source coding with a fidelity criterion

Abstract

Although coding of sources with respect to a fidelity criterion was first proposed by Shannon in 1959, only recently have effective schemes been developed. Another such algorithm based on a stack is described. Like the author's other algorithms (Jelinek and Anderson, 1971, 1973), this algorithm is conceived with tree codes in mind. It has various strengths and weaknesses relative to the earlier two, which are explored. By itself, the algorithm has an interesting mathematical analysis that yields insight into the faster-than-exponential behavior of tree variables and provides a new characterization of the rate-distortion function.