Memory-Efficient Recursive Interleaved Entropy Coding

Recursive interleaved entropy coding is a promising new adaptable binary entropy coding technique that offers fast encoding and decoding at high compression efficiency [1]. However, with the encoding methods presented in [1], the maximum memory required by the encoder is proportional to the source sequence length. In this article, we present a new encoding technique, and corresponding decoding technique, that limits encoder memory usage without limiting the source sequence length. We illustrate the improvement in compression efficiency over the alternative memory-efficient technique that partitions the source sequence into smaller blocks. We present a generalized encoding framework that encompasses both memoryefficient encoding methods and includes additional variations as well.