Two-level context based compression of binary images

Abstract

Variable order models coupled to an arithmetic coder have proved a successful paradigm for text compression. Here we explore the usefulness of a similar scheme for binary images. An algorithm due to Langdon and Rissanen is used as test bed for an experimental investigation, and with a two-level scheme based upon a conditioning context of 18 pixels a compression gain of about 20% can be achieved. Other experiments show that, for the test data used, this compression is at most about 30% inefficient.