In this paper we describe a new lossless image compression algorithm called
HBB. HBB has been designed to see how much further we can push the current
one-pass DPCM approach to lossless image coding. Its features include a
novel scan ordering and a complex adaptive history-based mechanism which
blends sub-predictors for prediction. Its performance is not sensitive to
the values of its parameters, and it extends easily to different pixel
depths. We show HBB to out-perform current state-of-the-art lossless codecs,
but at a computational cost.