A comparison of methods for improving the lossless compression of images with sparse histograms

The importance of efficiently compressing images that are characterized by having sparse histograms is been increasing. A good indicator of this fact is the recent introduction of an extended prediction mode in Part 2 of the JPEG-LS standard, with the aim of improving the compression of this class of images. In this paper, we describe and compare several methods that share the same objective: to overcome the lack of compression efficiency that most of the general purpose image compression techniques usually show when handling images that have sparse histograms. This comparison allow us to conclude that a recently proposed online preprocessing technique, albeit quite simple, is also the most effective and versatile.