A colour reindexing algorithm for lossless compression of digital images

The efficiency of lossless compression algorithms for fixed palette images (also called indexed images) changes if a different indexing scheme is adopted. Indeed, these algorithms adopt a differential-predictive approach of some sort: if the spatial distribution of the indexes over the image is smooth, greater compression ratios may be obtained. It hence becomes relevant to find an indexing scheme that realizes such a smooth distribution. This seems to be a hard problem, and only approximate answers can be provided if a realistic run-time has to be achieved. In this paper, we propose a new indexing scheme, based on an approximate algorithm that maximizes the cost of a Hamiltonian path in a weighted graph. The proposed technique compares favourably with the algorithm proposed by W. Zeng et al. (2000). The computational complexity of the two algorithms is compared and experimental tests that show that relative compression rates are reported.