Hamming Graph based Image Compression with variable Threshold

Abstract

In this paper a lossy image compression methodusing the adjacency matrix of a graph is proposed.We instantiate the method using the Hamming graph,which gives rise to an image compression algorithmthat uses a Hadamard Matrix together with a thresholdapproach. In threshold methods, selection of thecorrect (static) threshold value is difficult. In our approachwe use a new threshold determination thatchanges from block to block obtaining an algorithmthat uses only integer and modular operations andobtains good results in some standard images. Thenew method has better stability under recompressioniteration than jpg.