Lossless Image Compression Using Agglomerative Hierarchical Clustering and the Jensen-Shannon Divergence
Sameh Samir, Sherif F. Fahmy, Gamal Ibrahim Selim · 2015
In this paper, an approach to lossless image compression using minimum entropy clustering is shown. In such a clustering, blocks are grouped such that the overall entropy of the clusters is minimized when compared to the single cluster case. The approach in this paper utilizes an agglomerative hierarchical clustering mechanism with the Jensen-Shannon divergence as a similarity measure. Experimental results show that this approach outperforms both k-medoids minimum entropy clustering and the single cluster cases in compression ratios.