Fast image compression using matrix K-L transform

Open Webpage

Daoqiang Zhang, Songcan Chen

Published: 2005, Last Modified: 13 Nov 2024Neurocomputing 2005EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: A novel matrix K–L transform (MatKLT) is proposed as an extension to the conventional Karhunen–Loeve transform (K–L) transform (KLT) in this paper for fast image compression. Experimental results on 18 publicly available benchmark images show that the MatKLT is tens to hundreds times faster than standard KLT with comparable image compression quality.