A new algorithm for lossless still image compression

Abstract

This paper presents a spatial domain method for lossless still image compression using a new scheme: base switching (BS). The given image is partitioned into non-overlapping fixed-size subimages. Different subimages then get different compression ratios according to the base values of the subimages. In order to increase the compression ratio, a hierarchical technique is also used. It is found that the compression ratio of the proposed algorithm can compete with that of the VBSS and the international standard algorithms known as JBIG and Lossless JPEG. In addition, when the BS method is compared with the S + P method, which is an excellent frequency domain method that used EZW, although S + P method gains about 9% increase in the compression ratio, its encoding time (excluding I/O) is about three times longer than ours. The math theory needed to build up the proposed compression scheme is also provided. (C) 1998 Published by Elsevier Science Ltd on behalf of the Pattern Recognition Society. All rights reserved.