A Multiresolution Markov Approach to Model-Based Image Compression

✍ Scribed by Robert J. Bonneau; Henry E. Meadows

• Publisher: Elsevier Science
• Year: 2001
• Tongue: English
• Weight: 266 KB
• Volume: 11
• Category: Article
• ISSN: 1051-2004
• DOI: 10.1006/dspr.2001.0495

✦ Synopsis

Currently a large area of research is being devoted to content-based compression due to the JPEG-2000 and MPEG-7 requirements. Recently there has been much work in wavelet and fractal methods for texture and shape segmentation as well as data compression. While these methods do not give optimal least mean square noise performance for a given compression ratio, they contain implicit models for shape and texture coding as a natural part of the compression process. We thus develop an approach for wavelet fractal compression that incorporates these shape and texture models during quantization. Upon decoding, the model regions are preserved for visual or automatic inspection. Our compression models make use of the Mallat Gaussian derivative basis set and an implicit Markov shape and texture structure.