Abstract
JPEG2000 is a new still image compression standard. It has better compression performance than conventional JPEG standard, and it provides many useful features. The hardware implementation of JPEG2000, therefore, becomes essential technique of digital still camera. In this thesis, we have implemented a complete JPEG2000 encoder and performed power minimization on this design through an improved memory decomposition tool we developed. This tool is a refinement over previous methods in a number of aspects. For example, we build more accurate power models for the building blocks supporting memory decomposition methodology, including the external bank decoder, output multiplexor, address driver, and each memory sub-bank. With such accurate power models, we can then embark on a realistic search to decide the optimal bank number for decomposing an arbitrary given large memory. The experiments results show that about 30% power reduction can be achieved when utilizing this memory decomposition tool for our JPEG2000 encoder.