Parallel adaptive wavelet analysis

Adapted wavelet analysis in the sense of wavelet packet algorithms is a hig hly relevant procedure in different types of applications, like, e.g. data compression, feature extraction, classification problems, data analysis, nu merical mathematics, etc. Given a large or high-dimensional data set the co mputational demand is too high for interactive or "nearly-interactive" proc essing. Therefore, parallel processing is one of the possibilities to accel erate the processing speed. In this case, special attention has to be paid towards handling of the large amount of data in addition to the proper orga nization of the computations. We investigate different data decomposition a pproaches, border handling techniques and programming paradigms. The memory consuming decomposition into a given arbitrary basis after adaptive basis choice is resolved by a localized decomposition strategy. (C) 2001 Elsevier Science B.V. All rights reserved.