High performance computing for spatial outliers detection using parallel wavelet transform
Intelligent Data Analysis
Hi-index | 0.00 |
Abstract: Wavelet transform has become a popular method extensively used nowadays. One important tool when working with wavelets is the Fast Wavelet Transform (FWT). However, in the last few years a new method for constructing wavelets has achieved great success: the lifting scheme. This paper describes a message-passing based parallel strategy, suitable for both algorithms (FWT and lifting scheme), in which high efficiency is achieved by means of a modified data-swapping approach allowing communications to overlap computations. The method is illustrated with its application to the well known Daubechies (D4) wavelet. Timing and speed-up results for the Cray T3E are presented.