RLS algorithm based sub-band decomposition for low memory multimedia applications

In this paper a neural-network based recursive algorithm for computation of the principal component of a stationary vector stochastic process is discussed. The optimal criteria is to retain the maximum information contained of the input, so as to be able to reconstruct the network inputs from the corresponding outputs with minimum mean square error. In this paper a multi-stage principal component extraction (PCE) algorithm is proposed for sub-band decomposition. In multimedia applications though wavelet or sub-band coding of image has been shown to be superior to more traditional transform coding techniques, little attention has been paid to the important issue of whether both the wavelet transforms and the subsequent can be implemented in low memory without significant loss in performance. This paper addresses the problem of low memory image compression and proposes a block based, reduced memory PCE based, image compression algorithm suitable for multimedia online applications