A new approach to array denoising

Abstract

We present a novel approach for suppressing additive noise in multichannel signal processing environments. Inspired by a neurophysiological data environment, where an array of recording electrodes is surrounded by multiple neural cell sources, significant spatial correlation of source signals motivated the need for an efficient technique for reliable multichannel signal estimation. The technique described is based on thresholding an array discrete wavelet transform (ADWT) representation of the multichannel data. We show that by spatially decorrelating the ADWT, spatially correlated noise components in each frequency subband spanned by the corresponding wavelet orthonormal bases are converted to uncorrelated ones that are eventually suppressed by the thresholding stage. Recorrelation and reconstruction of the resulting ADWT is then performed yielding a significant improvement in SNR on all channels. The advantage of this technique lies in the fact that no apriori assumptions are required about the signal parameters or the noise process. Results of applying the technique to simulated and real multiunit neural recordings are presented and compared to existing techniques.