For nonstationary time series the fixed Fourier basis is no longer canonical. This article shows how the choice of analysis basis influences the detection of nonstationarities within time series. Rather than limit our basis choice to wavelet or Fourier functions we develop a new stationarity test using a (multiple) bootstrap hypothesis test based on non-decimated wavelet packets. Non-decimated ...

The most widely used speech representation is based on the mel-frequency cepstral coefficients, which incorporates biologically inspired characteristics into artificial recognizers. However, the recognition performance with these features can still be enhanced, specially in adverse conditions. Recent advances have been made with the introduction of wavelet based representations for different ki...

The deconvolution of blurred and noisy satellite images is an ill-posed inverse problem. The direct inversion leads to unacceptable noise amplification. Usually, either the problem is regularized during the inversion process, or the noise is filtered after deconvolution and decomposition in the wavelet transform domain. Herein, we have developed the second solution, by thresholding the coeffici...

INTRODUCTION One of the fundamental signal processing techniques in spindle metrology is the separation of synchronous and asynchronous parts of spindle motion. While in many cases, the synchronous parts of the spindle motion are more important, there are many cases where knowledge of the asynchronous motion is desired. However, it is common that a time series describing the spindle motion is d...

Alternative ways to represent speaker’s voice individuality are studied for the task of speaker verification. We exploit a set of orthonormal bases provided by wavelet packets that allow an effective manipulation of the frequency subbands according to the critical bands concept. Novel wavelet packet based sets of speech features are contrasted with existing wavelet features as well as with the ...

posed both spatially (S) and spectrally (F). The spatial decomposition is carried out using a quad-tree split. Since the spatial partitions do not overlap, this segmentation allows the four children nodes to be further analyzed independently. The frequency decomposition is performed using a four channel quadrature mirror filter (QMF) bank. This produces a library of wavelet packet bases which c...

The partitioning of turbulent perturbations into a ‘‘low-dimensional’’ active part responsible for much of the turbulent energy and fluxes and a ‘‘high-dimensional’’ passive part that contributes little to turbulent energy and transport dynamics is investigated using atmospheric surface-layer (ASL) measurements. It is shown that such a partitioning scheme can be achieved by transforming the ASL...

The perfect reconstruction conditions for a timevarying lapped transform (paraunitary filter bank) are developed through the factorization of the transform matrix into sparse factors. A general formulation is presented allowing one to switch between two paraunitary filter banks. However, the extended lapped transform (ELT) is often used as an example. Furthermore, an adaptive wavelet packet is ...

Many studies have focused on the nonlinear analysis of electroencephalography mainly for the characterization of epileptic brain states. The spatial and temporal dynamics of the epileptogenic process is still not clear completely especially the most challenging aspects of epileptology which is the anticipation of the seizure. Despite all the efforts we still don’t know how and when and why the ...

We use the eigenvalues of a version of the autocovariance matrix to recognize directions at which the Fourier transform of a function is slowly decreasing, which provides us with a technique to detect singularities in images. In very high dimensions, we show how the wavelet packet best-basis algorithm can be used to compute these eigenvalues approximately, at relatively low computational comple...