Showing papers in "IEEE Signal Processing Letters in 2001"
TL;DR: A novel adaptive operator is devises, which forms estimates based on the differences between the current pixel and the outputs of center-weighted median (CWM) filters with varied center weights, which consistently works well in suppressing both types of impulses with different noise ratios.
Abstract: Previous median-based impulse detection strategies tend to work well for fixed-valued impulses but poorly for random-valued impulse noise, or vice versa. This letter devises a novel adaptive operator, which forms estimates based on the differences between the current pixel and the outputs of center-weighted median (CWM) filters with varied center weights. Extensive simulations show that the proposed scheme consistently works well in suppressing both types of impulses with different noise ratios.
741 citations
TL;DR: This paper gives an alternative derivation and explanation for the result by Kingsbury (1999), that the dual-tree DWT is (nearly) shift-invariant when the scaling filters satisfy the same offset.
Abstract: This paper considers the design of pairs of wavelet bases where the wavelets form a Hilbert transform pair. The derivation is based on the limit functions defined by the infinite product formula. It is found that the scaling filters should be offset from one another by a half sample. This gives an alternative derivation and explanation for the result by Kingsbury (1999), that the dual-tree DWT is (nearly) shift-invariant when the scaling filters satisfy the same offset.
356 citations
TL;DR: This work provides a textbook-like direct derivation of the stochastic CRB, indirectly derived as the (asymptotic) covariance matrix of the maximum likelihood (ML) estimator.
Abstract: The stochastic Cramer-Rao bound (CRB) for direction estimation in array processing applications was indirectly derived some ten years ago as the (asymptotic) covariance matrix of the maximum likelihood (ML) estimator. Attempts to obtain the stochastic CRB directly via the CRB theory fell short of providing a simple derivation and consequently, no direct derivation of this useful performance bound was available in the open literature. we correct this situation by providing a textbook-like direct derivation of the stochastic CRB.
334 citations
TL;DR: In this paper, the design of a fractional order FIR differentiator is investigated and one example is used to demonstrate that the fractional derivatives of digital signals are easily computed by using the proposed filtering technique.
Abstract: In this paper, the design of a fractional order FIR differentiator is investigated. First, the fractional derivative of the power function is defined. Then, the impulse response of the fractional order differentiator is obtained by solving linear equations of the Vandermonde form. Finally, one example is used to demonstrate that the fractional derivatives of digital signals are easily computed by using the proposed filtering technique.
197 citations
TL;DR: A new speech enhancement method based on the time adaption of wavelet thresholds that does not require an explicit estimation of the noise level or of the a priori knowledge of the SNR, which is usually needed in most of the popular enhancement methods.
Abstract: We propose a new speech enhancement method based on the time adaption of wavelet thresholds The time dependence is introduced by approximating the Teager energy of the wavelets coefficients This technique does not require an explicit estimation of the noise level or of the a priori knowledge of the SNR, which is usually needed in most of the popular enhancement methods Performance of the proposed method is evaluated on speech recorded in real conditions and with artificial noise
195 citations
TL;DR: The SM-AP algorithm generalizes the idea of the set-membership NLMS (SM-NLMS) algorithm to include constraint sets constructed from the past input and desired signal pairs, and can be seen as a set- membership version of the affine-projection (AP) algorithm with an optimized step size.
Abstract: This letter presents a new data selective adaptive filtering algorithm, the set-membership affine projection (SM-AP) algorithm. The algorithm generalizes the idea of the set-membership NLMS (SM-NLMS) algorithm to include constraint sets constructed from the past input and desired signal pairs. The resulting algorithm can be seen as a set-membership version of the affine-projection (AP) algorithm with an optimized step size. Also, the SM-AP algorithm does not trade convergence speed with misadjustment and computational complexity as most adaptive filtering algorithms. Simulations show the good performance of the algorithm, especially for colored input signals, in terms of convergence, final misadjustment, and reduced computational complexity.
187 citations
TL;DR: Wavelet packet transform's multiresolution capabilities are used to derive new sets of features, which are found to be superior to Mel frequency cepstral coefficients (MFCC) in unvoiced phoneme classification problems.
Abstract: A new filter structure using admissible wavelet packet analysis is presented. These filters have the advantage of having frequency bands spacing similar to the Mel scale. Further wavelet packet transform's multiresolution capabilities are used to derive new sets of features, which are found to be superior to Mel frequency cepstral coefficients (MFCC) in unvoiced phoneme classification problems.
170 citations
TL;DR: An efficient carrier frequency offset (CFO) estimation algorithm for the orthogonal frequency-division multiplexing (OFDM)-based wireless local area networks (WLANs) is presented.
Abstract: We present an efficient carrier frequency offset (CFO) estimation algorithm for the orthogonal frequency-division multiplexing (OFDM)-based wireless local area networks (WLANs). The packet preamble information we use is based on the high rate WLAN standards adopted by the IEEE 802.11 standardization group. Numerical results are presented to demonstrate the effectiveness of the proposed algorithm.
166 citations
TL;DR: A blind source separation algorithm is proposed that is based on minimizing Renyi's mutual information by means of nonparametric probability density function (PDF) estimation, which avoids the problems of PDF inaccuracy due to truncation of series expansion and the estimation of joint PDFs in high-dimensional spaces given the typical paucity of data.
Abstract: A blind source separation algorithm is proposed that is based on minimizing Renyi's mutual information by means of nonparametric probability density function (PDF) estimation. The two-stage process consists of spatial whitening and a series of Givens rotations and produces a cost function consisting only of marginal entropies. This formulation avoids the problems of PDF inaccuracy due to truncation of series expansion and the estimation of joint PDFs in high-dimensional spaces given the typical paucity of data. Simulations illustrate the superior efficiency, in terms of data length, of the proposed method compared to fast independent component analysis (FastICA), Comon's (1994) minimum mutual information, and Bell and Sejnowski's (1995) Infomax.
154 citations
TL;DR: This work develops a new HMM, called local contextual HMM (LCHMM), by introducing the Gaussian mixture field where wavelet coefficients are assumed to locally follow theGaussian mixture distributions determined by their neighborhoods.
Abstract: Wavelet domain hidden Markov models (HMMs) have been proposed and applied to image processing, e.g., image denoising. We develop a new HMM, called local contextual HMM (LCHMM), by introducing the Gaussian mixture field where wavelet coefficients are assumed to locally follow the Gaussian mixture distributions determined by their neighborhoods. The LCHMM can exploit both the local statistics and the intrascale dependencies of wavelet coefficients at a low computational complexity. We show that the LCHMM combined with the "cycle-spinning" technique can achieve state-of-the-art image denoising performance.
137 citations
TL;DR: This work analyzes the behavioural mechanism of the likelihood ratio, identifies the reason for the unwanted phenomenon, and proposes a solution based on a smoothed likelihood ratio that gives a significant improvement to the original VAD.
Abstract: From an investigation of a statistical model-based voice activity detector (VAD), it is found that the likelihood ratio defined in the VAD has a fundamental problem at the offset regions of the speech signals. Thus, we analyze the behavioural mechanism of the likelihood ratio, identify the reason for the unwanted phenomenon, and propose a solution based on a smoothed likelihood ratio. Objective test results show that the proposed method gives a significant improvement to the original VAD. Additionally, the improved VAD results in a performance that is even superior to G.729B and comparable to AMR VAD option 2.
TL;DR: In this paper, the authors studied space-time block coding for single carrier zero-padded (ZP) block transmissions through finite impulse response (FIR) multipath fading channels of order L. They proved that the maximum diversity of order 2(L+1) is achieved with two transmit and one receive antenna.
Abstract: We study space-time (ST) block coding for single carrier zero-padded (ZP) block transmissions through finite impulse response (FIR) multipath fading channels of order L. We prove that the maximum diversity of order 2(L+1) is achieved with two transmit and one receive antenna. Simulations demonstrate that joint exploitation of multi-antenna diversity and multipath diversity leads to significantly enhanced performance.
TL;DR: The estimation of the channel and the design of optimal preambles (training sequences) for an OFDM system with two transmit and multiple receive antennas are considered.
Abstract: The so-called orthogonal frequency division multiplexing (OFDM) technique has received considerable interest, especially in the area of wireless local area networks (WLANs). One way of meeting the demands for increased data rates in WLANs is to provide the transmitter and receiver with multiple antennas. In this letter, we consider the estimation of the channel and the design of optimal preambles (training sequences) for an OFDM system with two transmit and multiple receive antennas.
TL;DR: The generalized likelihood ratio test (GLRT) for this problem is stated and related to a simpler ad-hoc detector and is compared to the conventional multichannel subspace detector and shows its robustness to nonidentical channels on data collected with the Westerbork radio telescope.
Abstract: We consider the detection of unknown Gaussian signals received by an array of uncalibrated nonidentical sensors, which is a problem that appears in radio astronomy. The problem is formulated as a test on the covariance structure. The generalized likelihood ratio test (GLRT) for this problem is stated and related to a simpler ad-hoc detector. We compare the method to the conventional multichannel subspace detector and show its robustness to nonidentical channels on data collected with the Westerbork radio telescope.
TL;DR: An extended split-radix fast Fourier transform (FFT) algorithm is proposed that has the same asymptotic arithmetic complexity as the conventional split- Radix FFT algorithm but has the advantage of fewer loads and stores.
Abstract: An extended split-radix fast Fourier transform (FFT) algorithm is proposed. The extended split-radix FFT algorithm has the same asymptotic arithmetic complexity as the conventional split-radix FFT algorithm. Moreover, this algorithm has the advantage of fewer loads and stores than either the conventional split-radix FFT algorithm or the radix-4 FFT algorithm.
TL;DR: The most efficient implementation of theforward and inverse MDCT computation for layer III in MPEG-1 and MPEG-2 international audio coding standards is proposed, based on a new fast algorithm for the forward and inverseMDCT computation in the oddly stacked system.
Abstract: The modified discrete cosine transform (MDCT) is employed in subband/transform coding schemes as the analysis/synthesis filter bank based on time domain aliasing cancellation (TDAC). The most efficient implementation of the forward and inverse MDCT computation for layer III in MPEG-1 and MPEG-2 international audio coding standards is proposed. It is based on a new fast algorithm for the forward and inverse MDCT computation in the oddly stacked system. The complete signal flow graphs for the implementation of MDCT and inverse MDCT in layer III are also provided.
TL;DR: Two normalized versions of Oja's (1992) algorithm (NOja and NOOja), which can be used for the estimation of minor and principal subspaces of a vector sequence, offer a faster convergence, orthogonality, and a better numerical stability.
Abstract: We present two normalized versions of Oja's (1992) algorithm (NOja and NOOja), which can be used for the estimation of minor and principal subspaces of a vector sequence. The new algorithms offer, as compared to Oja, a faster convergence, orthogonality, and a better numerical stability with a slight increase in computational complexity.
TL;DR: A simplified version of lattice factorization for linear-phase perfect reconstruction filter bank (LPPRFB) derived by T.D. Tran et al. is proposed, substantially reducing free parameters in nonlinear optimization and saving computation cost in hardware implementation.
Abstract: We propose a simplified version of lattice factorization for linear-phase perfect reconstruction filter bank (LPPRFB) derived by T.D. Tran et al. (see IEEE Trans. Signal Processing. vol.48, no.1, p.133-47, Jan. 2000). The proposed new lattice structure spans the same class of LPPRFB, while substantially reducing free parameters in nonlinear optimization and saving computation cost in hardware implementation. To further address the importance of our proposed structure, we generalize our factorization to multidimensional LPPRFB (MD-LPPRFB), and show its effectiveness.
TL;DR: A two-stage blind multiuser antenna array detector is proposed for the uplink of a wideband CDMA system subject to large delay spread and it is shown that it is possible to identify the symbols of all users without knowledge of their codes, channels, or directions.
Abstract: A two-stage blind multiuser antenna array detector is proposed for the uplink of a wideband CDMA system subject to large delay spread. It is shown that it is possible to identify the symbols of all users without knowledge of their codes, channels, or directions. The key idea is that low-rank decomposition of the three-dimensional (3-D) received data array can reduce the FIR multiple-input multiple-output (FIR-MIMO) CDMA problem into standard FIR single-input multiple-output (FIR-SIMO) problems.
TL;DR: A phase unwrapping method using a maximum likelihood estimation technique together with frequency diversity information to reconstruct highly discontinuous ground elevation profiles is presented.
Abstract: We present a phase unwrapping method using a maximum likelihood estimation technique together with frequency diversity information to reconstruct highly discontinuous ground elevation profiles. Frequency diversity can be obtained by considering the interferograms obtained by different couples of subband images.
TL;DR: This letter proposes to jointly optimize the TFR and distance measure by minimizing the (estimated) probability of classification error, and the resulting optimized classification method is applied to multicomponent chirp signals and real speech records (speaker recognition).
Abstract: Time-frequency representations (TFRs) are efficient tools for nonstationary signal classification. However, the choice of the TFR and of the distance measure employed is critical when no prior information other than a learning set of limited size is available. In this letter, we propose to jointly optimize the TFR and distance measure by minimizing the (estimated) probability of classification error. The resulting optimized classification method is applied to multicomponent chirp signals and real speech records (speaker recognition). Extensive simulations show the substantial improvement of classification performance obtained with our optimization method.
TL;DR: A fully adaptive normalized nonlinear gradient descent (FANNGD) algorithm for online adaptation of nonlinear neural filters is proposed and is shown to converge faster than previously introduced algorithms of this kind.
Abstract: A fully adaptive normalized nonlinear gradient descent (FANNGD) algorithm for online adaptation of nonlinear neural filters is proposed. An adaptive stepsize that minimizes the instantaneous output error of the filter is derived using a linearization performed by a Taylor series expansion of the output error. For rigor, the remainder of the truncated Taylor series expansion within the expression for the adaptive learning rate is made adaptive and is updated using gradient descent. The FANNGD algorithm is shown to converge faster than previously introduced algorithms of this kind.
TL;DR: In this article, supervised texture classification was performed using kernel PCA for texture feature extraction, and the principal components were computed within the product space of the input pixels making up the texture patterns, thereby producing a good performance.
Abstract: Kernel principal component analysis (PCA) has recently been proposed as a nonlinear extension of PCA. The basic idea is to first map the input space into a feature space via a nonlinear map and then compute the principal components in that feature space. This letter illustrates the potential of kernel PCA for texture classification. Accordingly, supervised texture classification was performed using kernel PCA for texture feature extraction. By adopting a polynomial kernel, the principal components were computed within the product space of the input pixels making up the texture patterns, thereby producing a good performance.
TL;DR: Two effective algorithms that reduce the computational complexity of state likelihood computation in mixture-based Gaussian speech recognition systems by exploiting the high dependence exhibited among subsequent feature vectors to predict the best scoring mixture for each state.
Abstract: This paper describes two effective algorithms that reduce the computational complexity of state likelihood computation in mixture-based Gaussian speech recognition systems. We consider a baseline recognition system that uses nearest-neighbor search and partial distance elimination (PDE) to compute state likelihoods. The first algorithm exploits the high dependence exhibited among subsequent feature vectors to predict the best scoring mixture for each state. The method, termed best mixture prediction (BMP), leads to further speed improvement in the PDE technique. The second technique, termed feature component reordering (FCR), takes advantage of the variable contribution levels made to the final distortion score for each dimension of the feature and mean space vectors. The combination of two techniques with PDE reduces the computational time for likelihood computation by 29.8% over baseline likelihood computation. The algorithms are shown to yield the same accuracy level without further memory requirements for the November 1992 ARPA Wall Street Journal (WSJ) task.
TL;DR: This paper proposes an efficient class of perfect reconstruction (PR) modulated filter banks (MFB) using SOPOT coefficients, based on a modified factorization of the DCT-IV matrix and the lossless lattice structure of the prototype filter.
Abstract: This paper proposes an efficient class of perfect reconstruction (PR) modulated filter banks (MFB) using sum-of-powers-of-two (SOPOT) coefficients. This is based on a modified factorization of the DCT-IV matrix and the lossless lattice structure of the prototype filter, which allows the coefficients to be represented in SOPOT form without affecting the PR condition. A genetic algorithm (GA) is then used to search for these SOPOT coefficients. Design examples show that SOPOT MFB with a good frequency characteristic can be designed with very low implementation complexity. The usefulness of the approach is demonstrated with a 16 channel design example.
TL;DR: This paper provides a stability analysis of a class of acoustic noise control algorithms by showing that the adapted models have more in common with nonlinear, finite impulse response (FIR) equation error models than with the infinite impulse response output error models they superficially resemble.
Abstract: This paper provides a stability analysis of a class of acoustic noise control algorithms by showing that the adapted models have more in common with nonlinear, finite impulse response (FIR) equation error models than with the infinite impulse response (IIR) output error models they superficially resemble. Stability results from the adaptive control literature are applied to show global stability in the noise free case, and to show exponential stability when the input is persistently excited. The latter demonstrates a robustness to mismodeling errors, disturbances such as noises, and allows results to be applied to the tracking of time-varying systems.
TL;DR: It is shown that the convergence properties of PPM remain valid under a simple summability condition on the relaxed averages of the errors, suggesting that the signal that least violates constraints in an average squared-distance sense remains valid.
Abstract: The parallel projection method (PPM) uses successive averages of projections onto constraint sets to construct a signal that least violates these constraints in an average squared-distance sense. In this paper, we study the robustness of PPM to errors in the computation of the projections. It is shown that the convergence properties of PPM remain valid under a simple summability condition on the relaxed averages of the errors.
TL;DR: It is concluded that using cross channel linear prediction in the multichannel perceptual audio coding does not provide a net coding gain and the whitening effect of the prediction filter increases either the energy or the energy peaks in the high frequency region.
Abstract: We have studied and concluded that the time domain cross channel prediction is generally not applicable to perceptual audio coding. From the statistical analysis, correlation among certain channels seems to be enough to provide some coding gain and the prediction also successfully reduces the total energy of the signal to be coded. But this energy decrease and its resulting bit reduction mainly happens in the low frequency bands. In fact, the whitening effect of the prediction filter increases either the energy or the energy peaks in the high frequency region. As a result, more bits are required to code the high-frequency part of the signal and this increase outpaces the bit reduction realized in the low frequency region. Therefore, using cross channel linear prediction in the multichannel perceptual audio coding does not provide a net coding gain.
TL;DR: This work considers a two-stage knot selection scheme for adaptively fitting splines to data subject to noise and shows that the proposed method is well suited for a variety of smoothing and noise filtering needs.
Abstract: A critical component of spline smoothing is the choice of knots, especially for curves with varying shapes and frequencies in its domain. We consider a two-stage knot selection scheme for adaptively fitting splines to data subject to noise. A potential set of knots is chosen based on information from certain wavelet decompositions with the intention of placing more points where the curve shows rapid changes. The final knot selection is then made based on statistical model selection ideas. We show that the proposed method is well suited for a variety of smoothing and noise filtering needs.
TL;DR: A new method for blind phase estimation that uses eight-order statistics is described, which provides about the same root mean square error (RMSE) as the fourth power method with about one-sixth to one-fourth the number of samples.
Abstract: A new method for blind phase estimation that uses eight-order statistics is described. For cross QAM constellations, it provides about the same root mean square error (RMSE) as the fourth power method with about one-sixth to one-fourth the number of samples. Monte Carlo simulations are provided to demonstrate the usefulness of the method.