Showing papers in "IEEE Signal Processing Letters in 2003"
TL;DR: The proposed robust Capon beamformer can no longer be expressed in a closed form, but it can be efficiently computed and its excellent performance is demonstrated via a number of numerical examples.
Abstract: The Capon beamformer has better resolution and much better interference rejection capability than the standard (data-independent) beamformer, provided that the array steering vector corresponding to the signal of interest (SOI) is accurately known. However, whenever the knowledge of the SOI steering vector is imprecise (as is often the case in practice), the performance of the Capon beamformer may become worse than that of the standard beamformer. We present a natural extension of the Capon beamformer to the case of uncertain steering vectors. The proposed robust Capon beamformer can no longer be expressed in a closed form, but it can be efficiently computed. Its excellent performance is demonstrated via a number of numerical examples.
367 citations
TL;DR: An image enhancement algorithm for images compressed using the JPEG standard is presented, based on a contrast measure defined within the discrete cosine transform (DCT) domain that does not affect the compressibility of the original image.
Abstract: An image enhancement algorithm for images compressed using the JPEG standard is presented. The algorithm is based on a contrast measure defined within the discrete cosine transform (DCT) domain. The advantages of the psychophysically motivated algorithm are 1) the algorithm does not affect the compressibility of the original image because it enhances the images in the decompression stage and 2) the approach is characterized by low computational complexity. The proposed algorithm is applicable to any DCT-based image compression standard, such as JPEG, MPEG 2, and H. 261.
317 citations
TL;DR: It is proved that speech samples during voice activity intervals are Laplacian random variables, and all marginal distributions of speech are accurately described by LD in decorrelated domains.
Abstract: It is demonstrated that the distribution of speech samples is well described by Laplacian distribution (LD). The widely known speech distributions, i.e., LD, Gaussian distribution (GD), generalized GD, and gamma distribution, are tested as four hypotheses, and it is proved that speech samples during voice activity intervals are Laplacian random variables. A decorrelation transformation is then applied to speech samples to approximate their multivariate distribution. To do this, speech is decomposed using an adaptive Karhunen-Loeve transform or a discrete cosine transform. Then, the distributions of speech components in decorrelated domains are investigated. Experimental evaluations prove that the statistics of speech signals are like a multivariate LD. All marginal distributions of speech are accurately described by LD in decorrelated domains. While the energies of speech components are time-varying, their distribution shape remains Laplacian.
302 citations
TL;DR: A new forward-backward algorithm is proposed whose computational complexity is only O((MD + M/sup 2/)T), a reduction by almost a factor of D when D > M and whose memory requirement is O(MT).
Abstract: Existing algorithms for estimating the model parameters of an explicit-duration hidden Markov model (HMM) usually require computations as large as O((MD/sup 2/ + M/sup 2/)T) or O(M/sup 2/ DT), where M is the number of states; D is the maximum possible interval between state transitions; and T is the period of observations used to estimate the model parameters. Because of such computational requirements, these algorithms are not practical when we wish to construct an HMM model with large state space and large explicit state duration and process a large amount of measurement data to obtain high accuracy. We propose a new forward-backward algorithm whose computational complexity is only O((MD + M/sup 2/)T), a reduction by almost a factor of D when D > M and whose memory requirement is O(MT). As an application example, we discuss an HMM characterization of access traffic observed at a large-scale Web site: we formulate the Web access pattern in terms of an HMM with explicit duration and estimate the model parameters using our algorithm.
230 citations
TL;DR: A fast algorithm to approximate the Kullback-Leibler distance (KLD) between two dependence tree models is presented, which offers a saving of hundreds of times in computational complexity compared to the commonly used Monte Carlo method.
Abstract: We present a fast algorithm to approximate the Kullback-Leibler distance (KLD) between two dependence tree models. The algorithm uses the "upward" (or "forward") procedure to compute an upper bound for the KLD. For hidden Markov models, this algorithm is reduced to a simple expression. Numerical experiments show that for a similar accuracy, the proposed algorithm offers a saving of hundreds of times in computational complexity compared to the commonly used Monte Carlo method. This makes the proposed algorithm important for real-time applications, such as image retrieval.
229 citations
TL;DR: It is demonstrated that the denoising performance of Wiener filtering may be increased by preprocessing images with a thresholding operation and the approximate analysis of the errors occurring in empiricalWiener filtering is presented.
Abstract: The approximate analysis of the errors occurring in empirical Wiener filtering is presented. We demonstrate that the denoising performance of Wiener filtering may be increased by preprocessing images with a thresholding operation.
196 citations
TL;DR: Experimental results show that this approach is better than the conventional approach, which only uses the term-by-term multiwavelet denoising, and it outperforms neighbor single wavelet Denoising for some standard test signals and real-life images.
Abstract: Multiwavelets give better results than single wavelets for signal denoising. We study multiwavelet thresholding by incorporating neighboring coefficients. Experimental results show that this approach is better than the conventional approach, which only uses the term-by-term multiwavelet denoising. Also, it outperforms neighbor single wavelet denoising for some standard test signals and real-life images. This is an extension to Cai and Silverman's (see Sankhya: Ind. J. Stat. B, pt.2, vol.63, p.127-148, 2001) work.
176 citations
TL;DR: A superimposed periodic pilot scheme for finite-impulse response (FIR) channel estimation is proposed and the variance expression of the linear channel estimate is derived and compared with the Cramer-Rao bound.
Abstract: Multipath is a major impairment in a wireless communications environment, and channel estimation algorithms are of interest. We propose a superimposed periodic pilot scheme for finite-impulse response (FIR) channel estimation. A simple first-order statistic is used, and any FIR channel can be estimated. There is no loss of information rate but a controllable increase in transmission power. We derive the variance expression of our linear channel estimate and compare with the Cramer-Rao bound. Numerical examples illustrate the effectiveness of the proposed method.
151 citations
TL;DR: Experimental results demonstrate the efficiency of the proposed method, since it clearly detects the time location and duration of LS and BS, despite their variation either in time duration and/or amplitude.
Abstract: An efficient technique for detecting explosive lung sounds (LS) (fine/coarse crackles and squawks) or bowel sounds (BS) in clinical auscultative recordings is presented. The technique is based on a fractal-dimension (FD) analysis of the recorded LS and BS obtained from controls and patients with pulmonary and bowel pathology, respectively. Experimental results demonstrate the efficiency of the proposed method, since it clearly detects the time location and duration of LS and BS, despite their variation either in time duration and/or amplitude. A noise stress test justifies the noise robustness of the FD-based detector, indicating its potential use in everyday clinical practice.
128 citations
TL;DR: A new approach to adaptive beamforming with sidelobe control is developed that minimizes the array output power while maintaining the distortionless response in the direction of the desired signal and a sidelobe level that is strictly guaranteed to be lower than some given threshold value.
Abstract: A new approach to adaptive beamforming with sidelobe control is developed. The proposed beamformer represents a modification of the popular minimum variance distortionless response (MVDR) beamformer. It minimizes the array output power while maintaining the distortionless response in the direction of the desired signal and a sidelobe level that is strictly guaranteed to be lower than some given (prescribed) threshold value. The resulting modified MVDR problem is shown to be convex, and its second-order cone (SOC) formulation is obtained that facilitates a computationally efficient way to implement our beamformer using the interior point method.
117 citations
TL;DR: In continuous-speech recognition experiments using SRI International's DECIPHER recognition system, both using artificially added noise and using recorded noisy speech, the combined-microphone approach significantly outperforms the single- microphone approach.
Abstract: We present a method to combine the standard and throat microphone signals for robust speech recognition in noisy environments. Our approach is to use the probabilistic optimum filter (POF) mapping algorithm to estimate the standard microphone clean-speech feature vectors, used by standard speech recognizers, from both microphones' noisy-speech feature vectors. A small untranscribed "stereo" database (noisy and clean simultaneous recordings) is required to train the POF mappings. In continuous-speech recognition experiments using SRI International's DECIPHER recognition system, both using artificially added noise and using recorded noisy speech, the combined-microphone approach significantly outperforms the single-microphone approach.
TL;DR: The signal subspace approach for speech enhancement is extended to colored-noise processes and explicit forms for the linear time-domain- and spectral- domain-constrained estimators are presented.
Abstract: The signal subspace approach for speech enhancement is extended to colored-noise processes. Explicit forms for the linear time-domain- and spectral-domain-constrained estimators are presented. These estimators minimize the average signal distortion power for given constraints on the residual noise power in the time and spectral domains, respectively. Equivalent implementations of the two estimators using the whitening approach are described.
TL;DR: The proposed stochastic gradient for Shannon's entropy can be used in online adaptation problems where the optimization of an entropy-based cost function is necessary.
Abstract: Entropy has found significant applications in numerous signal processing problems including independent components analysis and blind deconvolution. In general, entropy estimators require O(N/sup 2/) operations, N being the number of samples. For practical online entropy manipulation, it is desirable to determine a stochastic gradient for entropy, which has O(N) complexity. In this paper, we propose a stochastic Shannon's entropy estimator. We determine the corresponding stochastic gradient and investigate its performance. The proposed stochastic gradient for Shannon's entropy can be used in online adaptation problems where the optimization of an entropy-based cost function is necessary.
TL;DR: A new region-of-interest (ROI) coding method called partial significant bitplanes shift (PSBShift) that combines the advantages of the two standard ROI coding methods defined in JPEG2000 is proposed.
Abstract: We propose a new region-of-interest (ROI) coding method called partial significant bitplanes shift (PSBShift) that combines the advantages of the two standard ROI coding methods defined in JPEG2000. The PSBShift method not only supports arbitrarily shaped ROI coding without coding the shape, but also enables the flexible adjustment of compression quality in ROI and background. Additionally, the new method can efficiently code multiple ROIs with different degrees of interest in an image.
TL;DR: A new technique for achieving blind source separation when given only a single-channel recording is presented based on exploiting the inherent time structure of sound sources by learning a priori sets of time-domain basis functions that encode the sources in a statistically efficient manner.
Abstract: We present a new technique for achieving blind source separation when given only a single-channel recording. The main idea is based on exploiting the inherent time structure of sound sources by learning a priori sets of time-domain basis functions that encode the sources in a statistically efficient manner. We derive a learning algorithm using a maximum likelihood approach given the observed single-channel data and sets of basis functions. For each time point, we infer the source parameters and their contribution factors using a flexible but simple density model. We show the separation results of two music signals as well as the separation of two voice signals.
TL;DR: It is proved that filter length and the envelope of impulse response are the important factors in convergence rate control for the impulse response with an exponential decay envelope.
Abstract: This letter analyzes the mean-square convergence of a deficient-length least mean-square adaptive filter whose length is less than that of the unknown system and proves that filter length and the envelope of impulse response are the important factors in convergence rate control For the impulse response with an exponential decay envelope, which covers a large set of physical systems, eg, acoustic echo path, an optimal filter length sequence is figured out to achieve the fastest convergence The simulations of an exact exponential decay envelope and of a real-life echo path in a car environment are performed via computer, and the results validate our analysis well
TL;DR: A simple analytic expression for the change in coherent weighted integration gain due to a white Gaussian error or noise in the phase of the integrated samples is developed and shown by simulation to be very accurate for any reasonable value of phase noise standard deviation.
Abstract: We develop a simple analytic expression for the change in coherent weighted integration gain due to a white Gaussian error or noise in the phase of the integrated samples. Our expression is shown by simulation to be very accurate for any reasonable value of phase noise standard deviation. The result is useful in estimating the performance impact on coherent signal processing systems of oscillator noise, residual motion compensation errors, and other system imperfections that are manifested primarily as phase errors.
TL;DR: A simple modification to the CIC filter that enhances its SRC performance at the expense of requiring a few extra computations per output sample is proposed.
Abstract: Cascaded-integrator-comb (CIC) filters perform sample rate conversion (SRC) efficiently using only additions/subtractions. However, the limited number of tuning parameters may make conventional CIC filters unsuitable for SRC in software radio (SWR) systems. A simple modification to the CIC filter that enhances its SRC performance at the expense of requiring a few extra computations per output sample is proposed. Simulation results show that the modified CIC filter outperforms the conventional CIC filter for the purpose of SRC in SWR systems.
TL;DR: The generalized likelihood ratio test (GLRT) is invariants with respect to transformations for which the hypothesis testing problem itself is invariant, an important property of the GLRT in light of its widespread use and the recent interest in invariant tests applied to signal processing applications.
Abstract: The generalized likelihood ratio test (GLRT) is invariant with respect to transformations for which the hypothesis testing problem itself is invariant. This result from the statistics literature is presented in the context of some simple signal models. This is an important property of the GLRT in light of its widespread use and the recent interest in invariant tests applied to signal processing applications. The GLRT is derived for some examples in which the uniformly most powerful invariant (UMPI) test does and does not exist, including one in which the UMPI test exists and is not given by the GLRT.
TL;DR: The adaptive Landweber method (ALM) to reconstruct an image from a blurred observation has a higher convergence rate and lower MSE and mean absolute error than the standard LM and emphasizes speed at the beginning stages and stability at late stages of iteration.
Abstract: We present an adaptive Landweber method (ALM) to reconstruct an image from a blurred observation. The standard Landweber method (LM) is an iterative method to solve "ill-posed" problems encountered in image restoration. The standard LM uses a constant update parameter. It has the disadvantage of slow convergence. Instead of using a constant update parameter, the adaptive method computes the update parameter at each iteration. In the ALM, the adaptive update parameter is calculated as the ratio of the L/sub 2/ norm of the first-order derivatives of the restored images at current and previous iterations. The adaptive LM emphasizes speed at the beginning stages and stability at late stages of iteration. The ALM has a higher convergence rate and lower MSE and mean absolute error than the standard LM. We use examples to demonstrate the performance of the ALM.
TL;DR: A new algorithm for fast computation of the modulated complex lapped transform (MCLT) is presented, based on computing a length-2M fast Fourier transform plus M butterfly-like stages, without data shuffling, which reduces the number of operations and memory accesses when compared to previous algorithms.
Abstract: A new algorithm for fast computation of the modulated complex lapped transform (MCLT) is presented. For a length-M MCLT, the algorithm is based on computing a length-2M fast Fourier transform plus M butterfly-like stages, without data shuffling. That reduces the number of operations and memory accesses needed to compute the MCLT when compared to previous algorithms. Compared to the original type-4 discrete-cosine-transform-based algorithm, an implementation of the new algorithm leads to 25% faster execution.
TL;DR: It is shown that copula theory leads to a noniterative method for constructing positive TFDs, and Cohen-Posch (1985) theory of positive time-frequency distributions andCopula theory are formally equivalent.
Abstract: We establish connections between Cohen-Posch (1985) theory of positive time-frequency distributions (TFDs) and copula theory. Both are aimed at designing joint probability distributions with fixed marginals, and we demonstrate that they are formally equivalent. Moreover, we show that copula theory leads to a noniterative method for constructing positive TFDs. Simulations show typical results.
TL;DR: A method for the postprocessing of JPEG-2000 compressed images at very low bitrates that counter-intuitively employs further compression to achieve image enhancement and demonstrates its applicability to wavelet coders.
Abstract: Motivated by error concealment applications, this letter proposes a method for the postprocessing of JPEG-2000 compressed images at very low bitrates. The proposed method counter-intuitively employs further compression to achieve image enhancement. This approach, although not widely known, is not entirely new: it is an adaptation of a technique originally designed for the removal of block-transform coding artifacts. The contribution of this work is to demonstrate its applicability to wavelet coders. In its simplest form, this algorithm uses existing system components with little or no additional hardware or software. Experimental results show a distinct reduction of ringing artifacts at very low bitrates.
TL;DR: It is shown that the problem addressed can be recast to a convex optimization one characterized by linear matrix inequalities (LMIs), and therefore a numerically attractive LMI approach can be exploited to test the robust stability of the uncertain discrete-time 2-D systems.
Abstract: We deal with the robust stability problem for linear two-dimensional (2-D) discrete time-invariant systems described by a 2-D local state-space (LSS) Fornasini-Marchesini (1989) second model. The class of systems under investigation involves parameter uncertainties that are assumed to be norm-bounded. We first focus on deriving the sufficient conditions under which the uncertain 2-D systems keep robustly asymptotically stable for all admissible parameter uncertainties. It is shown that the problem addressed can be recast to a convex optimization one characterized by linear matrix inequalities (LMIs), and therefore a numerically attractive LMI approach can be exploited to test the robust stability of the uncertain discrete-time 2-D systems. We further apply the obtained results to study the robust stability of perturbed 2-D digital filters with overflow nonlinearities.
TL;DR: It is shown that when estimating the a posteriori probability density of a possible signal in noise by means of a particle filter, the output of the filter can be used to approximately construct the likelihood ratio, which arises in many different detection schemes.
Abstract: In this paper, we present a new result that can be used for detection purposes. We show that when estimating the a posteriori probability density of a possible signal in noise by means of a particle filter, the output of the filter, i.e., the unnormalized weights, can be used to approximately construct the likelihood ratio, which arises in many different detection schemes.
TL;DR: This article proposes an efficient implementation of the Farrow (1988) structure using sum-of-powers- of-two (SOPOT) coefficients and multiplier-block (MB), and a novel algorithm for designing the F arrow coefficients in SOPOT form is detailed.
Abstract: This article proposes an efficient implementation of the Farrow (1988) structure using sum-of-powers-of-two (SOPOT) coefficients and multiplier-block (MB). In particular, a novel algorithm for designing the Farrow coefficients in SOPOT form is detailed. Using the SOPOT coefficient representation, coefficient multiplication can be implemented with limited number of shifts and additions. Using MB, the redundancy between multipliers can be fully exploited through the reuse of the intermediate results generated. Design examples show that the proposed method can greatly reduce the complexity of the Farrow structure while providing comparable phase and amplitude responses.
TL;DR: A novel adaptive runlength (ARL) coding scheme that encodes RUN and LEVEL separately using adaptive binary arithmetic coding and simple context modeling is introduced that outperforms the conventional runlength coding scheme by a large margin in the rate-distortion sense.
Abstract: Runlength coding is the standard coding technique for block transform-based image/video compression. A block of quantized transform coefficients is first represented as a sequence of RUN/LEVEL pairs that are then entropy coded-RUN being the number of consecutive zeros and LEVEL being the value of the following nonzero coefficient. We point out the inefficiency of conventional runlength coding and introduce a novel adaptive runlength (ARL) coding scheme that encodes RUN and LEVEL separately using adaptive binary arithmetic coding and simple context modeling. We aim to maximize compression efficiency by adaptively exploiting the characteristics of block transform coefficients and the dependency between RUN and LEVEL. Coding results show that with the same level of complexity, the proposed ARL coding algorithm outperforms the conventional runlength coding scheme by a large margin in the rate-distortion sense.
TL;DR: A new method is proposed to approximate an IIR filter by an FIR filter, which directly yields an optimal approximation with respect to the H/sup /spl infin// error norm, and it is shown that this design problem can be reduced to a linear matrix inequality.
Abstract: Finite-impulse response (FIR) filters are often preferred to infinite-impulse response (IIR) filters because of their various advantages in respect of stability, phase characteristic, implementation, etc. This article proposes a new method to approximate an IIR filter by an FIR filter, which directly yields an optimal approximation with respect to the H/sup /spl infin// error norm. We show that this design problem can be reduced to a linear matrix inequality. We also make a comparison via a numerical example with an existing method, known as the Nehari shuffle.
TL;DR: A low-complexity algorithm for embedding watermarks into two or more error-diffused images and the two halftone images can be made from two totally different gray-tone images and still provide a clear and sharp visual decoding pattern.
Abstract: In this letter, we propose a low-complexity algorithm for embedding watermarks into two or more error-diffused images. The first one is only a regular error-diffused image, and the others are achieved by applying the proposed noise-balanced error diffusion technique (NBEDF) to the original gray-level image. The visual decoding pattern can be perceived when these two or more similar error-diffused images are overlaid each other. Furthermore, with the proposed modified version of NBEDF, the two halftone images can be made from two totally different gray-tone images and still provide a clear and sharp visual decoding pattern.
TL;DR: This letter proposes a new time-domain equalizer training algorithm aimed at minimizing the total interblock interference power seen by an orthogonal frequency division multiplexing receiver by emphasizing the suppression of the outer edges of the channel response more with the help of a weight matrix.
Abstract: This letter proposes a new time-domain equalizer (TEQ) training algorithm aimed at minimizing the total interblock interference power seen by an orthogonal frequency division multiplexing receiver. The new TEQ achieves this goal by emphasizing the suppression of the outer edges of the channel response more with the help of a weight matrix.