In this letter, a micro-expression recognition method is investigated by integrating both spatio-temporal facial features and a regression model. To this end, we first perform a multi-scale facial region division for each facial image and then extract a set of local binary patterns on three orthogonal planes (LBP-TOP) features corresponding to divided facial regions of the micro-expression videos. Furthermore, we use GSLSR model to build the linear regression relationship between the LBP-TOP facial feature vectors and the micro expressions label vectors. Finally, the learned GSLSR model is applied to the prediction of the micro-expression categories for each test micro-expression video. Experiments are conducted on both CASME II and SMIC micro-expression databases to evaluate the performance of the proposed method, and the results demonstrate that the proposed method is better than the baseline micro-expression recognition method.

In this paper, we propose a localization algorithm that uses the time difference of arrival (TDOA) and the angle of arrival (AOA). The problem is formulated in a hybrid linear matrix equation. TDOA and AOA measurements are used for estimating the target's position. Although it is known that the accuracy of TDOA based localization is superior to that of AOA based localization, TDOA based localization has a poor vertical accuracy in deteriorated geometrical conditions. This paper, therefore, proposes a localization algorithm in which the vertical position is estimated by AOA measurements and the horizontal position is estimated by TDOA measurement in order to achieve high location accuracy in three dimensions. In addition, the Lagrange multipliers are obtained efficiently and robustly. The simulation analysis shows that the proposed constrained linear squares (CLS) algorithm is an unbiased estimator, and that it approaches the Cramer-Rao lower bound (CRLB) when the measurement noise and the sensor's location errors are sufficiently small.

Ultra high definition (UHD) imaging systems have attracted much attention as a next generation television (TV) broadcasting service and video streaming service. However, the state of the art video coding standards including H.265/HEVC has not enough compression rate for streaming, broadcasting and storing UHD. Existing coding standard such as H.265/HEVC normaly use RGB-YCbCr color transform before compressing RGB color image since that procedure can decorrelate color components well. However, there is room for improvement on the coding efficiency for color image based on an observation that the luminance and chrominance components changes in same locations. This observation inspired us to propose a new post-processing method for compressed images by using weighted least square (WLS) filter with coded luminance component as a guide image, for refining the edges of chrominance components. Since the computational cost of WLS tends to superlinearly increase with increasing image size, it is difficult to apply it to UHD images. To overcome this problem, we propose slightly overlapped block partitioning and a new variant of WLS (constrained WLS, CWLS). Experimental results of objective quality comparison and subjective assessment test using 4K images show that our proposed method can outperform the conventional method and reduce the bit amount for chrominance component drastically with preserving the subjective quality.

In recent years, various applications based on propagation characteristics have been developed. They generally utilize the locality of the fading characteristics of multipath environments. On the other hand, if a received signal at a remote location can be estimated beyond the correlation distance of the multipath fading environment, a wide variety of new applications can be possible. In this paper, we attempt to estimate a received signal at a remote location using the MUSIC method and the least squares method. Based on the plane wave assumption for each arriving wave, multipath environment is analyzed through estimation of the directions of arrival by the MUISC method and the complex amplitudes of the received signals by the least squares method, respectively. We present evaluation results on the estimation performance of the method by computer simulations.

This paper presents a downlink interference mitigation framework for two-tier heterogeneous networks, that consist of spectrum-sharing macrocells and femtocells*. This framework establishes cooperation between the two tiers through two algorithms, namely, the restricted waterfilling (RWF) algorithm and iterative reweighted least squares interference alignment (IRLS-IA) algorithm. The proposed framework models the macrocell-femtocell two-tier cellular system as an overlay cognitive radio system in which the macrocell system plays the role of the primary user (PU) while the femtocell networks play the role of the cognitive secondary users (SUs). Through the RWF algorithm, the macrocell basestation (MBS) cooperates with the femtocell basestations (FBSs) by releasing some of its eigenmodes to the FBSs to do their transmissions even if the traffic is heavy and the MBS's signal to noise power ratio (SNR) is high. Then, the FBSs are expected to achieve a near optimum sum rate through employing the IRLS-IA algorithm to mitigate both the co-tier and cross-tier interference at the femtocell users' (FUs) receivers. Simulation results show that the proposed IRLS-IA approach provides an improved sum rate for the femtocell users compared to the conventional IA techniques, such as the leakage minimization approach and the nuclear norm based rank constraint rank minimization approach. Additionally, the proposed framework involving both IRLS-IA and RWF algorithms provides an improved total system sum rate compared with the legacy approaches for the case of multiple femtocell networks.

Least squares error (LSE) method adopted recursively can be used to track the frequency and amplitude of signals in steady states and kinds of non-steady ones in power system. Taylor expansion is used to give another version of this recursive LSE method. Aided by variable-windowed short-time discrete Fourier transform, recursive LSEs with and without Taylor expansion converge faster than the original ones in the circumstance of off-nominal input singles. Different versions of recursive LSE were analyzed under various states, such as signals of off-nominal frequency with harmonics, signals with step changes, signals modulated by a sine signal, signals with decaying DC offset and additive Gaussian white noise. Sampling rate and data window size are two main factors influencing the performance of method recursive LSE in transient states. Recursive LSE is sensitive to step changes of signals, but it is in-sensitive to signals' modulation and singles with decaying DC offset and noise.

An accurate time-of-arrival (TOA) estimation method for isolated pulses positioning system is proposed in this paper. The method is based on a multi-level crossing timing (MCT) digitizer and least square (LS) criterion, namely LS-MCT method, in which TOA of the received signal is directly described as a parameterized combination of a set of MCT samples of the leading and trailing edges of the signal. The LS-MCT method performs a receiver training process, in which a GPS synchronized training pulse generator (TPG) is used to obtain training data and determine the parameters of the TOA combination. The LS method is then used to optimize the combination parameters with a minimization criterion. The proposed method is compared to the conventional TOA estimation methods such as leading edge level crossing discriminator (LCD), adaptive thresholding (ATH), and signal peak detection (PD) methods. Simulation results show that the proposed algorithm leads to lower sensitivity to signal-to-noise ratio (SNR) and attains better TOA estimation accuracy than available TOA methods.

Most images obtained with imaging sensors contain Bayer patterns and suffer from blurring caused by the lens. In order to convert a blurred Bayer-patterned image into a viewable image, demosaicing and deblurring are needed. These concepts have been major research areas in digital image processing for several decades. Despite their importance, their performance and efficiency are not satisfactory when considered independently. In this paper, we propose a joint deblurring and demosaicing method in which edge direction and edge strength are estimated in the Bayer domain and then edge adaptive deblurring and edge-oriented interpolation are performed simultaneously from the estimated edge information. Experimental results show that the proposed method produces better image quality than conventional algorithms in both objective and subjective terms.

In this paper, we present a new four parameter estimator of sampled sinusoidal signals that does not require iteration. Mathematically, the four parameters (frequency, phase, magnitude, and dc offset) of sinusoidal signals can be obtained when four data points are given. In general, the parameters have to be calculated with iteration since the equations are nonlinear. In this paper, we point out that the four parameters can be obtained analytically if the four data points given are measured using a fixed sampling interval. Analytical expressions for the four parameters are derived using the signal differences. Based on this analysis, we suggest an algorithm of estimating the four parameters from N data samples corrupted by noise without iteration. When comparing with the IEEE-1057 method which is based on the least-square method, the proposed algorithm does not require the initial guess of the parameters for iteration and avoid the convergence problem. Also, the number of required numerical operations for estimation is fixed if N is determined. As a result, the processing time of parameter estimation is much faster than the least-square method which has been confirmed by numerical simulations. Simulation results and the quantitative analysis show that the estimation error of the estimated parameters is less than 1.2 times the square root of the Cramer-Rao bounds when the signal to noise ratio is larger than 20dB.

This letter deals with a sparse signal recovery problem and proposes a new algorithm based on the iterative reweighted least squares (IRLS) algorithm. We assume that the non-zero values of a sparse signal is always greater than a given constant and modify the IRLS algorithm to satisfy this assumption. Numerical results show that the proposed algorithm recovers a sparse vector efficiently.

In multi-static sonar systems, the least square (LS) and maximum likelihood (ML) are the typical estimation criteria for target location estimation. The LS localizaiton has the advantage of low computational complexity. On the other hand, the performance of LS can be degraded severely when the target lies on or around the straight line between the source and receiver. We examine mathematically the reason for the performance degradation of LS. Then, we propose a location adaptive — least square (LA-LS) localization that removes the weakness of the LS localizaiton. LA-LS decides the receivers that produce abnormally large measurement errors with a proposed probabilistic measure. LA-LS achieves improved performance of the LS localization by ignoring the information from the selected receivers.

Two channel correlation estimation (CCE) schemes exploiting pilots are presented for an OFDM system with a comb-type pilot pattern under the assumption that there exist virtual subcarriers in the OFDM block. Whereas the first scheme is designed based on the conventional regularized-least square (LS) approach, the second scheme is designed by a newly devised technique based on LS. As the second scheme removes the necessity of computing the matrix inverse by making the minimum eigenvalue of the inversed matrix positive, it leads to reduced implementation complexity and improved performance. It is shown by simulation that the proposed CCE schemes substantially enhance the mean equare error and symbol error rate performances of the MMSE based channel estimation by providing more accurate channel correlation information.

In this paper, we propose a rectangular weighting function that can be used in the method of iteratively reweighted least squares (IRWLS) for designing equiripple all-pass IIR filters. The purpose of introducing this weighting function is to improve the convergence performance in the solution of the IRWLS. The height of each rectangle is designed to be equal to the local maximum of each ripple, and the width of each rectangle is designed so that the area of each rectangle becomes equal to the area of each ripple. Here, the ripple is the absolute value of the phase error. We show experimentally that the convergence performance in the solution of the IRWLS can be improved by using the proposed weighting function.

This paper presents a method for learning an overcomplete, nonnegative dictionary and for obtaining the corresponding coefficients so that a group of nonnegative signals can be sparsely represented by them. This is accomplished by posing the learning as a problem of nonnegative matrix factorization (NMF) with maximization of the incoherence of the dictionary and of the sparsity of coefficients. By incorporating a dictionary-incoherence penalty and a sparsity penalty in the NMF formulation and then adopting a hierarchically alternating optimization strategy, we show that the problem can be cast as two sequential optimal problems of quadratic functions. Each optimal problem can be solved explicitly so that the whole problem can be efficiently solved, which leads to the proposed algorithm, i.e., sparse hierarchical alternating least squares (SHALS). The SHALS algorithm is structured by iteratively solving the two optimal problems, corresponding to the learning process of the dictionary and to the estimating process of the coefficients for reconstructing the signals. Numerical experiments demonstrate that the new algorithm performs better than the nonnegative K-SVD (NN-KSVD) algorithm and several other famous algorithms, and its computational cost is remarkably lower than the compared algorithms.

An instrumental variable (IV) based linear estimator is proposed for effective target localization in sensor network by using time-difference-of-arrival (TDOA) measurement. Although some linear estimation approaches have been proposed in much literature, the target localization based on TDOA measurement still has a room for improvement. Therefore, we analyze the estimation errors of existing localization estimators such as the well-known quadratic correction least squares (QCLS) and the robust least squares (RoLS), and demonstrate advantages of the proposition by comparing the estimation errors mathematically and showing localization results through simulation. In addition, a recursive form of the proposition is derived to consider a real time application.

An asymmetric classifier based on kernel partial least squares is proposed for software defect prediction. This method improves the prediction performance on imbalanced data sets. The experimental results validate its effectiveness.

In this paper, a new Hammerstein predistorter modeling for power amplifier (PA) linearization is proposed. The key feature of the model is that the cubic splines, instead of conventional high-order polynomials, are utilized as the static nonlinearities due to the fact that the splines are able to represent hard nonlinearities accurately and circumvent the numerical instability problem simultaneously. Furthermore, according to the amplifier's AM/AM and AM/PM characteristics, real-valued cubic spline functions are utilized to compensate the nonlinear distortion of the amplifier and the following finite impulse response (FIR) filters are utilized to eliminate the memory effects of the amplifier. In addition, the identification algorithm of the Hammerstein predistorter is discussed. The predistorter is implemented on the indirect learning architecture, and the separable nonlinear least squares (SNLS) Levenberg-Marquardt algorithm is adopted for the sake that the separation method reduces the dimension of the nonlinear search space and thus greatly simplifies the identification procedure. However, the convergence performance of the iterative SNLS algorithm is sensitive to the initial estimation. Therefore an effective normalization strategy is presented to solve this problem. Simulation experiments were carried out on a single-carrier WCDMA signal. Results show that compared to the conventional polynomial predistorters, the proposed Hammerstein predistorter has a higher linearization performance when the PA is near saturation and has a comparable linearization performance when the PA is mildly nonlinear. Furthermore, the proposed predistorter is numerically more stable in all input back-off cases. The results also demonstrate the validity of the convergence scheme.

In this paper, a linear optimization of the dual-tree complex wavelet transform (DTCWT) based on the least squares method is proposed. The proposed method can design efficient DTCWTs by improving the design degrees of freedom and solving the least square solution iteratively. Because the resulting DTCWTs have good approximation accuracy of the half sample delay condition and the stopband attenuation, they provide precise shift-invariance and directionality. Finally, the proposed DTCWTs are evaluated by applying to non-linear approximation and image denoising, and showed their effectiveness, compared with the conventional DTCWTs.

A new compact form of the sliding window recursive least squares (SWRLS) algorithm, the I-SWRLS algorithm, is derived using an indefinite matrix. The resultant algorithm has a form similar to that of the traditional recursive least squares (RLS) algorithm, and is more computationally efficient than the conventional SWRLS algorithm including two Riccati equations. Furthermore, a computationally reduced version of the I-SWRLS algorithm is developed utilizing a shift property of the correlation matrix of input data. The resulting fast algorithm reduces the computational complexity from O(N2) to O(N) per iteration when the filter length (tap number) is N, but retains the same tracking performance as the original algorithm. This fast algorithm is much easier to implement than the existing SWC FTF algorithms.

In this paper, we propose a spatially adaptive gradient-projection algorithm for the H.264 video coding standard to remove coding artifacts using local statistics. A hybrid method combining a new weighted constrained least squares (WCLS) approach and the projection onto convex sets (POCS) approach is introduced, where weighting components are determined on the basis of the human visual system (HVS) and projection set is defined by the difference between adjacent pixels and the quantization index (QI). A new visual function is defined to determine the weighting matrices controlling the degree of global smoothness, and a projection set is used to obtain a solution satisfying local smoothing constraints, so that the coding artifacts such as blocking and ringing artifacts can be simultaneously removed. The experimental results show the capability and efficiency of the proposed algorithm.