Hiroshi Ijima

Deterministic regression methods for unbiased estimation of time-varying autoregressive parameters from noisy observations

A great deal of interest has been paid to autoregressive parameter estimation in the noise-free case or when the observation data are disturbed by random noise. Tracking time-varying autoregressive (TVAR) parameters has been also discussed, but few papers deal with this issue when there is an additive zero-mean white Gaussian measurement noise. In this paper, one considers deterministic regression methods (or evolutive methods) where the TVAR parameters are assumed to be weighted combinations of basis functions. However, the additive white measurement noise leads to a weight-estimation bias when standard least squares methods are used. Therefore, we propose two alternative blind off-line methods that allow both the variance of the additive noise and the weights to be estimated. The first one is based on the errors-in-variable issue whereas the second consists in viewing the estimation issue as a generalized eigenvalue problem. A comparative study with other existing methods confirms the effectiveness of the proposed methods.

Prediction error method to estimate the ar parameters when the AR process is disturbed by a colored noise

Estimating the autoregressive parameters from noisy observations has been addressed by various authors for the last decades. Although several on-line or off-line approaches have been proposed when the additive noise is white, few papers deal with the additive moving average noise. In this paper, we suggest estimating the model parameters by using the prediction error method. Despite its high computational cost, the method has the advantage of being efficient in the Gaussian case. A comparative study with existing methods is then carried out and points out the efficiency of our approach especially when the number of samples is small.

Quantification of carbohydrate based on scan image analysis for TLC technique compensating lack of spot overlaps

In this paper, a quantification method of the carbohydrate is proposed for samples of spot tests using the thin-layer chromatography (TLC) based on the scanned image analysis compensating lack of overlap between each spot. The color density of image data is modeled by 2-dimensional Gaussian function. Parameters in the Gaussian function are estimated by calculating marginal distribution and using least squares (LS) method. Sample image of the spots on the TLC plate are quantified as calculating the volume of Gaussian function. A numerical example is then carried out for quantifying the glucose which is an important and commonly unit of the carbohydrate. The efficiency of our method is pointed out comparing with two non-compensated methods. In addition, in order to investigate the performance of proposed method, calibrated values are also compared with results using the high performance liquid chromatography (HPLC).

Evolutive method based on a generalized eigenvalue decomposition to estimate time varying autoregressive parameters from noisy observations

A great deal of interest has been paid to the estimation of time-varying autoregressive (TVAR) parameters. However, when the observations are disturbed by an additive white measurement noise, using standard least squares methods leads to a weight-estimation bias. In this paper, we propose to jointly estimate the TVAR parameters and the measurement-noise variance from noisy observations by means of a generalized eigenvalue decomposition. It extends to the TVAR case an off-line method that was initially proposed for AR parameter estimation from noisy observations. A comparative study is then carried out with existing methods such as the recursive errors-in-variable approach and Kalman based algorithms.

Detection of Signals in Nonstationary Random Noise via Stationarization of Data Incorporated with Kalman Filter

Recently, the authors have proposed a method for the detection of signals corrupted by nonstationary random noise based on stationarization of the observation data which can be modeled by the first-order Ito stochastic differential equation. In this paper, in order to apply this method to more general situation, we propose a stationarization method incorporated with Kalman filter. To test the proposed method simulation experiments are presented.

Estimation of Motion Parameters of Moving Target using Wigner Distribution

The purpose of this paper is to estimate unknown motion parameters, acceleration and initial velocity, from the radar signal corrupted by random noise. The principal attack of the approach is to use the (pseudo)-Wigner distribution which is computed from the noisy observation data. Parameters are estimated by least-squares method for the noisy instantaneous frequency of returned signal. Numerical simulations are presented to verify the efficacy of the proposed method.