This paper proposes an algorithm that adaptively estimates time-varying noise variance used in Kalman filtering for real-time speech signal enhancement. In the speech signal contaminated by white noise, the spectral components except dominant ones in high frequency band are expected to reflect the noise energy. Our approach is first to find the dominant energy bands over speech spectrum using LPC. We then calculate the average value of the actual spectral components over the high frequency region excluding the dominant energy bands and use it as the noise variance. The resulting noise variance estimate is then applied to Kalman filtering to suppress the background noise. Experimental results indicate that the proposed approach achieves a significant improvement in terms of speech enhancement over those of the conventional Kalman filtering that uses the average noise power over silence interval only. As a refinement of our results, we employ multiple-Kalman filtering with multiple noise models and improve the intelligibility.

This paper describes how the image sequences taken by a stationary video camera may be effectively processed to detect and track moving objects against a stationary background in real-time. Our approach is first to isolate the moving objects in image sequences via a modified adaptive background estimation method and then perform token tracking of multiple objects based on features extracted from the processed image sequences. In feature based multiple object tracking, the most prominent tracking issues are track initialization, data association, occlusions due to traffic congestion, and object maneuvering. While there are limited past works addressing these problems, most relevant tracking systems proposed in the past are independently focused to either "occlusion" or "data association" only. In this paper, we propose the KL-IMMPDA (Kanade Lucas-Interacting Multiple Model Probabilistic Data Association) filtering approach for multiple-object tracking to collectively address the key issues. The proposed method essentially employs optical flow measurements for both detection and track initialization while the KL-IMMPDA filter is used to accept or reject measurements, which belong to other objects. The data association performed by the proposed KL-IMMPDA results in an effective tracking scheme, which is robust to partial occlusions and image clutter of object maneuvering. The simulation results show a significant performance improvement for tracking multi-objects in occlusion and maneuvering, when compared to other conventional trackers such as Kalman filter.

This paper presents a new data hiding scheme under fractal image generation via Fourier filtering method for Computer Graphics (CG) applications. The data hiding operations are achieved in the frequency domain and a method similar to QAM used in digital communication is introduced for efficient embedding in order to explore both phase and amplitude components simultaneously. Consequently, this scheme enables us not only to generate a natural terrain surface without loss of fractalness analogous to the conventional scheme, but also to embed larger amounts of data into an image depending on the fractal dimension. This scheme ensures the correct decoding of the embedded data under lossy data compression such as JPEG by controlling the quantization exponent used in the embedding process.

New designs of MAXFLAT discrete and differentiating Hilbert transformers are presented using their interrelationships with digital differentiators. The new designs have the explicit formulas for their tap-coefficients, which are further modified to obtain a new class of narrow transition band filters, with a performance comparable to the Chebyshev filters.

The signed regressor algorithm, a variation of the least mean square (LMS) algorithm, is characterized by the estimation way of using the clipped reference signals, namely, its sign (). This clipping, equivalent to quantizing the reference signal to 1, only increases the estimation error by about 2 dB. This paper proposes to increase the number of the quantization steps to three, namely, 1 and 0, and shows that the 'tri-quantized-x' normalized least mean square (NLMS) algorithm with three quantization steps improves the convergence property.

This paper presents the design and simulation of a power efficient 1:4 interpolation FIR filter with partitioned look Up Table (LUT) structure. Using the symmetry of the filter coefficients and the contents of the LUT, the area of the proposed filter is minimized. The two filters share the partitioned LUT and activate the LUT selectively to realize the low power operation. Experimental results suggest that the proposed filter reduces the power consumption by 25% and simultaneously reduces the gate area by 7% compared to the previously proposed single-architecture dual-channel filter.

Aged people who live alone are in particular need of a daily health check, medication, and of warm communication with family and friends. The authors have been developing a life-support computer system with such functions. Among them, a daily health check function with the capability of measuring blood pressure, detecting diseases from coughing, and so on would in particular be very powerful for primary care. As a first step to achieving quick services for a daily health check with a personal computer, utilization of cough information is considered. Features of cough data are analyzed aiming at developing an automatic cough data detection method. This paper proposes a novel method for extracting cough signals from other types of signals. The differential coefficient of a low-pass filtered waveform is first shown to be an effective parameter for discriminating between vowel and cough signals, and the relationship between cut-off frequency and cough detection rate is clarified. This parameter is then applied to cough signals mixed with vowel signals or white noises to evaluate robustness. The evaluation tests show that the cough feature can be perfectly detected for a 20 dB S/N ratio when the cut-off frequency is set to 24 [Hz]. The experimental results suggest that the proposed cough detection method can be a useful tool as a primary care for aged people with a bronchitis like an asthmatic bronchitis and a bronchopneumonia.

A signal processing approach is discussed which has the potential for imaging strongly scattering objects from a series of scattering experiments. The method is based on a linear spectral estimation technique to replace the filtered backpropagation for limited discrete data and a subsequent nonlinear signal processing step to remove the contribution of multiple scattering my means of homomorphic filtering. Details of this approach are discussed and illustrated by applying the imaging algorithm to both simulated and real data.

A new current-mode (CM) multifunction filter with single input and three outputs (SITO) employing only three dual output current conveyors (DO-CCII) and four passive elements is presented. The proposed filter realizes three basic filter functions simultaneously all at high impedance outputs. No component matching is required and all the passive and active sensitivities are low.

This letter proposes a modified synchronous acquisition method using code-orthogonalizing filters (COFs) in an asynchronous direct sequence (DS) / CDMA. Improvements on the average acquisition time for several conditions are shown.

New explicit formulas for tap-coefficients of halfband low/high pass MAXFLAT non-recursive filters are presented by using their relationship with already presented maximally linear type IV differentiators. These formulas are modified to give a new class of narrow transition band filters, with a performance comparable to that of optimal filters.

This paper addresses the L-gain filtering problem for continuous-time linear systems with time-varying structured uncertainties and non-zero initial conditions. We propose a full order linear filter that renders the L-gain from disturbance to filtering error within a prescribed level by solving a linear matrix inequality (LMI) feasibility problem. The filter gain is specified by the solution to a set of LMI's. A numerical example is given to illustrate the proposed method.

As a preprocessor of the automatic speech recognizer in a noisy environment, a microphone array system has been investigated to reduce the environmental noise. In usual microphone array design, a plane wave is assumed for the sake of simplicity (far-field assumption). However, this far-field assumption does not always hold, resulting in distortion in the array output. In this report, the subspace method, which is one of the high resolution spectrum estimator, is applied to the near-field source localization problem. A high resolution method is necessary especially for the near-field source localization with a small-sized array. By combining the source localization technique with a spatial inverse filter, the signal coming from the multiple sources in the near-field range can be separated. The modified minimum variance beamformer is used to design the spatial inverse filter. As a result of the experiment in a real environment with two sound sources in the near-field range, 60-70% of word recognition rate was achieved.

This paper reviews the recent progress made by those working on optical filters and switches for photonic networks based on dense wavelength division multiplexing (WDM). While various kinds of optical devices have been developed for flexible and large-capacity networks, the key components for the WDM networks are narrow-band filters and switches. Three kinds of optical filter are described in this paper: thin-film interference filters, fiber grating filters and arrayed-waveguide-grating (AWG) filters. The optical switches reviewed here are mechanical fiber-type switches, thermo-optic switches made using planar-lightwave-circuit technologies, total-internal-reflection switches and micro-electromechanical-system switches. Each device has its own advantages, and has been or will be used in point-to-point WDM, optical add/drop mupliplexing systems and optical crossconnect systems. Further advances in optical components and technologies are expected to contribute greatly to the construction of future photonic networks.

In this paper, we propose an adaptive IIR equalizer based on prefiltering techniques. The proposed equalizer has a cascade structure of an ARMA prefilter and an adaptive FIR equalizer. The ARMA prefilter is designed based on the transfer function estimated by the gradient-type instrumental variable algorithm. Simulation results are shown to confirm the performance of the proposed adaptive IIR equalizer.

Recently the applications of MEMS (micro electro mechanical systems) have made remarkable progress in many filelds. The optical application of MEMS is one of the most promising because it provides micro mechano optical devices, the key components for high-perfromance systems in optical communication networks and data storage devices. This paper disucces the impacts of MEMS techologies on optical systems. Furthermore, state-of-the-art exmaples of micro optical switches, pig-tailed tunable filters and two-dimensional MEMS optical scanners are described.

A tunable optical Fabry-Perrot filter was designed by setting a single-mode optical fiber normal to the diaphragm of a capacitive pressure sensor. The silicon diaphragm is deflected by the electrostatic force generated by applying a voltage to the capacitive electrodes. According to the movement of the diaphragm, the peak wavelength changed from 1546 to 1551 nm when applied voltage was increased from 20 to 50 V. The relationship of the wavelength change to the applied voltage was derived from the silicon diaphragm deflection theory. That measured change of the wavelength agrees well with the wavelength change calculated from this relationship. The commercial pressure sensors are expected to be able to be used as a tunable optical filter.

We report a directly deposited dielectric multilayer onto an end face of a fluorinated polyimide optical waveguide by ion beam sputtering process. This dielectric multilayer (Ta2O5/SiO2) acts on a wavelength separation filter which passes 1310 nm wavelength signal and reflects 1550 nm wavelength signal.

This paper presents a three-dimensional polarimetric detection result of targets buried in snowpack by synthetic aperture FM-CW radar system. Since the FM-CW radar is suitable for short range sensing and can be equipped with fully polarimetric capability, we further extended it to a polarimetric three-dimensional SAR system. A field experiment was carried out to image and detect targets in a natural snowpack of 280 cm deep. The polarimetric detection and identification schemes are the polarimetric filtering, three-component decomposition, and the power polarization anisotropy coefficient. These approaches to acquired data show the usefulness of three-dimensional polarimetric FM-CW SAR system.

A constraint resolution scheme in the Hopfield-type neural network named "Neuron Filter" is presented for efficiently solving combinatorial optimization problems. The neuron filter produces an output that satisfies the constraints of the problem as best as possible according to both neuron inputs and outputs. This paper defines the neuron filter and shows its introduction into existing neural networks for N-queens problems and FPGA board-level routing problems. The performance is evaluated through simulations where the results show that our neuron filter improves the searching capability of the neural network with the shorter computation time.