Search

In this paper, a method using a subadaptive filter playing the supplementary role of acoustic echo cancellation is discussed. Since the updating state of the adaptive filter can be calculated approximately as a characteristic of the subadaptive filter, the step size parameter can be optimally controlled depending on the estimation accuracy. Also, for prevention of echo path variation, we discuss a variable detector in which the variation of the filter coefficients at the time of variation is considered. The proposed method is found to be robust to channel variations and can improve the ERLE by 10 to 15 dB in comparison with the conventional method. © 2006 Wiley Periodicals, Inc. Electron Comm Jpn Pt 3, 90(2): 9–21, 2007; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecjc.20231

The subband adaptive Volterra filter (SBAVF), which can assign the filter length of each subband arbitrarily, is proposed. If the frequency characteristics of a system to be identified are shifted to a certain bandwidth, it is usually not possible to carry out efficient identification by taking account of the shift of the bandwidth in a (full band) adaptive Volterra filter. On the other hand, in the proposed SBAVF, efficient system identification is possible by assigning a longer filter length to the dominant frequency range of the system of interest. When the proposed method is applied to a speaker system, it is shown through simulation results that the error attenuation can be improved from 1.5 dB to 7.5 dB in comparison with the conventional method. © 2004 Wiley Periodicals, Inc. Electron Comm Jpn Pt 3, 87(7): 45–54, 2004; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecjc.10135

In the problem of restoration of images degraded by impulsive noise, a method has been proposed to eliminate impulsive noise by using the noise position image formed with noise detectors. However, such a method estimates the signal values of the pixels to be processed by using only the nondegraded pixels within the filter window. Hence, if the majority of the window is occupied by impulsive noise, the signal information needed for estimation becomes insufficient and the estimation accuracy is significantly reduced. In addition, in the previous method, average value filters and median filters are used for estimation of the signal value of the processing points. No processing has been included that takes account of the directionality (shape) of image signals such as edge components and detail components. In the present paper, a new noise elimination filter is proposed that attempts to improve restoration accuracy by addressing these two problems. In the proposed filter, first the degraded pixels within the filter window are precompensated for the problem of insufficient original signal pixels in the filter window, and then the signal values at the processing points are estimated by the filter once sufficient signal information is obtained. Preservation of the directionality of the image signals is improved by extending the directional difference filter. Its effectiveness is verified by real image processing results. © 2003 Wiley Periodicals, Inc. Electron Comm Jpn Pt 3, 86(5): 65–75, 2003; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecjc.1149

In this paper, an automatic method for the design of compact acoustic systems, using the genetic algorithm, is proposed for improvement of degraded sensitivity caused by leakage. In the present method, an analysis based on the acoustic equivalent circuit is used. The equivalent circuit and the acoustic parameters of the circuit are designed in such a way that the sound pressure frequency characteristics are contained with the range based on the GSM (ETSI EN 300 903) specifications and that the sensitivity deterioration is improved. The PfGA (Parameter-free Genetic Algorithm) is used in the design of the circuit configuration, and the genetic algorithm of Michalewicz is used for the design of acoustic parameters. As a result, an equivalent circuit is obtained that realizes characteristics not significantly affected by leakage. The present method can be applied to various compact acoustic systems such as cellular phones and provides useful information for actual design. © 2003 Wiley Periodicals, Inc. Electron Comm Jpn Pt 3, 87(3): 1–9, 2004; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecjc.10149

An active noise control (ANC) system is proposed that uses a new frequency domain simultaneous perturbation (FDSP) method automatically controlling the magnitude of perturbation. The proposed method automatically controls the magnitude of perturbation so that the convergence speed can be improved substantially over the conventional case where the magnitude of perturbation is fixed. Specifically, when the error signal is large, the magnitude of the perturbation is made large, while the latter is reduced when the error signal is small. As a result, the present system resolves the problem of instability caused by the block process and also the step size can be set to a larger value. The present control is effective when the magnitude of perturbation needs to be set small. In particular, if the magnitude of perturbation is 0.08, the convergence speed is improved by more than 50 times. Also, the present control is effective for suppression of the noise generated due to addition of perturbation. © 2003 Wiley Periodicals, Inc. Electron Comm Jpn Pt 3, 87(2): 43–53, 2004; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecjc.10112

Since a speaker system has nonlinearity, it is necessary to place a nonlinear inverse system in front of the speaker system so that the nonlinear distortion is corrected. However, in realization with DSP, the amount of computation in the Volterra filter used in the nonlinear inverse system is enormous. There exists a region containing aliasing components in the output signal of the Volterra filter. There is a way to band-limit the input signal in order to avoid aliasing. However, since the sampling frequency is set to twice the Nyquist frequency, an unnecessary band is processed and the efficiency is poor. Hence, in this paper, a configuration is proposed that does not cause aliasing even if the sampling frequency is chosen as the Nyquist frequency in order to reduce the amount of computation in the Volterra filter. Next, an identification method for the speaker system in order to efficiently obtain the Volterra filter used in the proposed configuration is presented. The proposed configuration is applied to the nonlinear inverse system. Finally, the effectiveness of the present method is demonstrated by simulation. © 2002 Wiley Periodicals, Inc. Electron Comm Jpn Pt 3, 86(1): 19–26, 2003; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecjc.10031

In this paper, a formulation of the convergence property of the NLMS method, an updating algorithm for an adaptive Volterra filter, is discussed. To date, much research has been carried out on the convergence property of the updating algorithm in a linear adaptive filter. However, there has been insufficient study of the updating algorithm for the second-order adaptive Volterra filter. In this paper, the convergence property of the E-NLMS method, an updating algorithm in the second-order adaptive Volterra filter, is formulated and the convergence property of the second-order adaptive Volterra filter is theoretically determined. In the formulation of the convergence property we use the first-order recursive filter expression, which is effective for studying various characteristics of the NLMS method in linear adaptive filters. In addition, it is recognized that the nature of the input signal to the diagonal (D) elements of the second-order Volterra system is different from that of the input to the other (ED) components. By means of computer simulation, it is confirmed that the convergence property given by the derived theoretical equation can sufficiently express the actual convergence property. © 2002 Wiley Periodicals, Inc. Electron Comm Jpn Pt 3, 85(10): 40–50, 2002; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecjc.1124

In this paper, a new median filter is proposed that has robustness without relying on the noise generation probability. The proposed filter has a configuration with an impulse detector in the prior stage of the filter. Since this detector has a superior signal estimation capability of the directional difference filter, impulse noise can be detected effectively even if generated at a high generating ratio and as a burst. In the median filter in the later stage, inclusion of the output values of the directional filter and successive updating of the signal values (successive process) are introduced. Hence, while the noise elimination capability of the conventional median filter is retained, the signal preservation at the details and edges can be improved. Therefore, the noise elimination filter proposed in this paper is abundant in robustness against variations of the gray scale, the generation probability, and the impulse adding situation. Its restoration results for images with superposed impulse noise at a high generation probability confirm the improvement of performance. In this paper, the effectiveness of the proposed impulse detector is first demonstrated. It is proven through processing examples that a better image processing result can be obtained from its combination with a newly configured median filter. At the same time, the robustness of the proposed procedure under various conditions is demonstrated. The effectiveness of the proposed method in the final restored image is verified. © 2001 Scripta Technica, Electron Comm Jpn Pt 3, 85(3): 22–32, 2002

In this paper, we propose a directional difference filter that possesses superior noise removal for images corrupted by additive impulse noise and a high level of original signal preservation. By estimating the pixel value of the processing point from directional shape information about the edges and successively updating the estimate, the proposed filter can prevent residual noise even when impulse noise generated at a high probability was added. In addition, this process can perform original signal preservation robust to the detection errors of the noise detector and is extremely useful as a postprocess for a noise detector that is used to remove noise generated at a high probability. In this paper, we demonstrate the robustness of the proposed method to the detection errors generated when using an actual noise detector and use processing examples to verify the superior image processing results over conventional noise removal filters. © 2001 Scripta Technica, Electron Comm Jpn Pt 3, 85(2): 74–82, 2002

In this paper, we propose an active noise control (ANC) system without secondary path model (Ĉ filter). The proposed system is based on the simultaneous perturbation optimization method with block process. Consequently, the coefficients of the adaptive filter in the proposed system are updated by error signals only. The conventional ANC system using the Filtered-x algorithm becomes unstable due to the error between the secondary path from a secondary source to an error sensor and its model. On the other hand, the proposed ANC system has the advantage of becoming stable because of not using the model. In this paper, we show the principle of the proposed ANC system and also the effectiveness of this system by computer simulations. © 2000 Scripta Technica, Electron Comm Jpn Pt 3, 83(10): 47–55, 2000

A problem in the telephone handset is that there is no design formula that can relate the values of the parameters in the acoustic equivalent circuit (acoustic parameters) of the earphone system directly to the shape (dimensions) of the actual handset. As a result, the design of the handset has been made very difficult. to derive such a formula, the values of the acoustic parameters must be estimated accurately from the actual measurement of the handset. Previously, the authors have estimated that acoustic parameters, using the nonlinear optimizatoin technique, but a problem is still left that the number of parameters that can be simultaneously estimated is limited. This paper tries to solve this problem and proposes and estimation technique using the Monte Carlo method, which can also realize a high estimation accuracy. In the proposed method, an allowable range for the actual response is defined and the acoustic parameters are estimated by modifying the upper and the lower limits of each acoustic parameter so that the response stays within the allowable range. The proposed method has the advantages that the number of parameters to be estimated is not limited, a good agreement can be realized with the actual response, and the decision can be made without intuition. the approach also provides a useful means for the design of the whole system. Although this paper discusses the estimation for the piezoelectric telephone handset, the proposed method can be applied effectively to any type of telephone handset.

Until now, no formula has been derived for computing the acoustic parameters of the transmitting and receiving systems in the actual piezoelectric telephone handset of small size and light weight. The lack of such a formula was a significant hindrance to the design of the handset. In this paper, estimation of these parameters is made possible by the application of the nonlinear optimization technique. In this optimization problem, convergence is extremely slow since there is a magnitude difference of about 1010 among the parameters. Hence, for each parameter, a hypothetical maximum and minimum are assumed so that the problem is replaced with a hypothetical conditional optimization problem. By means of the change of variables, the variable ranges of all the parameters are aligned. At the same time, the problem is transformed back to the optimization problem without constraints so that the convergence is accelerated. Since the performance function is constructed in which all the measured characteristics of the transmitting and receiving sensitivities under different measurement conditions are used as esimating information, the resultant optimization problem is of a significant size. The simulation characteristics computed based on the obtained estimation values agreed well with the measured characteristics. This implies that the estimated value derived by the present method is useful in actual design of the handset.