In this paper, we propose a formula to compute the exact autocorrelations of the autoregressive (AR) process. For an arbitrary value of N, we first review the Yule-Walker equation and some basic properties of the AR model. We then modify the Yule-Walker equation to construct a new system of N+1 linear equations that can be used to solve for the N+1 autocorrelation coefficients for lags 0, 1, …, N, provided that the AR parameters of order N and the power of the white noise of the AR process are given.

The pseudomedian filter was designed to be a computationally effecient alternative to the median filter. However, a thorough analysis ofter pseudomedian filter reveals some important differences between its response and that of the median filter. This paper develops some formal properties of the pseudomedian filter. The root signal analysis demonstrates the close relatonship between the median and pseudomedian filters while pointing out important and useful differences.

This paper is a on the neural network to recognize the phonemes, the weight initialization to reduce learning speed, and LR parser for continuous speech recognition. The neural network spots the phonemes in continuous speech and LR parser parses the output of neural network. The whole phonemes recognized in neural network are divided into several groups which are grouped by the similarity of phonemes, and then each group consists of neural network. Each group of neural network to recognize the phonemes consisits of that recognize the phonemes of their own group and VGNN(Verify Group Neural Network) which judges whether the inputs are their own group or not. The weights of neural network are not initialized with random values but initialized from learning data to reduce learning speed. The LR parsing method applied to this paper is not a method which traces a unique path, but one which traces several possible paths because the output of neural network is not accurate. The parser processes the continuous speech frame by frame as accumulating the output of neural network through several possible paths. If this accumulated path-value drops below the threshold value, this path is deleted in possible parsing paths. This paper applies the continuous speech recognition system to the threshold value, this path is deleted in possible parsing paths. This paper applies the continuous speech recognition system to the continuous Korea digits recognition. The recognition rate of isolated digits is 97% in speaker dependent, and 75% in speaker dependent. The recognition rate of continuous digits is 74% in spaker dependent.

A rigorous theory for modeling, analysis, optimum nonseparable filter bank in multidimensional M-band quantized subband codec are developed in this paper. Each pdf-optimized quantizer is modeled by a nonlinear gain-plus-additive uncorrelated noise and embedded into the subband structure. We then decompose the analysis/synthesis filter banks into their polyphase components and shift the down-and up-samplers to the right and left of the analysis/synthesis polyphase matrices respectively. Focusing on the slow clock rate signal between the samplers, we derive the exact expression for the output mean square quantization error by using spatial-invariant analysis. We show that this error can be represented by two uncorrelated components : a distortion component due to the quantizer gain, and a random noise component due to fictitious uncorrelated noise at the uantizer. This mean square error is then minimized subject to perfect reconstruction (PR) constraints and the total bit allocation for the entire filter bank. The algorithm gives filter coefficients and subband bit allocations. Numerical design example for the optimum nonseparable orthonormal filter bank is given with a quincunx subsampling lattice.

This paper provides an approach for representing an optimum vector quantizer by a scalar nonlinear gain-plus-additive noise model. The validity and accuracy of this analytic model is confirmed by comparing the calcuated model quantization errors with actual simulation of the optimum Linde-Buzo-Gray(LBG) vector quantizer. Using this model we frm MSE measure of an M-band filter bank codec in terms of the equivalent scalar quantizatin model and find the optimum FIR filter coefficients for each channel in the M-band structure for a given bit rate, given filter length, and given input signal correlation model. Specific design examples are worked out for 4-tap filters in the two-band paraunitary case. These theoretical results are confirmed by extensive Monte Carlo simulation.

The performance of two-channel adaptive noise canceller is ofter degraded by the weights perturbation due to the speech signal. In this paper, an adaptive noise canceller employing a speech detector and two adaptation algorithms which are switched according to the speech detector is proposed. When highly correlated speech signal is detected, the tap weights of the adaptive filter are adapted by the sign algorithm. On the other hand, the weights are adapted by the NLMS algorithm when silence is detected or when the characteristics of the noise propagation channel is changed. The employed speech detector utilizes the power ratio of the input and the output of an adaptive linear prediction-error filter. According to the computer simulation, the proposed method yields better performance than conventional ones.

From the reverberation signals received in the shallower water, the surface scattered signals are identified by using the multipath eigneray model that provides launch angles, grazing angles and transmission loss from the high frequency directional source to and from the rough surface. For small scale surface waves, the perturbation method is used to compute the backscattering strength for various grazing angles and wind speeds. A scheme to inversely estimate the wind speed, by which the observed surface reverberation levels are produced, has been tested. In result, for low grazing angles the perturbation method can be used to predict the backscattering stregth, thereby the surface wind can be indirectly estimated.

In this paper, an adaptive precompensator, which can reduce the distortion of a Wiener system effectively, is proposed. The previous techniques for adaptive precompensation, based on the Volterra series modeling to compensate the distortion of a nonlinear system, are not suitable for real-time implementation due to high computational burden and slow convergence burden and slow convergence rate. This paper presents an adaptive precompensation technique for the class of nonlinear subsystem, referred to as Wiener system. An adaptive algorithm for adjusting the parameters of a precompensator, structured by a hammerstein model, is derived using the stochastic gradient method. Also, an adaptive precompensatin technique which can effectively reduce nonlinear distortion in μ-law type of saturation characteristics is proposed. The validity of the proposed algorithm is confirmed through simulation by applying it to known Wiener systmes and a typical loudspeaker model.

In this paper, we implement the door lock system based on pattern matching technique for speaker recognition using DTW. In this study, major features of our system are summarized as follows:(1) Make the average reference pattern using DTW. This method keeps the high recognition rate compared with the other systems whose performances degrade rapidly as time goes on. (2) Use F-ratio values of the cepstral coefficients. We find that the weighted cepstral reveals an effect on intensifying the difference between th customer and the imposter. The system hardware is composed of two parts : the door lock part and the speaker recognition processing part. We use an 8051 microprocessor in the door lock park for serial communication with host processor to open or close the lock. Using our system, we obtain speaker recognition rate of about 99.5%.

The squeal vibration of a drum is the major source of brake noise. In this system the binary flutter model of squeal vibration was employed for the drum brake of a passenger car. The vibration analysis of a drum brake was performed by using normal modes, which are obtained by variational method. An improved method for the estimation of shoe modes is proposed and the results are compared with the exact solutions. Numerical results for the coupled system of drum and shoes good agreement with the results of experimental model analysis and those obtained by FE analysis.

The experimental testing results of the large-scale version of a forward-looking ultrasound imaging catheter(FLUIC) are presented, along with the fabrication techniques used, experimental methods, and comparisons of the measured and simulated results. The transducer model is verified by measuring the electrical impedance of the transducer. The pulse width, beamwidth, and the dynamic range for both transmit and pulse-echo response of the fabricated FLUIC are also analyzed. The experimental results conformed its forward-looking imaging capability and the sources of discrepancies between the simulated and experimental beam profiles are addressed.

New very low bit rate segmentation video image coding technique is proposed by segmenting image into textually homogeneous regions. Regions are classified into one of three perceptually distinct texture classes(perceived constant intensity, smooth texture, and rough texture) using the Human Visual System(HVS) and the fractals. To design very low bit rate video image coder, it is very important to determine the best block size for estimation the fractal dimension and the thresholding of the fractal dimension for each texture class. Good quality reconstructed images are obtained with about 0.10 to 0.21 bit per pixel(bpp) for many different types of imagery.