• ETRI Journal
• /
• v.35 no.1
• /
• pp.100-108
• /
• 2013

In this paper, a feature extraction (FE) method is proposed that is comparable to the traditional FE methods used in automatic speech recognition systems. Unlike the conventional spectral-based FE methods, the proposed method evaluates the similarities between an embedded speech signal and a set of predefined speech attractor models in the reconstructed phase space (RPS) domain. In the first step, a set of Gaussian mixture models is trained to represent the speech attractors in the RPS. Next, for a new input speech frame, a posterior-probability-based feature vector is evaluated, which represents the similarity between the embedded frame and the learned speech attractors. We conduct experiments for a speech recognition task utilizing a toolkit based on hidden Markov models, over FARSDAT, a well-known Persian speech corpus. Through the proposed FE method, we gain 3.11% absolute phoneme error rate improvement in comparison to the baseline system, which exploits the mel-frequency cepstral coefficient FE method.

• The Transactions of the Korean Institute of Power Electronics
• /
• v.21 no.4
• /
• pp.343-350
• /
• 2016

Generally, the average phase current is sampled at the midpoint of a PWM signal for the vector control of an AC motor. The three-phase current can also be reconstructed from a DC-link shunt resistor by sampling the shunt voltage during the active vectors of the SVPWM. However, the reconstructed current is different from the average current because of the deviation of the sampling point from the midpoint of the PWM signal. This paper proposes an algorithm to estimate the average current from the reconstructed current in a single-shunt inverter. The proposed method is derived from the phase current slopes based on switching states and corresponding switching time. In addition, the proposed method is generalized for all the six sectors of the space vector hexagon. The validity of the proposed algorithm is verified with simulations and experiments.

• Journal of Power Electronics
• /
• v.12 no.6
• /
• pp.895-903
• /
• 2012

This paper presents a fast switching direct torque control (FS-DTC) using only a single DC-link current sensor. In FS-DTC, six new active voltage complex space vectors (CSVs) are synthesized by the conventional active voltage space vectors (SVs). The corresponding sectors are rotated in the anticlockwise direction by 30 degrees. A selection table is defined to select the CSVs. Based on the "Different Phase Mode", the output sequence of the selected CSV is optimized. Accordingly, a reconstruction method is proposed to acquire the phase currents. The core of the FS-DTC is that all of the three phase currents can be reliably reconstructed during every two sampling periods, which is the result of the fast switching between different phases. The errors between the reconstructed and actual currents are strictly limited in one sampling period. The FS-DTC has the advantages of the standard DTC scheme such as simple structure, quick torque response and robustness. As can be seen in the analysis, the FS-DTC can be thought of as an equivalent standard DTC scheme with 86.6% of the maximum speed, 173.2% of the torque ripple, and 115% of the response time of the torque. Based on a dSPACE DS1103 controller, the FS-DTC is implemented in an induction machine drive system. The results verify the effectiveness of the FS-DTC.

• Journal of Information Processing Systems
• /
• v.13 no.1
• /
• pp.114-127
• /
• 2017

This paper proposes a color image coding algorithm based on shape-adaptive all phase biorthogonal transform (SA-APBT). This algorithm is implemented through four procedures: color space conversion, image segmentation, shape coding, and texture coding. Region-of-interest (ROI) and background area are obtained by image segmentation. Shape coding uses chain code. The texture coding of the ROI is prior to the background area. SA-APBT and uniform quantization are adopted in texture coding. Compared with the color image coding algorithm based on shape-adaptive discrete cosine transform (SA-DCT) at the same bit rates, experimental results on test color images reveal that the objective quality and subjective effects of the reconstructed images using the proposed algorithm are better, especially at low bit rates. Moreover, the complexity of the proposed algorithm is reduced because of uniform quantization.

• 제어로봇시스템학회:학술대회논문집
• /
• 2000.10a
• /
• pp.282-282
• /
• 2000

The analysis of the radiation effect on matter has been performed using stochastic methods. Recently, It was discovered that the detector pulses of radiation can be analysed using deterministic method that utilizes the chaotic behaviour with an attractor found in a noise region. We acquired a time series for pulse tram of Am-241 using scintillation detector and reconstructed a phase space, then performed new analysis for the radiation detection signal by applying embedding theory, Lyapunov exponent, correlation dimension, autocorrelation dimension, and power spectrum.

• Proceedings of the KOSOMBE Conference
• /
• v.1996 no.11
• /
• pp.363-367
• /
• 1996

In general, the physiological systems have shown nonlinear complex phenomena. This study analyzes nonlinear characteristics of the flow of peripheral blood vessel dynamics in physiological systems using chaos theory. We performed this study by means of several quantity methods and power spectrum. The quantity methods are a phase space reconstruction and a poincare's map. And the power spectrum method is a conventional linear analysis. Experimental data have been acquired from examining 10 diabetic patients, and 10 control subjects in initial stable state. In acquisition experminetal data, we anlysized the differences of nonlinear characteristics between diabetic group and control group. The results of quality analysis methods showed the flow of peripheral blood vessel had the nonlinear and chaotic characteristics, screening a strange attractor on reconstructed phase space. In conclusion, the flow dynamics of peripheral blood vessel had a chaotic behavior of nonlinear dynamic systems, dynamic system, and differences of characteristic of nonlinear dynamic system.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.14 no.9
• /
• pp.2099-2104
• /
• 2010

In this paper, HRV signal which was appeared RR intervals from ECG was analyzed using return-map during anesthesia. We intended to estimate the depth of anesthesia observing the change of autonomic nervous activity(ANS) because HRV showed change of cardio-vascular system of the body according to state of ANS. Return-map analysis is to reconstruct time series of HRV to phase space after calculating delay time and embedded time. After approximating the signal distribution which was reconstructed in phase space in elliptic, we calculated the lengths of major and minor axises of the elliptic and the values was used to estimate the depth of anesthesia. Stages of the anesthesia were 7 levels to evaluate the depth of anesthesia. At induction stage of strong external stimulation, the length of major and minor axis were statistically high and at the operation stage of non-external stimulation, the values were statistically low. Conclusively, the stages of anesthesia were discriminated by HRV signal mapped in the phase space during operation.

• Proceedings of the Korean Society of Medical Physics Conference
• /
• 2002.09a
• /
• pp.429-431
• /
• 2002

Metabolic analysis of biological tissues, the interventional radiology in MRT (Magnetic Resonance Treatment) and for clinical diagnoses, representation of 4-Dimensional (4D) structural information (x,y,z,t) of biological tissues is required. This paper discusses image representation techniques for those 4D MR Images. We have proposed an image reconstruction method for ultra-fast 3D MRI. It is based on image interpolation and prediction of un-acquired pictorial data in both of the real and the k-space (the acquisition domain in MRI). A 4D MR image is reconstructed from only two 3D MR images and acquired a few echo signals that are optimized by prediction of the tissue motion. This prediction can be done by the phase of acquired echo signal is proportioned to the tissue motion. On the other hand, reconstructed 4D MR images are represented as a 3D-movie by using computer graphics techniques. Rendered tissue surfaces and/or ROIs are displayed on a CRT monitor. It is represented in an arbitrary plane and/or rendered surface with their motion. As examples of the proposed representation techniques, the finger and the lung motion of healthy volunteers are demonstrated.

• Journal of the Korean Society of Radiology
• /
• v.3 no.2
• /
• pp.23-26
• /
• 2009

This study proposed an improved center array-sequencing phase unwrapping (ICASPU) algorithm. 2% agarose phantom dopped with 0.6mM/l MnCl2 was used with clinical 1.5T MRI system and commercial knee coil. Obtained k-space data(raw-data) was transmitted to PC and reconstructed into phase image with MATLAB software. Previous center array-sequence phase unwrapping algorithm wascompared with proposed ICASP algorithm using second order regression analysis. As a result, we found that the amount of error on proposed ICASPU method is less about 5 times than that of previous CASPU method. In this study, we exploit improved Center array-sequence phase unwrapping algorithm and expect to apply to images including phase informations.

• Proceedings of the KIEE Conference
• /
• 1988.07a
• /
• pp.232-235
• /
• 1988

Recently, an iterative reconstruction-reprojection (IRR) algorithm has been suggested for application to incomplete data computed tomography (CT). In the IRR, the interpolation operation is performed in the image space during reconstruction-reprojection. The errors associated with the interpolation degrade the reconstructed image and may cause divergence unless a large number of rays is used. In this paper, we propose an improved IRR algorithm which eliminates the need for interpolation. The proposed algorithm adopts a new sampling scheme in which samples (projection data) is taken in phase with the samples of the Cartesian grid.