• Proceedings of the Korea Water Resources Association Conference
• /
• 2004.05b
• /
• pp.1331-1334
• /
• 2004

Wet/Dry phenomena typically incorporate a number of complex flow mechanism. These include a momentum transfer and turbulent mixing caused by the delivery of water. However currently available one dimensional schemes applicable to wet/dry process cannot effectively simulate such process. Two dimensional finite element model, SU/PG, is used to simulate complex flow in this study. The Wetted Area Method in SU/PG allows elements to transition gradually between wet and dry states. The model is applicable to a straight river reach with irregular bathymetry. Wet/dry calculation using the wetted area method can simulate simple numerical test. The computed results of velocity vectors and water depth agree with those of observed. The methodology Presented in this study will contributed to two-dimensional wet/dry analysis in a river in this country.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2007.06a
• /
• pp.153-156
• /
• 2007

In this paper, we propose a fast face detection approach using PCA and SVM. In our detection system, first we filter the face potential area using statistical feature which is generated by analyzing local histogram distribution. And then, we use SVM classifier to detect whether there are faces present in the test image. Support Vector Machine (SVM) has great performance in classification task. PCA is used for dimension reduction of sample data. After PCA transform, the feature vectors, which are used for training SVM classifier, are generated. Our tests in this paper are based on CMU face database.

• Magazine of the Korean Society of Agricultural Engineers
• /
• v.37 no.2
• /
• pp.64-72
• /
• 1995

A series of cubical triaxial tests with three independent principal stresses was per- formed on Baekma river sand( # 40~100). It was found that the major principal strain at failure remained approximately constant for b values larger than about 0.3 for both the drained and undrained condition, and thereafter increased as b value decreased. The test results showed that the direction of the strain increment at failure form acute angles with the failure surfaces for both the drained and undrained condition. The results were thus not in agreement with the normality condition from classic plasticity theory. Howev- er, it was found that the projections of the plastic strain increment vectors on the octahe- dral plane were perpendicular to the failure surface in that plane. Failure strength in terms of effective stress anlaysis was greatly influenced by the variation of intermediate principal stress and so was failure criterion. The effective stress failure surfaces for both the drained and undrained condition were estimated quite well by use of Lade's failure criterion.

• Speech Sciences
• /
• v.13 no.4
• /
• pp.177-186
• /
• 2006

In an effort to enhance the quality of feature vector classification and thereby reduce the recognition error rate of the speaker-independent speech recognition, we employ the Mahalanobis distance in the calculation of the similarity measure between feature vectors. It is assumed that the metric matrix of the Mahalanobis distance be diagonal for the sake of cost reduction in memory and time of calculation. We propose that the diagonal elements be given in terms of the variations of the feature vector components. Geometrically, this prescription tends to redistribute the set of data in the shape of a hypersphere in the feature vector space. The idea is applied to the speech recognition by hidden Markov model with fuzzy vector quantization. The result shows that the recognition is improved by an appropriate choice of the relevant adjustable parameter. The Viterbi score difference of the two winners in the recognition test shows that the general behavior is in accord with that of the recognition error rate.

• Phonetics and Speech Sciences
• /
• v.4 no.2
• /
• pp.79-86
• /
• 2012

In this paper, we propose a novel approach to improve histogram equalization for speaker identification. Our method collects all speech features of UBM training data to make a reference distribution. The ranks of the feature vectors are calculated in the sorted list of the collection of the UBM training data and the test data. We use the ranks to perform order-based histogram equalization. The proposed method improves the accuracy of the speaker recognition system with short utterances. We use four kinds of speech databases to evaluate the proposed speaker recognition system and compare the system with cepstral mean normalization (CMN), mean and variance normalization (MVN), and histogram equalization (HEQ). Our system reduced the relative error rate by 33.3% from the baseline system.

• Nuclear Engineering and Technology
• /
• v.30 no.4
• /
• pp.287-297
• /
• 1998

The elastic stiffness formula of leaf type holddown spring(HDS) assembly is verified by comparing the values of elastic stiffness with the characteristic test results of the HDS's specimens. The comparisons show that the derived elastic stiffness formula is useful in reliably estimating the elastic stiffness of leaf type HDS assembly. The elastic stiffness sensitivity of leaf type HDS assembly is analyzed using the formula and its gradient vectors obtained from the mid-point formula. As a result of sensitivity analysis, the elastic stiffness sensitivity with respect to each design variable is quantified and design variables of large sensitivity are identified. Among the design variables, leaf thickness is identified as the most sensitive design variable to the elastic stiffness of leaf type HDS assembly. In addition, the elastic stiffness sensitivity, with respect to design variable, is in power-law type correlation to the base thickness of the leaf.

• Transactions of Materials Processing
• /
• v.9 no.6
• /
• pp.590-597
• /
• 2000

A finite element formulation lot the simulation of axisymmetric sheet hydroforming is proposed, and an implicit program is coded. In order to describe normal anisotropy of steel sheet, Hill's non-quadratic yield function (Hill, 1979) is employed. Frictional contacts among sheet surface, rigid tool surface, and flexible hydrostatic pressure are considered using mesh normal vectors based on finite element of the sheet. Applied hydraulic pressure is also considered as a function of forming rate and time and treated as an external loading. The complete set of the governing relations comprising equilibrium and interfacial equations is approximately linearized for Newton-Raphson algorithm. In order to verify the validity of the developed finite element formulation, the axisymmetric bulge test is simulated. Simulation results are compared with other FEM results and experimental measurements and showed good agreements. In axisymmetric hydroforming processes of a disk cover, formability changes are observed according to the hydraulic pressure curve changes.

• Proceedings of the KOSOMBE Conference
• /
• v.1995 no.05
• /
• pp.37-41
• /
• 1995

In this paper, we proposed the 2-stage ultrasound liver image classifier which uses the fractal dimensions obtained from the original image and its 1/2 subsampled image, and the Normalized Fourier Power Spectrum. The fractal dimension based on Fractional Brownian Motion (FBM) is calculated from the variance of the same scale pixels instead of the mean of them. Since the actual ultrasound. liver images does not fully match the FBM, to get the fractal dimension, we use the scale vectors which satisfy the FBM model. In 2-stage classifier, we first classified normal and diffuse liver and then classified the fat liver and cirrhosis from the diffuse liver. For the test liver images. 70% of normal liver and 80% of fat liver and 90% of cirrhosis is classified classified with our 2-stage classifier.

• ETRI Journal
• /
• v.39 no.6
• /
• pp.841-850
• /
• 2017

We investigate neural network image reconstruction for magnetic particle imaging. The network performance strongly depends on the convolution effects of the spectrum input data. The larger convolution effect appearing at a relatively smaller nanoparticle size obstructs the network training. The trained single-layer network reveals the weighting matrix consisting of a basis vector in the form of Chebyshev polynomials of the second kind. The weighting matrix corresponds to an inverse system matrix, where an incoherency of basis vectors due to low convolution effects, as well as a nonlinear activation function, plays a key role in retrieving the matrix elements. Test images are well reconstructed through trained networks having an inverse kernel matrix. We also confirm that a multi-layer network with one hidden layer improves the performance. Based on the results, a neural network architecture overcoming the low incoherence of the inverse kernel through the classification property is expected to become a better tool for image reconstruction.

• Proceedings of the IEEK Conference
• /
• 2000.07a
• /
• pp.351-354
• /
• 2000

In the past, we proposed an IDDQ testable design method for static CMOS PLA circuits. All bridging faults can be detected in NOR planes of our testable designed PLA circuits by IDDQ testing with 4 kinds of test input vectors which are independent of the logical functions to be realized. However, the testable designed PLA circuits consume large power in the normal operation. In this paper, a new IDDQ testable design method is proposed and evaluated by some experiments. The experimental results show that the PLA circuit designed with our method can work with low power consumption than the previous one.