• 전자공학회논문지CI
• /
• 제40권6호
• /
• pp.95-105
• /
• 2003

본 논문에서는 지문인증 시스템에서 인증 성능을 향상시키기 위한 기준점 검출 알고리즘과 특징 추출에 있어서 새로운 filterbank방법을 제안한다. 제안한 기준점 검출 알고리즘 GPM(Gradient Probabilistic Method)은 4개의 방향성분을 추출하여 방향성분을 가장 균일하게 가지는 지점을 검출하는 방법이며, 기존의 Poincare index방법과 달리 수학적 통계적 방법을 사용하기 때문에 지문의 융선에 대한 세부적이고 세밀한 전처리 과정이 불필요하며, arch형태 지문의 기준점 검출에 대한 단점을 해결한다. 또한, 제안한 filterbank방법은 기존filterbank방법에서 특징의 불균일한 분포로 생기는 단점을 균일한 분포로 만들어 추출함으로써 해결한다. 제안한 GPM의 실험결과 기존의 Poincare index방법에 비해서, 일반환경뿐 아니라 잡음환경에서의 특징 추출 시간과 인증률에서 우수함을 보여준다. 특히, 제안한 GPM은 Poincare index방법에 비해서, arch type의 지문에 대한 FAR은 일반 환경에서 49%, 밝기 잡음환경에서 39.2%, salt and pepper 잡음환경에서 15.7%의 향상을 보여준다. 또한, 기준점 검출시간에 있어서, 제안한 GPM방법은 기존의 Poincare index방법보다 0.07초의 감소를 보여주며, 특징추출 시간에 있어서도 제안한 filterbank 알고리즘은 기존의 filterbank 방법에 비해서 0.06sec의 감소를 보여준다.

• The Journal of the Acoustical Society of Korea
• /
• 제20권1E호
• /
• pp.3-6
• /
• 2001

Sound pressure level (SPL) measurements were performed on a controlled test track vehicle coast-by runs of a passenger vehicle with six different sets of tires across a range of temperatures. A small but significant reduction of noise level with positive temperature increases was observed for some tires. The temperature gradient of the different tires at 80㎞/h ranged from -0.07 to + 0.01 dB/℃. Frequency analysis of the tire noise identified that noise content in the range of 1,300 to 1,900Hz was particularly sensitive to temperature changes. Differences in SPL due to speed and tire type were much greater than that due to temperature.

• Geomechanics and Engineering
• /
• 제8권1호
• /
• pp.33-52
• /
• 2015

Based on the Mohr-Coulomb failure criterion, a gradient-dependent plastic model that considers the strain-softening behavior is presented in this study. Both triaxial shear tests on conventional specimen and precut-specimen, which were obtained from an ancient landslide, are performed to plot the post-peak stress-strain entire-process curves. According to the test results of the soil strength, which reduces from peak to residual strength, the Mohr-Coulomb criterion that considers strain-softening under gradient plastic theory is deduced, where strength reduction depends on the hardening parameter and the Laplacian thereof. The validity of the model is evaluated by the simulation of the results of triaxial shear test, and the computed and measured curves are consistent and independent of the adopted mesh. Finally, a progressive failure of the ancient landslide, which was triggered by slide of the toe, is simulated using this model, and the effects of the strain-softening process on the landslide stability are discussed.

• 대한의용생체공학회:학술대회논문집
• /
• 대한의용생체공학회 1990년도 추계학술대회
• /
• pp.19-23
• /
• 1990

A new technique for reducing the susceptibility artifact in gradient echo imaging which uses a tailored RF pulse is proposed. It is applied to the case of imaging where artifacts and distortions arise due to the high local magnetic field inhomogeneity i. e., the susceptibility. The signal loss and void phenomena due to susceptibility in a voxel are studied and a correction method using a tailored RF pulse is proposed. Applications of this method in imaging are given and experimental results obtained using an human volunteer with a 2.0 T KAIS NMR system are presented.

• 한국지능시스템학회논문지
• /
• 제5권4호
• /
• pp.33-40
• /
• 1995

This paper delas with the detection of freehand forgeries of signatures based on the averaged directional amplitudes of gradient vetor which are related to the overall shape of the handwritten signature and fuzzy ARTMAP neural network classifier. In the first step, signature images are extracted from the background by a process involving noise reduction and automatic thresholding. Next, twelve directional amplitudes of gradient vector for each pixel on the signature line are measure and averaged through the entire signature image. With these twelve averaged directional gradient amplitudes, the fuzzy ARTMAP neural network is trained and tested for the detection of freehand forgeries of singatures. The experimental results show that the fuzzy ARTMAP neural network cna lcassify a signature whether genuine or forged with greater than 95% overall accuracy.

• 대한전기학회논문지
• /
• 제39권1호
• /
• pp.22-28
• /
• 1990

This paper is a study on the reduction of computation time in case of nonlinear magnetic field analysis by finite element method and Newton-Raphson method. For the purpose, the nonlinear convergence equation is computed by the conjugate gradient method which is known to be applicable to symmetric positive definite matrix equations only. As the results, we can not prove mathematically that the system Jacobian is positive definite, but when we applied this method, the diverging case did not occur. And the computation time is reduced by 25-55% and 15-45% in comparison with the case of direct and successive over-relaxation method, respectively. Therefore, we proved the utility of conjugate gradient method.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 1994년도 추계학술대회 논문집 학회본부
• /
• pp.156-158
• /
• 1994

In recent days, the various adjustable speed drives are widely employed at the industrial applications for the purpose of energy saving and speed control. In particular, for the machine control applications. the switching frequency is required to be increased for better dynamic performance of the drive. Moreover, this also leads to the reduction of the switching loss of the device. For IGBT (Insulated Gate Bipolar Transistor), the most widely used switching device in the inverters below the 100[kW] range, the falling and falling time is of the order about $200{\sim}300[ns]$. Therefore unexpected phenomena occurs such as voltage spikes due to high gradient of current at the switching instant, the weakening of motor insulation due to high gradient of voltage. In this paper, a new voltage gradient suppression technique is presented in both theoretically and experimentally.

• Smart Structures and Systems
• /
• 제23권1호
• /
• pp.45-60
• /
• 2019

This paper is devoted to proposing a new approach for damage detection of structures. In this technique, the biconjugate gradient method (BCG) is employed. To remedy the noise effects, a new preconditioning algorithm is applied. The proposed preconditioner matrix significantly reduces the condition number of the system. Moreover, based on the characteristics of the damage vector, a new direct search algorithm is employed to increase the efficiency of the suggested damage detection scheme by reducing the number of unknowns. To corroborate the high efficiency and capability of the presented strategy, it is applied for estimating the severity and location of damage in the well-known 31-member and 52-member trusses. For damage detection of these trusses, the time history responses are measured by a limited number of sensors. The results of numerical examples reveal high accuracy and robustness of the proposed method.

• 한국정보전자통신기술학회논문지
• /
• 제10권2호
• /
• pp.141-146
• /
• 2017

본 논문에서는 저전력 영상 특징 추출 하드웨어 설계를 위한 하드웨어 폴딩 기법 기반 저면적 Gradient magnitude 연산기 구조를 제안한다. 하드웨어 복잡도를 줄이기 위해 Gradient magnitude 벡터의 특징을 분석하여 기존 알고리즘을 하드웨어를 공유하여 사용할 수 있는 알고리즘으로 변경하여 Folding 구조가 적용될 수 있도록 했다. 제안된 하드웨어 구조는 기존 알고리즘의 특징을 최대한 이용했기 때문에 데이터 품질의 열화가 거의 없이 구현될 수 있다. 제안된 하드웨어 구조는 Altera Quartus II v16.0 환경에서 Altera Cyclone VI (EP4CE115F29C7N) FPGA를 이용하여 구현되었다. 구현 결과, 기존 하드웨어 구조를 이용하여 구현한 연산기와의 비교에서 41%의 logic elements, 62%의 embedded multiplier 절감 효과가 있음을 확인했다.

• 말소리와 음성과학
• /
• 제7권1호
• /
• pp.151-163
• /
• 2015

In this study, we aim to describe kinematic characteristics of Korean /l/-flapping in two speech rates (fast vs. comfortable). Production data was collected from seven native speakers of Seoul Korean (four females and three males) using electromagnetic midsagittal articulometry (EMMA), which provided two dimensional data on the x-y plane. We examined kinematic properties of the vertical/horizontal tongue tip gesture, the vertical/horizontal (rear) tongue body gesture, and the jaw gesture in an /i/-/i/ context. Gestural landmarks of the vertical tongue tip gesture are directly measured. This serves as the actual anchoring time points to which relevant measures of other trajectories referred. The study focuses on velocity profiles, closing/opening spatiotemporal properties, constriction duration, and constriction minima were analyzed. The results are summarized as follows. First, gradiently distributed spatiotemporal values of the vertical tongue tip gesture were on a continuum. This shows more of a reduction in fast speech rate, but no single instance of categorical reduction (deletion). Second, Korean /l/-flapping predominantly exhibited a backward sliding tongue tip movement, in 83% of production, which is apparently distinguished from forward sliding movement in English. Lastly, there was an indication of vocalic reduction in fast rate, truncating spatial displacement of the jaw and the tongue body, although we did not observe positional variations with speech rate. The present study shows that Korean /l/-flapping is characterized by mixed articulatory properties with respect to flapping sounds of other languages such as English and Xiangxiang Chinese. Korean /l/ flapping demonstrates a language-universal property, such as the gradient nature of its flapping sounds that is compatible with other languages. On the other hand, Korean /l/-flapping also shows a language-particular property, particularly distinguished from English, in that a backward gliding movement occurs during the tongue tip closing movement. Although, there was no vocalic reduction in V2 observed in terms of jaw and tongue body height, spatial displacement of these articulators still suggests truncation in fast speech rate.