• The Journal of the Acoustical Society of Korea
• /
• 제22권2E호
• /
• pp.62-68
• /
• 2003

This paper addresses a novel noise-compensation scheme to solve the mismatch problem between training and testing condition for the automatic speech recognition (ASR) system, specifically in car environment. The conventional spectral subtraction schemes rely on the signal-to-noise ratio (SNR) such that attenuation is imposed on that part of the spectrum that appears to have low SNR, and accentuation is made on that part of high SNR. However, these schemes are based on the postulation that the power spectrum of noise is in general at the lower level in magnitude than that of speech. Therefore, while such postulation is adequate for high SNR environment, it is grossly inadequate for low SNR scenarios such as that of car environment. This paper proposes an efficient spectral subtraction scheme focused specifically to low SNR noisy environment by extracting harmonics distinctively in speech spectrum. Representative experiments confirm the superior performance of the proposed method over conventional methods. The experiments are conducted using car noise-corrupted utterances of Aurora2 corpus.

• 한국군사과학기술학회지
• /
• 제14권6호
• /
• pp.1073-1080
• /
• 2011

In this study, a target detection algorithm was proposed for using hyperspectral imagery. The proposed algorithm is designed to have minimal processing time, low false alarm rate, and flexible threshold selection. The target detection procedure can be divided into two steps. Initially, candidates of target pixel are extracted using matching ratio of spectral pattern that can be calculated by spectral derivation. Secondly, spectral distance is computed only for those candidates using Euclidean distance. The proposed two-step method showed lower false alarm rate than the Euclidean distance detector applied over the whole image. It also showed much lower processing time as compared to the Mahalanobis distance detector.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 1988년도 전기.전자공학 학술대회 논문집
• /
• pp.20-24
• /
• 1988

This paper proposed the phonemic segmentation method that employed two types of Likelihood Ratio that measures the change of spectral structure. By this method, isolated digits and words of VCV form are segmented into phoneme-unit and especially, first-burst part in an aspirated bilabial plosive is divided.

• Nuclear Engineering and Technology
• /
• 제49권6호
• /
• pp.1287-1300
• /
• 2017

Based on the observation that ignoring the angle dependency of multigroup resonance cross sections within a fuel pellet would result in nontrivial underestimation of the spatial self-shielding of flux, a parametrized spectral superhomogenization (SPH) factor library (PSSL) method is developed as a practical means of resolving the problem. Region-wise spectral SPH factors are calculated by the normal and transport corrected SPH iterations after ultrafine group slowing down calculations over various light water reactor pin-cell configurations. The parametrization is done with fuel temperature, U-238 number density, fuel radius, moderator source represented by ${\Sigma}_{mod}V_{mod}$, and the number density ratio of resonance nuclides to that of U-238 in a form of resonance interference correction factors. The parametrization is successful in that the root mean square errors of the interpolated SPH factors over the fuel regions of various pin-cells are within 0.1%. The improvement in reactivity error of the PSSL method is shown to be superior to that by the original SPH method in that the reactivity bias of -200 pcm to -300 pcm vanishes almost completely. It is demonstrated that the environment effect takes only about 4% in the reactivity improvement so that the pin-cell based PSSL method is effective in the assembly problems.

• Bulletin of the Korean Chemical Society
• /
• 제22권1호
• /
• pp.37-42
• /
• 2001

Ridge regression is compared with multiple linear regression (MLR) for determination of Research Octane Number (RON) when the baseline and signal-to-noise ratio are varied. MLR analysis of near-infrared (NIR) spectroscopic data usually encounters a collinearity problem, which adversely affects long-term prediction performance. The collinearity problem can be eliminated or greatly improved by using ridge regression, which is a biased estimation method. To evaluate the robustness of each calibration, the calibration models developed by both calibration methods were used to predict RONs of gasoline spectra in which the baseline and signal-to-noise ratio were varied. The prediction results of a ridge calibration model showed more stable prediction performance as compared to that of MLR, especially when the spectral baselines were varied. . In conclusion, ridge regression is shown to be a viable method for calibration of RON with the NIR data when only a few wavelengths are available such as hand-carry device using a few diodes.

• 한국멀티미디어학회논문지
• /
• 제4권4호
• /
• pp.306-315
• /
• 2001

잡음환경에서의 음성인식을 위하여 음성에 부가된 잡음을 제거하는 방법에 있어, 기존의 스펙트럼 차감법은 잡음과 음성을 정확히 구별하기 힘들고 정확한 잡음의 특성을 추정할 수 없는 단점이 있다. 또한 웨이브렛 변환영역에서의 잡음제거 방법은 임계값 적용시 저주파 영역보다는 고주파영역에 상대적으로 더 큰 영향을 미쳐 고주파영역에서 신호의 손실이 발생하는 단점이 있다. 본 논문에서는 스펙트럼 차감법 및 웨이브렛 변환을 이용한 잡음제거 방법의 단점을 개선하기 위하여 연속 웨이브렛 변환 영역에서 웨이브렛 계수의 스케일별 표준편차로 묵음구간과 음성 구간을 판별하여 끝점을 검출 후, 잡음이 섞인 음성신호를 이산 웨이브렛 변화에 의해 3개의 대역으로 분리하여 각각의 대역 내에서 스펙트럼 차감법을 적용시키는 방법을 제안한다. 끝점을 검출하고 대역을 나눔으로써 스펙트럼 차감을 적응할 잡음 신호의 특성을 정확히 추출할 수 있다. 실험을 통하여 제안한 방법이 기존의 스펙트럼 차감법 및 웨이브렛 변환을 이용한 잡음제거 방법보다 신호대 잡음비 및 Itakura-Saito거리 측면에서 향상됨을 확인할 수 있었다.

• Wind and Structures
• /
• 제19권4호
• /
• pp.353-370
• /
• 2014

A new method is proposed in this study for estimating the damping ratio of a super tall building under strong wind loads with short-time measured acceleration signals. This method incorporates two main steps. Firstly, the power spectral density of wind-induced acceleration response is obtained by the wavelet transform, then the dynamic characteristics including the natural frequency and damping ratio for the first vibration mode are estimated by a nonlinear regression analysis on the power spectral density. A numerical simulation illustrated that the damping ratios identified by the wavelet spectrum are superior in precision and stability to those values obtained from Welch's periodogram spectrum. To verify the efficiency of the proposed method, wind-induced acceleration responses of the Guangzhou West Tower (GZWT) measured in the field during Typhoon Usagi, which affected this building on September 22, 2013, were used. The damping ratios identified varied from 0.38% to 0.61% in direction 1 and from 0.22% to 0.59% in direction 2. This information is expected to be of considerable interest and practical use for engineers and researchers involved in the wind-resistant design of super-tall buildings.

• 한국지구물리탐사학회:학술대회논문집
• /
• 한국지구물리탐사학회 2009년도 학술대회 초록집
• /
• pp.60-61
• /
• 2009

• 한국정보통신학회논문지
• /
• 제5권3호
• /
• pp.443-449
• /
• 2001

잡음환경에서의 음성인식은 실제의 환경에서의 음성인식에서 매우 중요한 애로기술로써 이를 해결하기 위한 연구는 꾸준히 연구되고 있다. 따라서 본 연구는 음성인식분야에서 가장 많이 사용하고 있는 HMM처리 시잡음처리의 문제점을 주파수 가중치 부가 HMM으로 해결하는 방법을 제안하고 그 성능을 인식실험을 통하여 검토하였다. 그 결과 SS처리를 함께 사용하는 $MCE-\mu$, MCE-$\rho$가 가장 잡음에 강한 방식임을 알 수 있었다.

• 한국인쇄학회지
• /
• 제18권2호
• /
• pp.1-12
• /
• 2000

This paper proposes a new algorithm classifing same hues in order toe estimate the spectral reflectance of object from 3 band color image information. To estimate the spectral reflectance of object, the conventional estimation methods are required of 5 or 9 band digital color values. The 5 or 9 band image acquisition systems are required of 5 or 3 times same work for color image acquisition process. To solve the above problems, we propose a new method that can be estimated spectra reflectance estimation of object. The proposed method is to classify same hues corresponding a color stimulus, by using hue angle and chroma vector of a color stimulus. The classified same hues are used as the population corresponding a color stimulus. The range of same hue is estimated by the cumulative proportional ration according to the number of basis function.