• Journal of the Korea Organic Resources Recycling Association
• /
• v.5 no.2
• /
• pp.29-37
• /
• 1997

Because food waste have high moisture content, landfill or incineration is not most suitable. The reuse of food waste by composting contribute to solve resourse, environmental, aglicultural problems. The purpose of this research is to optimise operating conditions and to develop new microorganisms for recycling of food waste by ultilizing small-scale composting equipment. The reduction rate of food waste was 75~85% by weight, and it was coincided with moisture reduction. When initial moisture was 25%, 45% and 60%, C/N ratio on based net weight was 9.8, 10.7 and 11. 8, respectively. And it was suitable for composting. But, The developed microorganism, PNU2, was better than existing commercial seed in the activity based on $CO_2$ concentration.

• Transactions of the Korean Society of Mechanical Engineers B
• /
• v.39 no.5
• /
• pp.415-423
• /
• 2015

In this study, we present a grid refinement model in the lattice Boltzmann method (LBM) for two-dimensional incompressible fluid flow. That is, the model combines the desirable features of the lattice Boltzmann method and stream function-vorticity formulations. In order to obtain an accurate result, very fine grid (or lattice) is required near the solid boundary. Therefore, the grid refinement model is used in the lattice Boltzmann method for stream function-vorticity formulation. This approach is more efficient in that it can obtain the same accurate solution as that in single-block approach even if few lattices are used for computation. In order to validate the grid refinement approach for the stream function-vorticity formulation, the numerical simulations of lid-driven cavity flows were performed and good results were obtained.

• Proceedings of the Korea Contents Association Conference
• /
• 2005.05a
• /
• pp.341-345
• /
• 2005

In this paper, we propose an algorithm for the approximation of FIR filters by IIR filters. The algorithm is based on a concept of the balanced model reduction. Head-related transfer functions(HRTFs) of dummy-head are approximated by 32-order IIR filters. The binaural sounds using the approximated HRTFs are reproduced by headphone, and serves as a cue of sound image localization. Experiment of sound image are carried out for 10 participants with computer simulation and DSP board respectively. The results of the experiments show that the localization using the approximated HRTFs by IIR filters is the same accuracy as the case of FIR filters that simulate the HRTFs.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.7 no.7
• /
• pp.1380-1386
• /
• 2003

Speech recognition performance is degraded by the variation in vocal tract length among speakers. In this paper, we have used a vocal tract length normalization method wherein the frequency axis of the short-time spectrum associated with a speaker's speech is scaled to minimize the effects of speaker's vocal tract length on the speech recognition performance In order to normalize vocal tract length, we tried several frequency warping functions such as linear and piece-wise linear function. Variable interval piece-wise linear warping function is proposed to effectively model the variation of frequency axis scale due to the large variation of vocal tract length. Experimental results on TIDIGITS connected digits showed the dramatic reduction of word error rates from 2.15% to 0.53% by the proposed vocal tract normalization.

• Journal of Korea Multimedia Society
• /
• v.11 no.4
• /
• pp.462-470
• /
• 2008

This study applies fuzzy functions to improve image quality under the assumption that uncertainty of image information due to low contrast is based on vagueness and ambiguity of the brightness pixel values. To solve the problem of low contrast images whose brightness distribution is inclined, we use the k-means algorithm as a parameter of the fuzzy function, through which automatic critical points can be found to differentiate objects from background and contrast between bright and dark points can be improved. The fuzzy function is presented at the three main stages presented to improve image quality: fuzzification, contrast enhancement and defuzzification. To measure improved image quality, we present the fuzzy index and entropy index and in comparison with those of histogram equalization technique, it shows outstanding performance.

• KIPS Transactions on Software and Data Engineering
• /
• v.9 no.10
• /
• pp.303-308
• /
• 2020

The ReLU(Rectified Linear Unit) function has been dominantly used as a standard activation function in most deep artificial neural network models since it was proposed. Later, Leaky ReLU, Swish, and Mish activation functions were presented to replace ReLU, which showed improved performance over existing ReLU function in image classification task. Therefore, we recognized the need to experiment with whether performance improvements could be achieved by replacing the RELU with other activation functions in the super resolution task. In this paper, the performance was compared by changing the activation functions in EDSR model, which showed stable performance in the super resolution task. As a result, in experiments conducted with changing the activation function of EDSR, when the resolution was converted to double, the existing activation function, ReLU, showed similar or higher performance than the other activation functions used in the experiment. When the resolution was converted to four times, Leaky ReLU and Swish function showed slightly improved performance over ReLU. PSNR and SSIM, which can quantitatively evaluate the quality of images, were able to identify average performance improvements of 0.06%, 0.05% when using Leaky ReLU, and average performance improvements of 0.06% and 0.03% when using Swish. When the resolution is converted to eight times, the Mish function shows a slight average performance improvement over the ReLU. Using Mish, PSNR and SSIM were able to identify an average of 0.06% and 0.02% performance improvement over the RELU. In conclusion, Leaky ReLU and Swish showed improved performance compared to ReLU for super resolution that converts resolution four times and Mish showed improved performance compared to ReLU for super resolution that converts resolution eight times. In future study, we should conduct comparative experiments to replace activation functions with Leaky ReLU, Swish and Mish to improve performance in other super resolution models.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2012.11a
• /
• pp.254-255
• /
• 2012

임의의 위치에 음상을 정위하는 기본적인 방법으로, 머리전달함수(HRTF : Head Related Transfer Function)을 원음에 convolution하는 기법이 사용된다. 또한 머리전달함수의 주파수 영역에서의 보간을 이용하여 이동음 효과를 생성할 수 있다. 하지만 더미헤드를 이용하여, 각각의 고도각과 방위각에서 측정된 머리전달함수는 사용자에 따라 정위감을 저하시킬 수 있으며, 클릭음 발생, 보간 특성의 변화등의 문제점이 발생하게 된다. 또한 좌표로 표현되는 영상과는 달리, 소리는 들려오는 방향을 정확한 좌표로 표현하기가 힘들다. 이에 본 논문에서는, 이동음 궤적상의 머리전달함수와 그 주변의 머리전달함수를 선형 보간법으로 그룹화하여 생성된 새로운 머리전달함수를 이용하여 이동음을 구현하는 방법을 제안한다. 또한, 생성된 머리전달함수를 이용하여 구현한 이동음의 청감평가를 수행하여 개선 효과를 확인하였다.

• Proceedings of the Korean Information Science Society Conference
• /
• 2000.10a
• /
• pp.284-286
• /
• 2000

시간에 따라 변화하는 자료들을 저장하는 시간지원 데이터베이스에서 집단 함수는 시간지원 그룹화를 통하여 집단 함수 값이 변하지 않는 시간 구간을 구하고 그 각각의 구간마다 집단 함수의 결과를 생성해야 하는 복잡한 연산이다. 기존의 시간지원 집단 함수 처리 기법들은 집단 함수를 포함하는 질의가 요구되었을 때, 불변 구간을 구하기 위해 트리와 같은 자료구조를 생성하고 이 트리의 노드들을 순회함으로써 집단 함수의 결과를 생성하였다. 이 논문에서는 미리 데이터베이스를 한 번 스캔하여 투플의 시작 시간과 종료 시간들의 정렬된 집합인 시점 시퀀스를 생성하고, 이를 이용하여 시간지원 집단 함수를 처리하는 방법을 제안한다. 또한 데이터베이스에서 저장된 데이터의 삭제나 새로운 데이터의 삽입에 따른 시점 시퀀스의 갱신 방법도 제시한다.

• Proceedings of the Korean Information Science Society Conference
• /
• 2000.04b
• /
• pp.475-477
• /
• 2000

본 논문에서는 적응적 성분분석 기법을 이용하여 radial basis 함수 신경망의 학습시간과 분류성능을 개선한 새로운 기법을 제안하였다. 제안된 기법에서 적응적 성분분석 기법은 radial basis 함수 신경망의 은닉층 뉴런 개수와 중심값 설정을 위해 이용하였다. 제안된 기법의 radial basis 함수 신경망을 200명의 암환자를 2부류(초기와 악성)로 분류하는 문제에 적용하여 시뮬레이션한 결고, k-평균 군집화 알고리즘을 이용한 radial basis 함수 신경망과 비교할 때 학습시간과 시험 데이터의 분류에서 더욱 우수한 성능이 있음을 확인할 수 있었다.

• Proceedings of the Korean Vacuum Society Conference
• /
• 2010.08a
• /
• pp.225-225
• /
• 2010

적외선 영역에서의 밴드갭 에너지를 가지고 있는 III-V 족 화합물 반도체 물질인 $InAs_xSb_{1-x}$는 좋은 성장 안정성과 높은 전자, 홀 이동도를 가지며, 제작 비용이 적게 드는 등 적외선 광소자 제작에 많은 이점을 가지고 있기 때문에 그에 관한 연구가 최근 활발히 진행 되고 있다. 하지만 이러한 $InAs_xSb_{1-x}$를 소자 제작에 이용하기 위해서는 임의의 As 함량에 따른 InAsSb의 물질의 광학적 특성 정보가 필요하다. 본 연구에서는 1.5~6.0 eV 에너지 구간에서 $InAs_xSb_{1-x}$ ($0{\leq}x{\leq}1$) 화합물의 임의의 As 함량에 따른 유전함수를 분석하고 그 분석 변수들을 보고하고자 한다. 기성박막층착장치 (molecular beam epitaxy)를 이용하여 GaAs 기판 위에 성장 시킨 $InAs_xSb_{1-x}$ (x = 0.000, 0.127, 0.337, 0.491, 0.726, 1.000) 박막의 순수한 유전함수 $\varepsilon$을 화학적 에칭을 통해 산화막 층을 제거하여 타원편광분석법을 이용하여 얻었다. 측정된 유전율 함수는 Gaussian-broadened polynomial 들의 합으로서 반도체 물질의 유전함수를 정확히 기술하는 변수화 모델을 이용하여 재현하였다. 변수화 모델을 통해 얻어진 각각의 변수들을 As 조성비 x 에 대한 다항식으로 피팅하여 임의의 As 조성비에 대한 변수 값을 얻었다. 그 결과 임의의 조성비에 따른 $InAs_xSb_{1-x}$ ($0{\leq}x{\leq}1$) 의 유전율 함수를 얻어낼 수 있었다. 우리는 이러한 결과가 물질의 실시간 성장 모니터링이나 다층구조 분석, 광소자의 제작 등에 유용한 정보를 제공할 것으로 확신한다.