• Proceedings of the Korea Information Processing Society Conference
• /
• 2017.11a
• /
• pp.799-802
• /
• 2017

We attempt to combines concepts of Morphological Operator(MO) and Convolutional Neural Networks(CNN) to improve image-to-image translation. To do this, we propose an operation that approximates morphological operations. Also we propose S-Convolution, an operation that extends the operation to use multiple filters like CNN. The experiment result shows that it can learn MO with big filter using multiple S-convolution layer of small filter. To validate effectiveness of the proposed layer in image-to-image translation we experiment with GAN with S-convolution applied. The result showed that GAN with S-convolution can achieve distinct result from that of GAN with CNN.

• Korean Journal of Fisheries and Aquatic Sciences
• /
• v.26 no.5
• /
• pp.502-509
• /
• 1993

Submerged filter process was used to evaluate the nitrifying efficiency of ammonia in the recycling water of marine aquatic culture system. The ammonia removal efficiency was achieved as high as $99\%$ at the hydraulic surface loading rate of up to $4.3{\ell}/m^2-day$. And the nitrite accumulation did not occur in the reactor even when the hydraulic surface loading rate of up to $36.8{\ell}/m^2day$ was applied. In the present study, the relationship between the effluent ammonia concentration and ammonia surface loading rate was formulated as an equation. The attachment rate of biofilm on the filter media at the ammonia surface loading rate of 62.3 and $311.7mg/m^2day$ was 15 and $55mg/m^2-day$, respectively, showing the linear relationship between the attachment rate and ammonia loading rates. Biofilm thickness and density of the filter media were found to be the function of the ammonia loading rate.

• Journal of the Korean Ceramic Society
• /
• v.46 no.3
• /
• pp.306-312
• /
• 2009

The size of various alumina ceramics used in semiconductor and display industry is also required to increase with increase in wafer and panel size. In this research, large alumina ceramics was fabricated by filter pressing of alumina slurry using commercial powder and thereafter sintering at $1600^{\circ}C$ in gas furnace. The characteristics of large alumina ceramics thereby were compared to those of small alumina ceramics prepared by pressure forming such as uniaxial pressing and CIP. Careful control of properties of alumina slurry and filter pressing made the fabrication of large alumina ceramics possible, and its characteristics were equivalent to those of small alumina ceramics. The large alumina ceramics, prepared by sintering the green body of 63% relative density at $1600^{\circ}C$, exhibited both dense microstructure corresponding to 98.5% of relative density and 99.8% of high purity as in starting powder.

• Journal of Nuclear Fuel Cycle and Waste Technology(JNFCWT)
• /
• v.3 no.1
• /
• pp.9-16
• /
• 2005

To evaluate parameters influencing on the dust removal of the High Temperature Filter(HTF) system, a computer simulation of fluid dynamics inside the system had been performed. The results showed that the optimum pulse jet periods were 50ms and 90ms for the 1000mm and 1500mm long filter elements respectively. Dust removal effect was very excellent under the pulse jet pressure of 3 bar. But the distance between the pulse jet nozzle and the venturi of a filter element had no meaningful effect on the performance with the variation from 5mm to 10mm. Compared to the dispersion mode of pulse jet, the collective mode of pulse jet flow was preferable in maintaining the pressure inside the system stable.

• Transactions of the Korean Society of Mechanical Engineers B
• /
• v.24 no.11
• /
• pp.1523-1534
• /
• 2000

The performance of a number of existing dynamic subgrid-scale(SGS) models is evaluated in large-eddy simulations(LES) of two prototype transitional and turbulent shear flows, a planar jet and a channel flow. The dynamic SGS models applied include the dynamic Smagorinsky model(DSM);Germano et al. 1991, Lully 1992), the dynamic tow-component model(DTM; Akhavan et al. 2000), the dynamic mixed model(DMM;Zang et al, 1993). and the dynamic two-parameter model(DTPM; Salvetti & Banerjee 1995). The results are compared with those for DNS for their evaluation. The LES results demonstrate the superior performance of DTM with use of a sharp cutoff filter and DMM with use of a box filter, as compared to their respect counterpart DSM, in predicting the mean statistics, spectra and large-scale structure of the flow, Such features of DTM and DMM derive from the construction of the models in which tow separate terms are included to represent the SGS interactions; a Smagorinsky edd-viscosity term to account for the non-local interactions, and a local-interaction term to account for the nonlinear dynamics between the resolved and subgrid scales in the vicinity of the LES cutoff. As well, overall the SGS models using a sharp cutoff filter are more successful than those using a box filter in capturing the statistics and structure of the flow. Finally, DTPM is found to be compatible or inferior to DMM.

• Journal of Korea Multimedia Society
• /
• v.7 no.12
• /
• pp.1657-1664
• /
• 2004

We propose a method that classifies images into two object types man-made and natural objects. A central object is extracted from each image by using central object extraction method[1] before classification. A central object in an images defined as a set of regions that lies around center of the image and has significant color distribution against its surrounding. We define three measures to classify the object images. The first measure is energy of edge direction histogram. The energy is calculated based on the direction of only non-circular edges. The second measure is an energy difference along directions in Gabor filter dictionary. Maximum and minimum energy along directions in Gabor filter dictionary are selected and the energy difference is computed as the ratio of the maximum to the minimum value. The last one is a shape of an object, which is also represented by Gabor filter dictionary. Gabor filter dictionary for the shape of an object differs from the one for the texture in an object in which the former is computed from a binarized object image. Each measure is combined by using majority rule tin which decisions are made by the majority. A test with 600 images shows a classification accuracy of 86%.

• Proceedings of the Korean Institute of Electrical and Electronic Material Engineers Conference
• /
• 2002.11a
• /
• pp.95-98
• /
• 2002

전력선 통신 blocking filter용 자심 재료를 개발하기 위해서 MnO 24 mol%, ZnO 25 mol% and $Fe_{2}O_{3}$ 51 mol% 의 기본조성에 $MoO_{3}$, $SiO_{2}$, CaO를 첨가하여 $1350^{\circ}C$에서 대기압 상수 A를 7.8롤 고정하고 소결하여 미세구조를 제어하였으며 기본 조성에 $MoO_{3}$ 400 ppm, $SiO_{2}$ 100 ppm and CaO 200 ppm을 첨가한 경우 평균 입경 $25{\mu}m$ 의 균일한 결정립으로 구성된 미세구조를 얻었고 기공의 감소에 의한 치밀화로 $4.98g/cm^{2}$의 고밀도화가 이루어 졌다. 또한 소결체의 균일한 미세구조와 고밀도화로 인해서 8221(${25^{\circ}C}$, 1 KHz) 의 가장 높은 투자율 특성을 나타냈다. 시편의 온도가 증가함에 따라 투자율이 증가되어 ${110^{\circ}C}$에서 13904의 거대 투자융이 측정되었고, 코일의 인가주파수가 1 KHz에서 1 MHz까지 증가됨에 따라 최고 ${102^{\circ}C}$까지 시편 온도가 상승하였다. 가장 높은 투자율 특성을 나타낸 ferrite 코어를 사용하여 단상 및 3상용 블로킹 필터의 감쇄율을 측정한 결과 현재 국내의 전력선 통신용 주파수 대역으로 규정되어 있는 10 KHz ~ 450 KHz 대역에서 각각 -46.46 dB와 -73.9 dB의 최고 값을 얻었다.

• The Plant Pathology Journal
• /
• v.21 no.3
• /
• pp.258-261
• /
• 2005

We conducted a survey on pepper virus diseases in 31 regions in Korea from November 2001 to December 2004. Using electron microscopy, test plant reaction, rapid immuno-filter paper assay (RIPA), reverse transcription-polymerase chain reaction (RT-PCR) and/or analysis of viral nucleotide sequences, we found a number of viruses from 1,056 samples that we collected. These included Cucumber mosaic virus (CMV), Pepper mottle virus (PepMoV), Pepper mild mottle virus (PMMoV), Broad bean wilt virus 2 (BBWV2), Tobacco mild green mosaic virus (TMGMV), and Tomato spotted wilt virus (TSWV). Of the samples analyzed, $343(32.5\%)$ were infected with CMV, $209(19.8\%)$ with PepMoV, $141(13.4\%)$ with PMMoV, $12(1.1\%)$ with BBWV2, $40(3.8\%)$ with TMGMV, $5(0.5\%)$ with TSWV, $153(14.5\%)$ with CMV and PepMoV, $54 (5.1\%)$ with CMV and PMMoV, $31(2.9\%)$ with PepMoV and PMMoV, $3(0.3\%)$ with CMV and BBWV2, $1(0.1\%)$ with CMV, PepMoV and BBWV2, $8(0.8\%)$ with CMV, PepMoV and PMMoV, and $30 (2.8\%)$ samples were infected with viruses which were not identified. CMV was the most predominant virus in all inspected fields and the number of the samples infected with PMMoV was relatively low as compared PepMoV infection level in pepper. TMGMV was only found in the southern part of Korea, while TSWV was isolated in Anyang and Yesan. However, we did not encounter in this survey the Alfalfa mosaic virus (AMV), Potato virus Y (PVY), Tobacco mosaic virus (TMV), and Pepper vein chlorosis virus (PVCV).

• Journal of the Korean Institute of Intelligent Systems
• /
• v.23 no.5
• /
• pp.418-422
• /
• 2013

This paper has considered the location estimation problem in two dimension space by using a non-linear filter under non-Gaussian noise in underwater acoustic sensor networks(UASNs). Recently, the extended Kalman filter (EKF) is widely used in location estimation. However, the EKF has a lot of problems in the non-linear system under the non-gaussian noise environment like underwater environment. In this paper, we propose the improved Two-Dimension Particle Filter (TDPF) using the re-interpretation distribution techniques based on the maximum likelihood (ML). Through the simulation, we compared and analyzed the proposed TDPF with the EKF under the non-Gaussian underwater sensor networks. Finally, we determined that the TDPF's result shows more accurate localization than EKF's result.

• Journal of Radiation Protection and Research
• /
• v.38 no.2
• /
• pp.78-90
• /
• 2013

Recently, the interest on exposure to radiation is rising. The radiation exposure of mammography is higher in absorbed dose than of X-ray, therefore unnecessary exposure needs to be reduced, and higher image quality is needed. Generally, ray quality of the radiation imaging is an important factor that determines image quality and the amount of ray exposure, and they are affected by tube voltage and added filter. The X-ray energy that is exposed from mammography device is generally a continuous spectrum, which includes low energy that has minute influence on the image quality, and high energy that hinders contrast on image. Currently, molybdenum (Mo) and rhodium (Rh) are the most used added filters for mammography device, and they are used differently according to the energy region of X-ray. This study aims to find out the degree of reduction in exposure dose according to the thickness of aluminum (Al), and to study the changes in image quality and dose when the added filter plates that are made with niobium (Nb) or zirconium (Zr) are used, other than molybdenum (Mo) and rhodium (Rh), the two most used added filters that have similar atomic number and K-absorption regions as Nb and Zr. In this study, single-added filters of molybdenum (Mo), niobium (Nb), and zirconium (Zr) are used, and in some cases, Aluminum (Al) is combined with the single filters. In this case, image quality is considered to be improved depending on the type of added filters, and by using Aluminum (Al) filter together with the others, unnecessary X-ray of low energy would be absorbed, therefore the dose is expected to decrease without any influence when the concentration level becomes identical.