• Journal of the Institute of Electronics Engineers of Korea TE
• /
• v.37 no.3
• /
• pp.45-52
• /
• 2000

In this study, we propose the image segmentation algorithm for facial region segmentation. The proposed algorithm separates the mean image of low frequency band from the differential image of high frequency band in order to make a boundary using HWT, and then we reduce the isolation pixels, projection pixels, and overlapped boundary pixels from the low frequency band. Also the boundaries are detected and simplified by the proposed boundary detection algorithm, which are cleared on the thinning process of 1 pixel unit. After extracting facial image boundary by using the proposed algorithm, we make the mask and segment facial image through matching original image. In the result of facial region segmentation experiment by using the proposed algorithm, the successive facial segmentation have 95.88% segmentation value.

• Journal of radiological science and technology
• /
• v.8 no.2
• /
• pp.71-72
• /
• 1985

This study was performed on the dose distribution of various field size and the effect of penumbra shield in the telecobalt unit. The results obtained are as follows. 1. Errors of the light and ${\gamma}-ray$ field size was below the regulation as 0.52 percentage. 2. The coefficient of field area was increased with the larger field area, and this coefficient was showed the more difference in larger SSD. 3. The rectangular field areas, which were described by level of the same percentage depth does, were decreased with the more elongation factor. At the same elongation factor, the compensating factor was decreased with the larger field size. 4. The lead block or extension collimator was able to shield r-ray exposure of outside field size from 50 to 80 percentage. 5. On the matching adjacent fields, while the gap between beam edges are contacted, that overlapped beam edges indicated up to 140 percentage, and while the gap was 1 cm, it could be reduced to 90 Percentage. The lead-libocking on the overlapped area was more effective to lower dose, as 80 percentage in this case. 6. Percentage depth dose of various trimming field sizes were increased linearlly according to area 1 perimeter size, but the center split field size did not maintain linearlly.

• Journal of electromagnetic engineering and science
• /
• v.17 no.4
• /
• pp.178-180
• /
• 2017

This study presents a 2-20 GHz monolithic distributed power amplifier (DPA) using a $0.25{\mu}m$ AlGaN/GaN on SiC high electron mobility transistor (HEMT) technology. The gate width of the HEMT was selected after considering the input capacitance of the unit cell that guarantees decade bandwidth. To achieve high output power using small transistors, a 12-stage DPA was designed with a non-uniform drain line impedance to provide optimal output power matching. The maximum operating frequency of the proposed DPA is above 20 GHz, which is higher than those of other DPAs manufactured with the same gate-length process. The measured output power and power-added efficiency of the DPA monolithic microwave integrated circuit (MMIC) are 35.3-38.6 dBm and 11.4%-31%, respectively, for 2-20 GHz.

• Survey Research
• /
• v.10 no.1
• /
• pp.169-186
• /
• 2009

The propensity score method is used to minimize the bias level in social survey, which comes from nonresponse. The theoretical concept and the background of the propensity score method is discussed first. The propensity score method was first applied in the epidemiology observational study. I have summarized the process of the three propensity score methods that were used to reduce estimation bias in this study. Matching by propensity score is applied to the relatively large control group. Subclassification has the advantage of using whole control group data and regression adjustment is applied to multiple covariates as well as propensity score of each unit is computable and usable. Lastly, the application procedures of propensity score method to reduce the nonresponse bias is suggested and its applicability to real situation is reviewed with the existing data.

• Journal of Biosystems Engineering
• /
• v.40 no.3
• /
• pp.250-260
• /
• 2015

Purpose: Dual cylindrical microwave chambers equipped with an ohmic heating tube were designed and fabricated to maximize the electric field strength for expeditious heat treatment of particulate foods. Methods: The efficacy of the combination heater was investigated by simulating the electric field distribution by using COMSOL Multiphysics software. Results: All components of the designed microwave heating unit were suitable for transmitting maximal microwave power to the load. The simulated electric field distribution implied that single-mode microwave heating would be sufficient for the steady generation of a highly localized heating zone in the cavity. During impedance matching, the calculated reflection coefficient ($S_{11}$) was small, possibly implying minimal power loss and wave reflection in the designed microwave heating chamber. Conclusions: This study demonstrates the possibility of concentrating the microwave power at the centerline for a single-frequency microwave, for thermal treatment of multiphase foods without attenuating the microwave power.

• ETRI Journal
• /
• v.41 no.6
• /
• pp.771-781
• /
• 2019

We propose the VS³-NET model to solve the task of question answering questions with machine-reading comprehension that searches for an appropriate answer in a given context. VS³-NET is a model that trains latent variables for each question using variational inferences based on a model of a simple recurrent unit-based sentences and self-matching networks. The types of questions vary, and the answers depend on the type of question. To perform efficient inference and learning, we introduce neural question-type models to approximate the prior and posterior distributions of the latent variables, and we use these approximated distributions to optimize a reparameterized variational lower bound. The context given in machine-reading comprehension usually comprises several sentences, leading to performance degradation caused by context length. Therefore, we model a hierarchical structure using sentence encoding, in which as the context becomes longer, the performance degrades. Experimental results show that the proposed VS³-NET model has an exact-match score of 76.8% and an F1 score of 84.5% on the SQuAD test set.

• Proceedings of the KIEE Conference
• /
• 2003.07d
• /
• pp.2130-2133
• /
• 2003

For real-time recognizing of the load vehicles a new Neural Network algorithm is proposed. The proposed neural network technique is the real time computation method through the inter-node diffusion. In the network, a node corresponds to a state in the quantized input space. Each node is composed of a Processing unit and fixed weights from its neighbor nodes as well as its input terminal. The most reliable algorithm derived for real time recognition of vehicles, is a dynamic programming based algorithm based on sequence matching techniques that would process the data as it arrives and could therefore provide continuously updated neighbor information estimates. Through severa simulation experiments, real time reconstruction nonlinear image information is Processed. 1-D hardware has been composed and various experi with static and dynamic signals have implemented.

• Journal of the Korean Society for Nondestructive Testing
• /
• v.36 no.1
• /
• pp.9-17
• /
• 2016

The elastic modulus of a 3D-printed Kelvin foam plate is investigated by measuring the acoustic wave velocity of 1 MHz ultrasound. An isotropic tetrakaidecahedron foam with 3 mm unit cell is designed and printed layer upon layer to fabricate a Kelvin foam plate of 14 mm thickness with a 3D CAD/printer using ABS plastic. The Kelvin foam plate is completely filled with paraffin wax for impedance matching, so that the acoustic wave may propagate through the porous foam plate. The acoustic wave velocity of the foam plate is measured using the time-of-flight (TOF) method and is used to calculate the elastic modulus of the Kelvin foam plate based on acousto-elasticity. Finite element method (FEM) and micromechanics is applied to the Kelvin foam plate to calculate the theoretical elastic modulus using a non-isotropic tetrakaidecahedron model. The predicted elastic modulus of the Kelvin foam plate from FEM and micromechanics model is similar, which is only 3-4% of the bulk material. The experimental value of the elastic modulus from the ultrasonic method is approximately twice as that of the numerical and theoretical methods because of the flexural deformation of the cell edges neglected in the ultrasonic method.

• The Bulletin of The Korean Astronomical Society
• /
• v.44 no.1
• /
• pp.46.2-46.2
• /
• 2019

We propose a deep learning model that can generate a waveform of coalescing binary black holes in merging and ring-down phases in less than one second with a graphics processing unit (GPU) as an approximant of gravitational waveforms. Up to date, numerical relativity has been accepted as the most adequate tool for the accurate prediction of merger phase of waveform, but it is known that it typically requires huge amount of computational costs. We present our method can generate the waveform with ~98% matching to that of the status-of-the-art waveform approximant, effective-one-body model calibrated to numerical relativity simulation and the time for the generation of ~1500 waveforms takes O(1) seconds. The validity of our model is also tested through the recovery of signal-to-noise ratio and the recovery of waveform parameters by injecting the generated waveforms into a public open noise data produced by LIGO. Our model is readily extendable to incorporate additional physics such as higher harmonics modes of the ring-down phase and eccentric encounters, since it only requires sufficient number of training data from numerical relativity simulations.

• Journal of Positioning, Navigation, and Timing
• /
• v.10 no.3
• /
• pp.189-196
• /
• 2021

In this paper, we propose techniques to correct the altitude error and heading error of 3D Pedestrian Inertial Navigation (PIN). When a PIN is used to estimate the location of a pedestrian only with an Inetrial Measurement Unit (IMU) without infrastructure, there is a problem in that the location error gradually increases due to the limitation of the observability of the filter. To solve this problem without additional sensors, we propose two techniques in this paper. First, stair walking is recognized in consideration of the altitude difference that may occur during one step. If it is recognized as stair walking, only Zero-velocity UPdaTe (ZUPT) is performed, and if it is recognized as level walking, ZUPT + Altitude Damping (AD) is performed together to correct the altitude error. Second, the straight-line movement direction is calculated through the difference of the estimated position, and the heading error is corrected by matching this information with the link information of the digital map. By applying these techniques, it is verified through real tests that accurate three-dimensional location information of pedestrians can be estimated without infrastructure.