• Journal of Multimedia Information System
• /
• v.8 no.2
• /
• pp.101-110
• /
• 2021

In computer vision, single-image super resolution has been an area of research for a significant period. Traditional techniques involve interpolation-based methods such as Nearest-neighbor, Bilinear, and Bicubic for image restoration. Although implementations of convolutional neural networks have provided outstanding results in recent years, efficiency and single model multi-scalability have been its challenges. Furthermore, previous works haven't placed enough emphasis on real-number scalability. Interpolation-based techniques, however, have no limit in terms of scalability as they are able to upscale images to any desired size. In this paper, we propose a convolutional neural network possessing the advantages of the interpolation-based techniques, which is also efficient, deeming it suitable in practical implementations. It consists of convolutional layers applied on the low-resolution space, post-up-sampling along the end hidden layers, and additional layers on high-resolution space. Up-sampling is applied on a multiple channeled feature map via bicubic interpolation using a single model. Experiments on architectural structure, layer reduction, and real-number scale training are executed with results proving efficient amongst multi-scale learning (including scale multi-path-learning) based models.

• Journal of Korean Clinical Nursing Research
• /
• v.17 no.3
• /
• pp.399-410
• /
• 2011

Purpose: This study was conducted to investigate the incidence of blood hemolysis and repeated blood sampling and to identify factors contributing to hemolysis and repeated blood sampling in the emergency department. Methods: A cross-sectional descriptive design was used. Participants were the patients who came to emergency department and are required a blood sampling for electrolyte level. All blood samples were collected by emergency department nurses and determined for hemolysis by experienced laboratory technologists. Data were analyzed using $x^2$-test, Fisher's exact test, Mann-Whitney u test and Binary Logistic Regression to determine significant differences. Results: A total of 402 valid samples were collected. Of these, 30 blood samples (7.5%) were found to be hemolyzed and 9 (2.2%) to be recollected. Statistically significant factors affecting on hemolysis and repeated blood sampling included the time of bloods sampling (night), the time of tourniquet application, and too-fast blood draw into the test tube. Conclusion: We recommend that nurses who take the blood sampling to consider the findings of the study and take the related factors into account as they set up the standardized care protocol in order for nursing quality improvement.

• Journal of electromagnetic engineering and science
• /
• v.8 no.1
• /
• pp.17-22
• /
• 2008

It is presented that sampling rates for behavioral modeling of quasi-memory less nonlinear devices can be far less than the Nyquist rate of the input signal. Although it has been believed that the sampling rate of nonlinear device modeling should be at least the Nyquist rate of the output signal, this paper suggests that far less than the Nyquist rate of the input signal can be applied to the modeling of quasi-memoryless nonlinear devices, such as frequency multipliers. To verify, a QPSK signal at 820 MHz were applied to a frequency tripler, whereby the device can be utilized as an up-converting mixer into 2.46 GHz with the aid of digital predistortion. AM-AM, AM-PM and PM-PM can be successfully measured regardless of sampling rates.

• Journal of Korean Society for Atmospheric Environment
• /
• v.14 no.1
• /
• pp.43-62
• /
• 1998

This study was carried out to investigate the influence of different sampling methods on the measured concentrations of polycyclic aromatic hydrocarbons (PAH) both in the vapor and particulate phases, and to evaluate the effects of ambient temperature and sampling duration on the losses of PAH associated with particle samples due to volatilization. The experimental protocol of this study is consisted of two parts. The first part is related to the comparison of PAH concentrations measured by 4 different sampling systems, each of which involves different sampling principles for comparison purposes, including a medium-volume sampler with XAD-2 adsorbent, a high-volume sampler with polyurethane foam (PUF), two identical low-volume samplers: one with XAD-2 and the other with PUF, respectively. The second part of this study is to quantitatively estimate the losses of particulate PAH samples by volatilization during sampling, using two identical low-volume samplers: one was used for changing the filters every 3 hrs, 6 hrs, 12 hrs, and 24 hrs sampling, while the other was maintained for continuous 48 hours sampling without changing the filter. The concentrations of volatile PAH including 2-3 rings appeared to be significantly affected by the type of adsorbent. Measured levels of these lower-molecular weight PAH by XAD-2 adsorbent were much higher than those by PUF for both high-volume and low-volume sampling. PUF was found to give rise to unknown components that interfered with the PAH analysis, even after extensive clean-up. In addition, the retention efficiency of PUF for lower molecular weight PAH was subject to a large variation, being significantly influenced by sampling conditions such as ambient temperature. However, the effect of sampling methods with different adsorbents on the measured levels of semi-volatile compounds including 4 rings PAH such as fluoranthene, pyrene, BaA and chrysene, was not so much significant as more volatile PAH compounds. It was also clear from this study that volatilization losses of the semi-volatile PAH collected on the filters were inevitably occurred during prolonged sampling, and hence the results obtained from conventional sampling methods may not be expected to yield an accurate distribution of PAH between the vapor and particulate phases.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.20 no.2
• /
• pp.268-274
• /
• 2016

To find the location of a random noise source present in the three-dimensional space is required at least four microphones. Using four microphones distributed in a three-dimensional space, noise source imaging technique was applied and evaluated on their performance. To compensate resolution problem which comes from both the position of the sensor array is fixed and the sampling frequency is low, up-sampling technique and interpolation function were applied. Five different interpolation methods were applied such as zero-padding, zero-order hold, first-order hold, spline function, and random signal padding. The up-sampling rate were chosen by two, four, eight times, and counting up 16 times. As a result, it was possible to more accurately estimate the position of the noise source according to the higher of the up-sampling rate. It also found that the first-order hold and the spline function's performance were slightly falling relative to other methods.

• Journal of Broadcast Engineering
• /
• v.25 no.2
• /
• pp.242-251
• /
• 2020

Recently, many deep convolutional neural networks for image super-resolution have been studied. Existing deep learning-based super-resolution algorithms are architecture that up-samples the resolution at the end of the network. The post-upsampling architecture has an inefficient structure at large scaling factor result of predicting a lot of information for mapping from low-resolution to high-resolution at once. In this paper, we propose a single image super-resolution using Channel Attention Residual Dense Block based on an iterative up-down sampling architecture. The proposed algorithm efficiently predicts the mapping relationship between low-resolution and high-resolution, and shows up to 0.14dB performance improvement and enhanced subjective image quality compared to the existing algorithm at large scaling factor result.

• Journal of Korean Society of Occupational and Environmental Hygiene
• /
• v.7 no.1
• /
• pp.21-31
• /
• 1997

Toluene, n-hexane, and methyl ethyl ketone(MEK) were exposed to the activated carbon fiber(ACF) and 3M(Model 3500) diffusive samplers under low and high humidity levels. In order to evaluate these two samplers, the sampling capacity, sampling rate, reverse diffusion, and storage stability were obtained. At low humidity level($8{\pm}3%RH$), the adsorption amount of all three organic vapors to the ACF diffusive sampler showed a positive linear relationship up to 8 hours. However, at high humidity level($90{\pm}5%RH$), n-hexane and MEK maintained a positive linear relationship up to 1.5 hrs, but decreased in their adsorption amounts afterwards. On the other hand, the adsorption amount of n-hexane, MEK, and toluene to 3M diffusive sampler showed almost a positive linear relationship up to 8 hours at both humidity levels. At low humidity level, there was almost no reverse diffusion for both 3M and ACF diffusive samplers. However, when the ACF diffusive sampler was used at high humidity level, there was about 52.63% of MEK sample loss and about 92.59% of n-hexane sample loss. The storage stabilities of the ACF and 3M diffusive samplers were both relative stable except for MEK. In the case of MEK, the difference between the analysis of the organic vapor right after the sampling and that of 3 weeks later at room temperature was 45% for the ACF diffusive sampler and 18% for the 3M diffusive sampler. Since the storage stability of the samples stored in a refrigerator was relatively stable, they need to be refrigerated until the analysis is done.

• Journal of the Institute of Electronics Engineers of Korea TC
• /
• v.42 no.5 s.335
• /
• pp.13-22
• /
• 2005

IEEE 802.11a is a standard for the high-speed wireless local area network (WLAN), supporting from 6 up to 54 Mbps in a 5 GHz band. We propose a channel equalization algerian and a sampling clock recovery algorithm by utilizing the Viterbi decoder output of the IEEE 802.11a WLAN standard. The proposed channel equalizer adaptively compensates channel variations. The proposed system uses re-encoded Viterbi decoder outputs as reference symbols for the adaptation of the channel equalizer. It also extracts sampling phase information with the Viterbi decoder outputs for fine adjustment of the sampling clock. The proposed sampling clock recovery and equalizer are more robust to noise and frequency selective fading environments than conventional systems using only four pilot samples.

• Transactions of the Korean Society of Mechanical Engineers B
• /
• v.27 no.3
• /
• pp.398-405
• /
• 2003

A fast response analyzer for measuring carbon dioxide concentration has been developed for transient characteristic and researches tested on internal combustion engine. The analyzer uses the well known NDIR(Non-Dispersive Infrared) method with miniaturized detection system, giving a time constant of approximately 30 microsecond, and sampling module consists of capillary tube. Since the transit time and the time constant of the sampling system depend on the sampling conditions, it is necessary to investigate the characteristics of sampling system before applied to exhaust gas measurement in engine. A unique method was designed to study the influence of the diameter of transfer sample line and operating conditions of the FRNDIR on transit time and time constant. A database of transit time and time constant was built up for different measured and simulated pressure conditions. The database can be used for correcting eventual $CO_2$ concentration measurement.

• Journal of the Korea Society for Simulation
• /
• v.13 no.3
• /
• pp.1-10
• /
• 2004

The cell loss probability required in the ATM network is in the range of 10$^{-9}$ ∼10$^{-12}$ . If Monte Carlo simulation is used to analyze the performance of the ATM node, an enormous amount of computer time is required. To obtain large speed-up factors, importance sampling may be used. Since the Markov-modulated processes have been used to model various high-speed network traffic sources, we consider discrete time single server queueing systems with Markov-modulated arrival processes which can be used to model an ATM node. We apply importance sampling based on the Large Deviation Theory for the performance evaluation of, MMBP/D/1/K, ∑MMBP/D/1/K, and two stage tandem queueing networks with Markov-modulated arrival processes and deterministic service times. The simulation results show that the buffer overflow probabilities obtained by the importance sampling are very close to those obtained by the Monte Carlo simulation and the computer time can be reduced drastically.