• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2021.06a
• /
• pp.303-306
• /
• 2021

이미지 초해상도는 영상 취득 과정에서 센서와 렌즈의 물리적인 한계 등으로 인하여 의해 화질이 저하된 이미지를 더 높은 배율로 복원하는 문제이다. 이미지 초해상도는 딥러닝을 통해 놀라운 성능향상을 이루었지만, 카메라로 촬영된 실제 이미지에서는 좋은 성능을 내지 못하였다. 이는 딥러닝에서는 'bicubic' 커널로 down-sampling된 합성 이미지 데이터를 사용하였던 것과 달리 실제 이미지에서는 'bicubic' 커널을 통한 화질 저하와는 다른 화질 저하, 즉 다른 커널을 통한 화질 저하가 발생하기 때문이다. 따라서 실제 이미지에 대한 성능을 높이기 위해서는 이에 대한 정확한 커널 예측이 필요하다. 최근 주목받기 시작한 이미지 초해상도를 위한 커널 예측은 초해상도를 잘 시켜주는 커널을 직접 찾는 방법[10, 13]과 이미지의 분포와 커널을 통해 다운샘플된 이미지에 대한 분포를 일치시켜주면서 커널을 예측하는 방법[14]으로 나누어져 있다. 그러나 두 방법 모두 ill-posed problem 인 커널 예측 문제를 한 장의 이미지만으로 해결하려는 것이기 때문에 정확한 예측에는 어려움이 발생한다. 따라서 본 논문에서는 두 장의 이미지를 활용한 이미지 화질 저하 커널 예측 방법을 제안한다. 제안된 방법은 두 장의 이미지가 같은 카메라를 통해 촬영되었으며 이때 이미지 화질 저하는 카메라에 의해서만 영향을 받는다는 가정을 기반으로 한다. 즉, 두 장의 이미지는 같은 커널을 통해 저하된 이미지라는 가정을 한다. 제안된 방법은 [14]에서처럼 이미지 분포를 기반으로 한 커널 예측을 진행하며, 이미지 초해상도를 진행하고자 하는 이미지 외에 참고 이미지 또한 같은 커널에서 화질 저하를 시켰을 때 본래의 이미지와 같은 분포에 있도록 학습을 진행한다. 결과적으로 본 논문에서는 두 장의 이미지를 사용하였을 때 더욱 정확하게 커널을 찾을 수 있음을 보여준다. 두 장의 이미지를 활용하는 방식이 한 장의 이미지만을 활용하는 기존의 최고 수준의 방법에 비해 합성된 다양한 커널 데이터셋[14]에서 약 0.17dB 성능 향상이 있었다.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.51 no.8
• /
• pp.87-96
• /
• 2014

In this paper, we propose a directionally adaptive aliasing and noise removal using dictionary learning based on space-frequency analysis. The proposed aliasing and noise removal algorithm consists of two modules; i) aliasing and noise detection using dictionary learning and analysis of frequency characteristics from the combined wavelet-Fourier transform and ii) aliasing removal with suppressing noise based on the directional shrinkage in the detected regions. The proposed method can preserve the high-frequency details because aliasing and noise region is detected. Experimental results show that the proposed algorithm can efficiently reduce aliasing and noise while minimizing losses of high-frequency details and generation of artifacts comparing with the conventional methods. The proposed algorithm is suitable for various applications such as image resampling, super-resolution image, and robot vision.

• Journal of The Korean Astronomical Society
• /
• v.49 no.3
• /
• pp.73-81
• /
• 2016

We report the characterization of a massive (m_p = 3.9±1.4M_jup) microlensing planet (OGLE-2015-BLG-0954Lb) orbiting an M dwarf host (M = 0.33 ± 0.12M_⊙) at a distance toward the Galactic bulge of $0.6^{+0.4}_{-0.2}kpc$, which is extremely nearby by microlensing standards. The planet-host projected separation is a⊥ ~ 1.2AU. The characterization was made possible by the wide-field (4 deg²) high cadence (Γ = 6 hr^–1) monitoring of the Korea Microlensing Telescope Network (KMTNet), which had two of its three telescopes in commissioning operations at the time of the planetary anomaly. The source crossing time t_* = 16 min is among the shortest ever published. The high-cadence, wide-field observations that are the hallmark of KMTNet are the only way to routinely capture such short crossings. High-cadence resolution of short caustic crossings will preferentially lead to mass and distance measurements for the lens. This is because the short crossing time typically implies a nearby lens, which enables the measurement of additional effects (bright lens and/or microlens parallax). When combined with the measured crossing time, these effects can yield planet/host masses and distance.

• The Bulletin of The Korean Astronomical Society
• /
• v.38 no.2
• /
• pp.71.1-71.1
• /
• 2013

Among quadruples or higher multiplicity stars, only a few binary systems have been discovered. They are important targets to understand the formation and evolution of multiple stellar systems because we can obtain accurate stellar parameters from photometric and spectroscopic studies. We present the observational results of this kind of rare object 1SWASP J093010.78+533859.5, for which the doubly eclipsing feature had been detected previously from the SuperWASP photometric archive. Individual PSF photometry for two objects with a separation of about 1.9 arcsec was performed for the first time in this study. Our time-series photometric data show that the brighter object A is an Algol-type detached eclipsing binary with an orbital period of 1.3 days and the fainter B is a W UMa-type contact eclipsing binary with a period of 0.23 days. Using the high-resolution optical spectra, we obtained well-defined radial velocity variations of the system A. Furthermore, stationary spectral lines were detected and should have originated from the other stellar component, which was confirmed by the third object contribution from the light curve analysis. No spectral feature of the system B was detected, probably due to its faintness. We obtained the binary parameters and the absolute dimensions from each light curve synthesis. The primary and secondary components of the system A have a spectral type of K1 and K5 main sequences, respectively. Two components of system B have nearly the same type of K3 main sequence. Light variations at out of eclipses were appeared in both systems, interpreting as the effect of stellar spots on these late spectral type stars. We estimated the distances to the systems A and B individually. They may have similar distances of about 70 pc and seem to be gravitationally bound with a separation of about 130 AU. In conclusion, we suggest that 1SWASP J093010.78+533859.5 is a quintuple stellar system with a hierarchical structure of a triple system A(ab)c and a binary system B(ab).

• Journal of Korean Society for Atmospheric Environment
• /
• v.32 no.5
• /
• pp.485-500
• /
• 2016

We examined acidity state of inorganic aerosol and oxidation state of organic aerosol by High Resolution Time of Flight Aerosol Mass Spectrometer (HR-ToF-AMS) at Baengnyeong Super site from Jan 2012 to Dec 2013. Additionally, we carried out the analysis for the aerosol component group of organic matter ($C_xH_y$, $C_xH_yO_1$, $C_xH_yO_z$, $C_xH_yO_zN_p$) and elemental composition to calculate H/C, O/C, N/C, OM/OC and identify the oxidation state. The aerosol chemical composition in this study is dominated by sulfate ($SO_4{^{2-}}$), nitrate ($NO_3{^-}$) plays a smaller role in aerosol acidity. Ammonium ($NH_4{^+}$) was found in a formation of $(NH_4)_3H(SO_4)_2$. However, the binding formations of $NH_4NO_3$ and $NH_4Cl$ increase in the winter. $C_xH_yO_1$ indicating the oxidized state of $PM_{1.0}$ has the highest ratio of 41% while $C_xH_y$ indicating the non-oxidized state has a lower ratio of 36%, meaning that the oxidation level of $PM_{1.0}$ in Baengnyeong Island is high. The ratio between H/C and O/C was 1.33 and 0.78 respectively, showing the characteristic of LV-OOA (Low volatility-Oxygenated Organic Aerosol). Acidic and oxidized aerosols sampled during this field study were largely anthropogenic in origin from Chinese continent and photochemically aged.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.37C no.11
• /
• pp.1077-1083
• /
• 2012

The jamming countermeasures in GNSS includes anti-jamming technique and jammer localization technique. In both techniques, direction of jamming signal is important and generally the MUSIC algorithm is used to find the direction of jamming signal. The MUSIC is super-resolution algorithm for detecting incident direction of signal. But, the search time of MUSIC algorithm is too long because all candidates of incidence angle are searched. This paper proposes the new method that has less computational burdens and therefore faster than the conventional MUSIC algorithm. The proposed method improves performance speed by reducing unnecessary calculations. In the proposed method, the cost function of conventional MUSIC algorithm is decomposed into the sum of squares and if the partial sum of cost function is larger than the minimum cost function so far, then the candidate is rejected and next candidates are searched. If the computed cost function is less than the minimum cost function so far, the minimum cost function so far is replaced with newly computed value. The performance of the proposed method was compared with the conventional MUSIC algorithm using the simulation. The accuracy of the estimaed direction of jamming signal was same as the conventional MUSIC while the search speed of the proposed method was 1.15 times faster than the conventional MUSIC.

• Korean Journal of Remote Sensing
• /
• v.13 no.1
• /
• pp.13-30
• /
• 1997

Recent improvements of satellite remote sensing sensors which are represented by hyperspectral imaging sensors and high spatial resolution sensors provide a large amount of data, typically several hundred megabytes per one scene. Moreover, increasing information exchange via internet and information super-highway requires the developments of more active service systems for processing and analysing of remote sensing data in order to provide value-added products. In this sense, an advanced satellite data processing system is being developed to achive high performance in computing speed and efficieney in processing a huge volume of data, and to make possible network computing and easy improving, upgrading and managing of systems. JAVA internet programming language provides several advantages for developing software such as object-oriented programming, multi-threading and robust memory managent. Using these features, a satellite data processing system named as GeoPixel has been developing using JAVA language. The GeoPixel adopted newly developed techniques including object-pipe connect method between each process and multi-threading structure. In other words, this system has characteristics such as independent operating platform and efficient data processing by handling a huge volume of remote sensing data with robustness. In the evaluation of data processing capability, the satisfactory results were shown in utilizing computer resources(CPU and Memory) and processing speeds.

• KIPS Transactions on Software and Data Engineering
• /
• v.9 no.12
• /
• pp.403-410
• /
• 2020

In addition to the research on noise removal and super-resolution using the data restoration (Output result) function of Autoencoder, research on the performance improvement of clustering using the dimension reduction function of autoencoder are actively being conducted. The clustering function and data restoration function using Autoencoder have common points that both improve performance through the same learning. Based on these characteristics, this study conducted an experiment to see if the autoencoder model designed to have excellent data recovery performance is superior in clustering performance. Skip connection technique was used to design autoencoder with excellent data recovery performance. The output result performance and clustering performance of both autoencoder model with Skip connection and model without Skip connection were shown as graph and visual extract. The output result performance was increased, but the clustering performance was decreased. This result indicates that the neural network models such as autoencoders are not sure that each layer has learned the characteristics of the data well if the output result is good. Lastly, the performance degradation of clustering was compensated by using both latent code and skip connection. This study is a prior study to solve the Hanja Unicode problem by clustering.

• Geophysics and Geophysical Exploration
• /
• v.25 no.3
• /
• pp.140-161
• /
• 2022

Seismic data with missing traces are often obtained regularly or irregularly due to environmental and economic constraints in their acquisition. Accordingly, seismic data interpolation is an essential step in seismic data processing. Recently, research activity on machine learning-based seismic data interpolation has been flourishing. In particular, convolutional neural network (CNN) and generative adversarial network (GAN), which are widely used algorithms for super-resolution problem solving in the image processing field, are also used for seismic data interpolation. In this study, CNN-based algorithm, U-Net and GAN-based algorithm, and conditional Wasserstein GAN (cWGAN) were used as seismic data interpolation methods. The results and performances of the methods were evaluated thoroughly to find an optimal interpolation method, which reconstructs with high accuracy missing seismic data. The work process for model training and performance evaluation was divided into two cases (i.e., Cases I and II). In Case I, we trained the model using only the regularly sampled data with 50% missing traces. We evaluated the model performance by applying the trained model to a total of six different test datasets, which consisted of a combination of regular, irregular, and sampling ratios. In Case II, six different models were generated using the training datasets sampled in the same way as the six test datasets. The models were applied to the same test datasets used in Case I to compare the results. We found that cWGAN showed better prediction performance than U-Net with higher PSNR and SSIM. However, cWGAN generated additional noise to the prediction results; thus, an ensemble technique was performed to remove the noise and improve the accuracy. The cWGAN ensemble model removed successfully the noise and showed improved PSNR and SSIM compared with existing individual models.

• Journal of the Korea Computer Graphics Society
• /
• v.26 no.3
• /
• pp.123-131
• /
• 2020

This study proposes a real-time rendering algorithm for lighting when each of more than 100,000 moving particles exists as a light source. Two 3D textures are used to dynamically determine the range of influence of each light, and the first 3D texture has light color and the second 3D texture has light direction information. Each frame goes through two steps. The first step is to update the particle information required for 3D texture initialization and rendering based on the Compute shader. Convert the particle position to the sampling coordinates of the 3D texture, and based on this coordinate, update the colour sum of the particle lights affecting the corresponding voxels for the first 3D texture and the sum of the directional vectors from the corresponding voxels to the particle lights for the second 3D texture. The second stage operates on a general rendering pipeline. Based on the polygon world position to be rendered first, the exact sampling coordinates of the 3D texture updated in the first step are calculated. Since the sample coordinates correspond 1:1 to the size of the 3D texture and the size of the game world, use the world coordinates of the pixel as the sampling coordinates. Lighting process is carried out based on the color of the sampled pixel and the direction vector of the light. The 3D texture corresponds 1:1 to the actual game world and assumes a minimum unit of 1m, but in areas smaller than 1m, problems such as stairs caused by resolution restrictions occur. Interpolation and super sampling are performed during texture sampling to improve these problems. Measurements of the time taken to render a frame showed that 146 ms was spent on the forward lighting pipeline, 46 ms on the defered lighting pipeline when the number of particles was 262144, and 214 ms on the forward lighting pipeline and 104 ms on the deferred lighting pipeline when the number of particle lights was 1,024766.