• Smart Structures and Systems
• /
• 제29권1호
• /
• pp.221-235
• /
• 2022

Manual inspection of steel box girders on long span bridges is time-consuming and labor-intensive. The quality of inspection relies on the subjective judgements of the inspectors. This study proposes an automated approach to detect and segment cracks in high-resolution images. An end-to-end cascaded framework is proposed to first detect the existence of cracks using a deep convolutional neural network (CNN) and then segment the crack using a modified U-Net encoder-decoder architecture. A Naïve Bayes data fusion scheme is proposed to reduce the false positives and false negatives effectively. To generate the binary crack mask, first, the original images are divided into 448 × 448 overlapping image patches where these image patches are classified as cracks versus non-cracks using a deep CNN. Next, a modified U-Net is trained from scratch using only the crack patches for segmentation. A customized loss function that consists of binary cross entropy loss and the Dice loss is introduced to enhance the segmentation performance. Additionally, a Naïve Bayes fusion strategy is employed to integrate the crack score maps from different overlapping crack patches and to decide whether a pixel is crack or not. Comprehensive experiments have demonstrated that the proposed approach achieves an 81.71% mean intersection over union (mIoU) score across 5 different training/test splits, which is 7.29% higher than the baseline reference implemented with the original U-Net.

• 한국전자통신학회논문지
• /
• 제3권1호
• /
• pp.39-45
• /
• 2008

모바일 환경에서의 지리정보 서비스는 현장에서 필요한 정보를 즉시 얻고 그것을 현장에 동적으로 적용하도록 함으로써 서비스 대상자의 비용과 효율을 높여줄 수 있다. 현재 모바일 환경에서의 지리정보 서비스는 단순한 차량 내비게이션 정도에 머물고 있으며, 대규모 지형을 포함한 복잡한 지리정보에 대한 서비스를 제공하기 위해서는 시스템 사양에 최적화된 콘텐츠의 제공과 처리 기술에 대한 연구가 필요하다.

• Journal of the Korean Society for Industrial and Applied Mathematics
• /
• 제5권1호
• /
• pp.55-78
• /
• 2001

Given the anisotropic Poisson equation $-{\nabla}{\cdot}{\mathcal{K}}{\nabla}p=f$, one can convert it into a system of two first order PDEs: the Darcy law for the flux $u=-{\mathcal{K}{\nabla}p$ and conservation of mass ${\nabla}{\cdot}u=f$. A very natural mixed finite volume method for this system is to seek the pressure in the nonconforming P1 space and the Darcy velocity in the lowest order Raviart-Thomas space. The equations for these variables are obtained by integrating the two first order systems over the triangular volumes. In this paper we show that such a method is really a standard finite element method with local recovery of the flux in disguise. As a consequence, we compare two approaches in analyzing finite volume methods (FVM) and shed light on the proper way of analyzing non co-volume type of FVM. Numerical results for Dirichlet and Neumann problems are included.

• Journal of the Korean Statistical Society
• /
• 제9권1호
• /
• pp.31-38
• /
• 1980

The problem considered in this paper can be defiened as follows. Consider observations $x_1, x_2, \cdot, x_n$ which are assumed to come from a mixed population of the density function, $$f(x) = \sum^m_{k=1} pkf_k(x)$$ where m is the number of subpoulations and $p_k$ is the proportion of subpopulation k such that $\sum^m_{k=1} pk=1, 0

• 생물환경조절학회지
• /
• 제27권4호
• /
• pp.294-301
• /
• 2018

겨울철에 열손실을 줄이기 위해 많은 온실에서 보온커튼을 사용하고 있다. 그러나 적절한 보온커튼을 선택할 때 판단 자료로 활용할 수 있는 명확한 기준이 없는 실정이며 이를 위해서는 보온재의 보온 특성에 대한 정량적인 값이 필요하다. 본 연구에서는 BES를 사용하여 보온커튼의 관류열전달계수를 산정하는 시뮬레이션 모델을 개발하였다. 일중 및 이중 PE필름 피복에 대한 관류열전달계수의 실험값을 사용하여 시뮬레이션 결과를 검증하였다. 검증된 모델을 사용하여 문헌에서 제시된 각종 열적 특성을 가진 보온커튼에 대한 관류열전달계수를 산정하고 비교분석하였다. 개발된 시뮬레이션 모델은 다양한 보온커튼의 관류열전달계수를 산정하는 데 활용될 수 있을 것이며, 제시된 관류열전달계수는 보온커튼의 성능을 정량적으로 비교하는데 유용하게 활용될 수 있을 것으로 판단된다.

• 한국통신학회논문지
• /
• 제19권1호
• /
• pp.10-23
• /
• 1994

온 라인 전송에 있어서 정보를 효율적으로 안전하게 전송하기 위해서는 자료압축과 기밀성 그리고 인증성이 요구된다. 이를 위해서 본 논문에서는 LZW를 완성형 한글로 적용하여 압축열이 2개 생성되는 LZWH4를 제안하며, DES를 개선하여 Differential Cryptanalysis에 대한 일부 대응방안으로 SAC과 상관계수에 만족하는 S-box를 S1~S8에서 S1~S16으로 확장하여 HDES1를 설계하며, 온 라인 전송에 있어서 디지털서명을 위해 이들을 효율적으로 결합한 LZWHDESI를 구현한다. 또한 일반적인 암호강도의 척도인 U.D.(Unicity Distance)에 있어서 HDES1이 DES와 HDES에 비해서 증가됨을 보이며, 본 논문에서 제안한 LZWHDES1가 LZW를 DES에 직접 연결한 LZWDES와 LZWH2를 HDES에 직접 연결한LZWHDES와 비교하여 효율적으로 수행시간이 단축됨을 보이며, LZWHDES1이 관용키 암호시스템으로써 디지털서명 시스템에 활용될 수 있음울 보인다.

• Asia Marketing Journal
• /
• 제15권1호
• /
• pp.43-55
• /
• 2013

Since movies are experience goods, consumers are easily influenced by other consumers' behavior. For moviegoers, box office rank is the most credible and easily accessible information. Many studies have found that the relationship between a movie's box office rank and its revenue departs from the Pareto distribution, and this phenomenon has been named "increasing returns to information." The primary objective of the current research is to apply the empirical model proposed by De Vany and Walls (1996) to the Korean movie market in order to examine whether the same phenomenon prevails in the Korean movie market. The other purpose of the present study is to provide managers with useful implications about the release timing of a movie by finding different curvatures that depend upon seasonality. The empirical test on the Korean movie market shows similar results as prior studies conducted on the U.S., Hong Kong, and U.K. movie markets. The phenomenon of increasing returns is generated by information transmission among consumers, which makes some movies become blockbusters and others bombs. The proposed model can also be interpreted in such a way that a change in the rank has a nonlinear effect on the movie's performance. If a movie climbs up the chart, it would be rewarded more than its proportion. On the other hand, if a movie falls down in the ranks, its performance would drop rapidly. The research result also indicates that the phenomenon of increasing returns occurs differently depending on when the movies are released. Since the tendency of the increasing returns to information is stronger during the peak seasons, movie marketers should decide upon the release timing of a movie based on its competitiveness. If a movie has substantial potential to incur positive word-of-mouth, it would be more reasonable to release the movie during the peak season to enjoy increasing returns. Otherwise, a movie should be released during the low season to minimize the risk of being dropped from the chart.

• 한국철도학회:학술대회논문집
• /
• 한국철도학회 2008년도 춘계학술대회 논문집
• /
• pp.46-55
• /
• 2008

Maglev is a system that a train runs levitated above a rail. Therefore it is very important to maintain a constant levitation gap for achieving serviceability and ride comfort. This study is a cooperation research subject of the 3-1 subject, performance improvement of maglev track structures, of the Center for Urban Maglev Program in Korea, started in 2006. The aim of this study is development of rapid constructions of bridge superstructure for maglev. At present, precast deck is widely used because of its superiority to cast-in-place concrete on quality and the term of works. The research group suggested basic systems of maglev guideway with PSC-U type and trapezoidal open steel box type girder, and precast deck, cooperating with Korea Railroad Research Institute, the managing institute of the 3-1 subject. In this study, longitudinally full-scale guideway girder system was fabricated and static/dynamic test of the girder was performed for the purpose of the performance evaluation.

• BMB Reports
• /
• 제51권12호
• /
• pp.613-622
• /
• 2018

RNA helicases DDX5 and DDX17 are multitasking proteins that regulate gene expression in different biological contexts through diverse activities. Special attention has long been paid to their function as coregulators of transcription factors, providing insight about their functional association with a number of chromatin modifiers and remodelers. However, to date, the variety of described mechanisms has made it difficult to understand precisely how these proteins work at the molecular level, and the contribution of their ATPase domain to these mechanisms remains unclear as well. In light of their association with long noncoding RNAs that are key epigenetic regulators, an emerging view is that DDX5 and DDX17 may act through modulating the activity of various ribonucleoprotein complexes that could ensure their targeting to specific chromatin loci. This review will comprehensively describe the current knowledge on these different mechanisms. We will also discuss the potential roles of DDX5 and DDX17 on the 3D chromatin organization and how these could impact gene expression at the transcriptional and post-transcriptional levels.

• Smart Structures and Systems
• /
• 제29권1호
• /
• pp.153-166
• /
• 2022

Identifying fine cracks in steel bridge facilities is a challenging task of structural health monitoring (SHM). This study proposed an end-to-end crack image segmentation framework based on a one-step Convolutional Neural Network (CNN) for pixel-level object recognition with high accuracy. To particularly address the challenges arising from small object detection in complex background, efforts were made in loss function selection aiming at sample imbalance and module modification in order to improve the generalization ability on complicated images. Specifically, loss functions were compared among alternatives including the Binary Cross Entropy (BCE), Focal, Tversky and Dice loss, with the last three specialized for biased sample distribution. Structural modifications with dilated convolution, Spatial Pyramid Pooling (SPP) and Feature Pyramid Network (FPN) were also performed to form a new backbone termed CrackDet. Models of various loss functions and feature extraction modules were trained on crack images and tested on full-scale images collected on steel box girders. The CNN model incorporated the classic U-Net as its backbone, and Dice loss as its loss function achieved the highest mean Intersection-over-Union (mIoU) of 0.7571 on full-scale pictures. In contrast, the best performance on cropped crack images was achieved by integrating CrackDet with Dice loss at a mIoU of 0.7670.