• 한국통신학회논문지
• /
• 제28권10C호
• /
• pp.958-966
• /
• 2003

동영상에서 움직이는 객체를 찾아내기 위해 프레임 차분에 기반을 둔 변화 검출 알고리즘이 많이 사용된다. 이러한 알고리즘들은 프레임의 변화를 추정된 통계적 배경 모델을 이용하여 검출한다. 그러나 이러한 추정된 배경 모델이 실제 통계적 분포와 다르면 잘못된 검출 결과들이 생성되게 된다. 본 논문에서는 오일러 수와 구조적 텐서를 이용한 개선된 변화 검출 알고리즘을 제안한다. 제안된 오일러 수에 기반을 둔 맵핑 방법은 Nonparametric 변화검출 알고리즘에 의해 잘못 검출된 결과를 감소시키는데 사용될 수 있다. 또한 본 논문에서 제안된 구조 텐서를 이용한 방법은 움직인 객체 영역 내부의 변화를 검출하는데 사용된다. 제안된 방법은 기존의 방법에 비해 Weather에서는 90%, Mother & daughter에서는 34% 그리고 Aisle에서는 43%의 검출 에러 감소 효과를 얻을 수 있음을 실험 결과로 확인한다.

• Earthquakes and Structures
• /
• 제10권1호
• /
• pp.211-238
• /
• 2016

The Maximum seismic responses of steel buildings with perimeter moment resisting frames (MRF), modeled as complex MDOF systems, are estimated for several incidence angles of the horizontal components and the critical one is identified. The accuracy of the existing rules to combine the effects of the individual components is also studied. Two and three components are considered. The critical response does not occur for principal components and the corresponding incidence angle varies from one earthquake to another. The critical response can be estimated as 1.40 and 1.10 times that of the principal components, for axial load and interstory shears, respectively. The rules underestimate the axial load but reasonably overestimate the shears. The rules are not always inaccurate in the estimation of the combined response for correlated components. On the other hand, totally uncorrelated (principal) components are not always related to an accurate estimation. The correlation of the individual effects (${\rho}$) may be significant, even for principal components. The rules are not always associated to an inaccurate estimation for large values of ${\rho}$, and small values of ${\rho}$ are not always related to an accurate estimation. Only for perfectly uncorrelated harmonic excitations and elastic analysis of SDOF systems, the individual effects of the components are uncorrelated and the rules accurately estimate the combined response. The degree of correlation of the components, the type of structural system, the response parameter under consideration, the location of the structural member and the level of structural deformation must be considered while estimating the level of underestimation or overestimation.

• 한국기계가공학회지
• /
• 제17권5호
• /
• pp.45-51
• /
• 2018

Roll forming refers to the production of long plate-molded products, such as panels, pipes, tubes, channels, and frames, by continuously causing the bending deformation to thin plates using rotating rolls. As the roll forming method has advantages in terms of mass production because of its excellent productivity, the size of the roll forming industry has been continuously increasing and the roll forming method is increasingly being used in diverse industrial fields as a very important processing method. Furthermore, as the roll forming method mainly depends on the continuous bending deformation of the plate materials, the time and the cost of the heterogeneous materials developed in the process are relatively large when considered from the viewpoint of plastic working because many processes are continuously implemented. The existing studies on roll forming manufacturing have reported the loss of large amounts of time and materials when the raw materials or product types were changed; further, they have stated that the use of this method can hardly guarantee the uniformity of the formed shapes and the consistency in terms of size and cannot detect all the defects occurring during the mass production and related to the dimensions. Therefore, in this research, a real-time process data-based smart roll forming method that can be applied to multiple products was studied. As a result, a roll forming system was implemented that remembers and automatically sets the changes in the finely adjusted values of the supplied quantities of individual heterogeneous materials so that the equipment setting changing time for heterogeneous material replacements or changes in the products being produced can be shortened. It also secures the uniformity of the products so that more competitive and precise slide-rail products can be mass-produced with improvements in the quality, price, and productivity of the products.

• Steel and Composite Structures
• /
• 제31권6호
• /
• pp.541-558
• /
• 2019

In Special Concentrically Braced Frames (SCBFs), vertical and horizontal components of the brace force must be resisted by column and beam, respectively but normal force component existing at the gusset plate-to-column and beam interfaces, creates out-of-plane action making distortion in column and beam faces adjacent to the gusset plate. It is a main concern in Hollow Structural Section (HSS) columns and beams where their webs and gusset plate are not in the same plane. In this paper, a new gusset plate passing through the HSS columns and beams, named as through gusset plate, is proposed to study the force transfer mechanism in such gusset plates of SCBFs compared to the case with conventional gusset plates. For this purpose, twelve SCBFs with diagonal brace and HSS columns and twelve SCBFs with chevron brace and HSS columns and beams are considered. For each frame, two cases are considered, one with through gusset plates and the other with conventional ones. Based on numerical results, using through gusset plates prevents distortion and out-of-plane deformation at HSS column and beam faces adjacent to the gusset plate helping the entire column and beam cross-sections to resist respectively vertical and horizontal components of the brace force. Moreover, its application increases energy dissipation, lateral stiffness and strength around 28%, 40% and 32%, respectively, improving connection behavior and raising the resistance of the normal force components at the gusset plate-to-HSS column and beam interfaces to approximately 4 and 3.5 times, respectively. Finally, using such through gusset plates leads to better structural performance particularly for HSS columns and beams with larger width-to-thickness ratio elements.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제13권6호
• /
• pp.3074-3091
• /
• 2019

With the widespread use of the Internet, services for providing large-capacity multimedia data such as video-on-demand (VOD) services and video uploading sites have greatly increased. VOD service providers want to be able to provide users with high-quality keyframes of high quality videos within a few minutes after the broadcast ends. However, existing keyframe extraction tends to select keyframes whose quality as a keyframe is insufficiently considered, and it takes a long computation time because it does not consider an HD class image. In this paper, we propose a keyframe selection method that flexibly applies multiple keyframe quality metrics and improves the computation time. The main procedure is as follows. After shot boundary detection is performed, the first frames are extracted as initial keyframes. The user sets evaluation metrics and priorities by considering the genre and attributes of the video. According to the evaluation metrics and the priority, the low-quality keyframe is selected as a replacement target. The replacement target keyframe is replaced with a high-quality frame in the shot. The proposed method was subjectively evaluated by 23 votes. Approximately 45% of the replaced keyframes were improved and about 18% of the replaced keyframes were adversely affected. Also, it took about 10 minutes to complete the summary of one hour video, which resulted in a reduction of more than 44.5% of the execution time.

• 전기전자학회논문지
• /
• 제24권4호
• /
• pp.1050-1057
• /
• 2020

실감형 360도 미디어는 기존 영상보다 고품질, 초대용량으로 영상의 크기가 크며, 다양한 렌더링 방식을 사용하여 기존방식으로 이미지 처리할 경우 영상인식 속도가 느려지는 문제가 있다. 또한, 실감형 360도 미디어의 특성상 특정 장소에서 카메라를 고정시켜 한 장면만 촬영하는 경우가 대부분이기 때문에, 모든 영상에서 특징정보를 추출할 필요가 없다. 본 논문에서는 실감형 360 미디어의 프레임 추출과정, 프레임 다운사이징, 구형 형태의 렌더링 과정을 거치고, 렌더링 과정에서 영상을 16개 프레임으로 분할 캡처하여 캡처된 프레임에서 객체 정보가 많은 중앙 부분에서 픽셀당 RGB 벡터와 딥 러닝을 이용하여 객체를 추출한 뒤, 객체 특징정보를 이용하여 대표 프레임을 선정하는 방법을 제안한다.

• 방송공학회논문지
• /
• 제28권2호
• /
• pp.185-193
• /
• 2023

최근 인터넷을 통한 동영상 제공 서비스가 확대됨에 따라 높은 품질의 온라인 컨텐츠에 대한 수요가 급증하고 있다. 그런데 넓은 동적 범위 (dynamic range)를 표현할 수 있는 high dynamic range (HDR) 컨텐츠의 공급은 수요를 따라가지 못하고 있는 실정이다. 따라서 본 논문에서는 HDR 영상 제작의 한 방법으로서, 여러 노출값에서 촬영된 프레임들로 구성된 low dynamic range (LDR) 동영상을 이용해 HDR 영상을 생성하는 방법을 제안한다. 우선, 프레임들 사이에 움직임이 존재하기 때문에 정렬 과정을 통해 이웃 프레임들을 중심 프레임에 맞추어 정렬한다. 이때 내용 (content) 기반의 정렬을 하여 정확도를 높이고, 원래 크기의 입력을 그대로 이용하는 모듈을 함께 사용하여 세부 정보도 잘 살려준다. 그러고 나서 잘 정렬된 다중 프레임들을 합쳐서 하나의 HDR 프레임으로 만들어 준다. 실험을 통해 기존 방법들에 비해 우수한 성능을 보임을 확인하였다.

• Steel and Composite Structures
• /
• 제47권2호
• /
• pp.167-184
• /
• 2023

In this study, a new recentering friction device (RFD) to retrofit steel moment frame structures is introduced. The device provides both self-centering and energy dissipation capabilities for the retrofitted structure. A hybrid performance-based seismic design procedure considering multiple limit states is proposed for designing the device and the retrofitted structure. The design of the RFD is achieved by modifying the conventional performance-based seismic design (PBSD) procedure using computational intelligence techniques, namely, genetic algorithm (GA) and artificial neural network (ANN). Numerous nonlinear time-history response analyses (NLTHAs) are conducted on multi-degree of freedom (MDOF) and single-degree of freedom (SDOF) systems to train and validate the ANN to achieve high prediction accuracy. The proposed procedure and the new RFD are assessed using 2D and 3D models globally and locally. Globally, the effectiveness of the proposed device is assessed by conducting NLTHAs to check the maximum inter-story drift ratio (MIDR). Seismic fragilities of the retrofitted models are investigated by constructing fragility curves of the models for different limit states. After that, seismic life cycle cost (LCC) is estimated for the models with and without the retrofit. Locally, the stress concentration at the contact point of the RFD and the existing steel frame is checked being within acceptable limits using finite element modeling (FEM). The RFD showed its effectiveness in minimizing MIDR and eliminating residual drift for low to mid-rise steel frames models tested. GA and ANN proved to be crucial integrated parts in the modified PBSD to achieve the required seismic performance at different limit states with reasonable computational cost. ANN showed a very high prediction accuracy for transformation between MDOF and SDOF systems. Also, the proposed retrofit showed its efficiency in enhancing the seismic fragility and reducing the LCC significantly compared to the un-retrofitted models.

• 전기전자학회논문지
• /
• 제27권3호
• /
• pp.313-318
• /
• 2023

본 논문은 Vision Transformer를 기반으로 하는 Video Classification의 성능을 개선하는 방법으로 fine-tuning를 적용한 신경망을 제안한다. 최근 딥러닝 기반 실시간 비디오 영상 분석의 필요성이 대두되고 있다. Image Classification에 사용되는 기존 CNN 모델의 특징상 연속된 Frame에 대한 연관성을 분석하기 어렵다는 단점이 있다. 이와 같은 문제를 Attention 메커니즘이 적용된 Vistion Transformer와 Non-local 신경망 모델을 비교 분석하여 최적의 모델을 찾아 해결하고자 한다. 또한, 전이 학습 방법으로 fine-tuning의 다양한 방법을 적용하여 최적의 fine-tuning 신경망 모델을 제안한다. 실험은 UCF101 데이터셋으로 모델을 학습시킨 후, UTA-RLDD 데이터셋에 전이 학습 방법을 적용하여 모델의 성능을 검증하였다.

• 대한토목학회논문집
• /
• 제28권4A호
• /
• pp.465-474
• /
• 2008

일반적으로, 보-기둥 부재로 구성된 강뼈대구조물의 설계는 개별부재의 유효좌굴길이를 고려하여 설계기준에서 제시한 안정성 평가식을 적용하고 있다. 그러나 이 방법은 구조물에서 상대적으로 작은 압축력이 적용되는 부재에서는 유효좌굴길이가 커지는 문제가 발생하게 된다. 이러한 문제를 극복하고자 본 연구에서는 대상 구조물의 초기결함(initial imperfection)을 고려한 2차 탄성해석법을 제시한다. 이 방법은 탄성좌굴 고유치해석으로 산정된 좌굴모드 및 좌굴고유치, 개별부재의 축력을 이용하여, 가장 작은 무차원 세장비를 가진 부재를 선정하고, 그 부재에 대하여 기하적, 재료적인 효과가 고려된 설계기준의 기준강도곡선으로부터 좌굴모드에 대한 증폭량을 산정한다. 이렇게 결정된 증폭량을 대상 구조물의 좌굴모드에 증폭시켜 2차 탄성해석을 수행하고, 개별부재의 안정성을 평가한다. 본 방법의 타당성을 확인하기 위하여, 8층 및 4층으로 이루어진 평면 강뼈대구조물에 적용시키고, 설계기준에서 제시하는 안정성 평가법과 비교한다.