http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
복셀 기반 광선추적 기법을 이용한 복사형상계수 계산 연구
김동건,권구용,장현성,하남구,송치오 한국전산유체공학회 2019 한국전산유체공학회지 Vol.24 No.4
IR signals are divided into self-emitted radiation, surface reflection and path radiation components. A self-emitted radiation is due to temperature and emissivity of the object surface, and it accounts for the highest proportion of the total IR signals. Therefore, in order to predict the IR signals of objects precisely, it must be preceded by an accurate calculation of the surface temperature. In the process of calculating surface temperature, it is important to evaluate the radiative heat transfer including the view factor between the object surfaces. The calculation of view factor requires the process of determining the geometry between meshes on the object surface and subdividing meshes, which needs a lot of computational time. In ordinary cases, accuracy and calculation time have conflicting relationships. To overcome these limitations, various studies show that the ray tracing is being applied to the view factor calculation. In this paper, an estimation method of the view factor calculation using the voxel-based ray tracing technique that had some strategies (hierarchical bounding volumes and space partitioning) to reduce computational expense. By applying this method, the view factor results show acceptable accuracy compared to the results of the commercial S/W, and the calculation time is also less than the results applied the traditional method of performing calculations sequentially. This advantage can be more pronounced as the nimber of meshes increases.
안개영상의 의미론적 분할 및 안개제거를 위한 심층 멀티태스크 네트워크
송태용,장현성,하남구,연윤모,권구용,손광훈 한국멀티미디어학회 2019 멀티미디어학회논문지 Vol.22 No.9
Image semantic segmentation and dehazing are key tasks in the computer vision. In recent years, researches in both tasks have achieved substantial improvements in performance with the development of Convolutional Neural Network (CNN). However, most of the previous works for semantic segmentation assume the images are captured in clear weather and show degraded performance under hazy images with low contrast and faded color. Meanwhile, dehazing aims to recover clear image given observed hazy image, which is an ill-posed problem and can be alleviated with additional information about the image. In this work, we propose a deep multi-task network for simultaneous semantic segmentation and dehazing. The proposed network takes single haze image as input and predicts dense semantic segmentation map and clear image. The visual information getting refined during the dehazing process can help the recognition task of semantic segmentation. On the other hand, semantic features obtained during the semantic segmentation process can provide cues for color priors for objects, which can help dehazing process. Experimental results demonstrate the effectiveness of the proposed multi-task approach, showing improved performance compared to the separate networks.
정형주,장현성,하남구,연윤모,권구용,손광훈 한국멀티미디어학회 2019 멀티미디어학회논문지 Vol.22 No.10
We present a novel deep learning architecture for obtaining a latent image from a single blurry image, which contains dynamic motion blurs through object/camera movements. The proposed architecture consists of two sub-modules: blur image restoration and optical flow estimation. The tasks are highly related in that object/camera movements make cause blurry artifacts, whereas they are estimated through optical flow. The ablation study demonstrates that training multi-task architecture simultaneously improves both tasks compared to handling them separately. Objective and subjective evaluations show that our method outperforms the state-of-the-arts deep learning based techniques.
야외 환경에 노출된 입체 구조물에 대한 열적외선 신호 예측 소프트웨어 개발 및 실험적 검증 연구
김동건(D.G. Kim),권구용(K.Y. Kwon),장현성(H.S. Jang),연윤모(Y.M. Yeon),하남구(N.K. Ha) 한국전산유체공학회 2021 한국전산유체공학회지 Vol.26 No.3
This paper contains experimental validation of surface temperature and IR(infrared) signal prediction performance for a 3D object placed in an outdoor environment as part of a software development process to predict thermal infrared signals based on computational thermal analysis with considering 3D thermal conduction and unsteady state. For experimental validation, CUBI targets, which are introduced as major verification cases abroad, were produced and installed on the experimental site. Also, thermocouple wires, infrared cameras, a solar tracker and a weather station were installed on the experimental site. Performance validation was carried out step by step, and as the first step, the suitability of the algorithm applied to the developed software was determined through comparison and analysis of prediction results for surface temperature. As a second step, the self-emitted radiance calculation result and the solar and sky reflected radiance calculation algorithm were validated through the analysis of the prediction result for infrared signals. In addition, by assigning the same input conditions to the other commercial software with similar concepts and comparing the calculated results together, the limitations of modeling inevitably appear in the comparison between modeling and measurement were considered. As a result of the validation, it is judged that the thermal infrared signal prediction performance of the software developed through this study is applicable to stereoscopic objects placed in an outdoor environment. In the future, it is necessary to prepare to become software with higher reliability through performance validation of 3D objects that have more complex forms or contain heat sources.
윤성준,심규진,고강욱,하남구,이민석,장현성,권구용,박가영,김창익 대한전자공학회 2022 전자공학회논문지 Vol.59 No.2
이미 촬영된 동영상의 품질을 카메라의 물리적 한계를 극복하여 향상하는 연구가 활발하게 진행되어왔다. 특히 동영상 안정화는, 외부적인 요인에 의해 흔들리는 카메라로부터 획득한 불안정한 동영상을 흔들리지 않는 동영상으로 변환하는 기술이며, 고품질의 동영상을 획득하기 위한 필수적인 연구 분야이다. 본 논문에서는 가시광선 정보가 부족한 저조도 환경에서 중요하게 사용되고 있는 적외선 동영상이, 차량 혹은 소형 비행 물체 등에서 촬영되어 물리적인 작은 흔들림이 발생하였을 때 이를 실시간으로 안정화할 수 있는 두 가지 딥러닝 기반 모델을 제안한다. 제안하는 모델은 불안정한 동영상과 정답 안정한 동영상 학습 데이터 페어가 없는 환경에서 학습할 수 있는 자기 지도 학습 모델이다. 우리가 제안하는 안정화 모델 두 가지는, 적외선 동영상(해상도: 640x512)에 대하여 실시간 프레임 처리율(>30 fps)과 함께 기존 딥러닝 기반 온라인 안정화 모델보다 더 좋은 안정화 정도를 보이는 것을 수치적 평가 및 결과 영상을 통해 확인하였다.
3차원 복합 열해석 기반 적외선 영상 생성 알고리즘 설계 및 구현
장현성(Hyun-Sung Jang),김동건(Dong-Geon Kim),권구용(Ku-Yong Kwon),하남구(Nam-Koo Ha) 융복합지식학회 2021 융복합지식학회논문지 Vol.9 No.4
적외선 영상은 일반적인 가시광 영상에 비해 사용 가능한 데이터셋이 상대적으로 적을 뿐만 아니라 영상획득 비용도 높다. 따라서, 머신러닝이나 시뮬레이션 분야에서는 적외선 영상을 생성하여 활용하려는 연구가 활발하다. 본 논문에서는 정밀한 표면온도 계산을 바탕으로 실제와 유사한 적외선 영상을 생성하는 방안과 그 구현 결과를 제시하였다. 실제 물리적 수식을 기반으로 한 3차원 복합 열해석을 바탕으로 복잡한 형상의 차량에 대한 열해석을 수행하였고, 제시한 방안의 오차가 약 1% 정도로 매우 정밀함을 확인하였다. 이러한 정확한 표면온도와 BRDF 값을 이용한 표면의 방향별 반사율을 이용하여 적외선 영상을 생성하면, 실제 적외선 카메라로 촬영된 영상과 유사한 적외선 영상의 생성이 가능함을 구현 결과로 확인할 수 있었다. Compared to general visible light images, infrared images have relatively few available data sets and also have high image acquisition costs. Therefore, in the field of machine learning or simulation, research to generate and utilize infrared images is active. In this paper, a method for generating infrared images based on precise surface temperature calculation and its implementation result are presented. Thermal analysis was performed on a vehicle with a complex shape based on a three-dimensional conjugate heat transfer analysis based on actual physical formulas, and it was confirmed that the error of the proposed method was about 1%. As a result of the implementation, it was confirmed that an infrared image similar to an image captured by an actual infrared camera can be generated when an infrared image is generated using the accurate surface temperature and BRDF (Bi-Directional Reflectance Distribution Function).