http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
- 中文 을 입력하시려면 zhongwen을 입력하시고 space를누르시면됩니다.
- 北京 을 입력하시려면 beijing을 입력하시고 space를 누르시면 됩니다.
RISS 인기검색어
- 검색키워드
- 저자 : 장훈석 (검색결과 19 건)
- 무료
- 기관 내 무료
- 유료
-
직업재활시설 종사자와 근로장애인의 직무만족도에 관한 연구
장훈석,나운환,박세진 대구대학교 특수교육재활과학연구소 2007 특수교육재활과학연구 Vol.46 No.3
The main thrust of this research is to provide more satisfactory vocational rehabilitation service for consumers, workers with disabilities in vocational rehabilitation facilities, as well as to present a baseline data to enhance job satisfaction of rehabilitation counselors in vocational rehabilitation facilities, by clarifying the relationships of the job satisfactions for the two groups. The participants of 178 rehabilitation counselors and 267 workers with disabilities were recruited from 89 facilities nationwide, in which individuals with physical disabilities, cerebral palsy, traumatic brain injury were working. The results of this research are as follows: The following were found about workers with disabilities who worked in vocational rehabilitation facilities. First, the group with high school degree showed higher satisfaction in coworker factors than those with elementary school education did. In addition, an unmarried group had more satisfaction in facility factors than a divorced or widowed group was. Also, people in their 20s showed more satisfaction with wage than those in their 30s and 40s. Lastly, individuals working in sheltered workshops indicated higher satisfaction with promotion than those in work activity center.
-
장훈석 한국정보전자통신기술학회 2022 한국정보전자통신기술학회논문지 Vol.15 No.4
관제 및 측정 시스템 등에서 고품질 영상을 획득하는 것은 중요한 요소 중 하나이다. 영상 획득 기술 중에서 SFF (Shape from Focus)는 물체를 광축 방향으로 사전에 결정된 스텝 사이즈로 이동시킴으로써 초점 정도가 다른 2D 영상들을 획득함으로써 삼차원 형상을 복원하는 기술이다. SFF를 위해 일정한 스텝 사이즈로 물체를 이동시킬 때 각 스텝에서 광축 방향으로 기계 진동, 즉 지터 노이즈가 발생한다. 본 논문에서는 이 지터 노이즈의 영향을 줄이기 위해 새로운 상태 추정 필터를 설계 및 적용한다. 제안된 방법의 적용을 위해 지터 노이즈와 초점 커브는 가우시안 함수로써 모델링한다. 실험 결과들을 통해 제안된 방법의 효율성을 증명한다.
-
장훈석(Hoon-Seok Jang) 한국정보기술학회 2021 한국정보기술학회논문지 Vol.19 No.9
Robust pupil detection is essential for binocular gaze estimation in the augmented reality. Since it is important to acquire a clear pupil image for accurate pupil detection, multi-focus image fusion has received a lot of attention. Multi-focus image fusion is a technique for extracting a focused region from the blurred images to obtain a image with high sharpness. In multi-focus image fusion, activity level measurement and fusion rule are important factors. Since the existing multi-focus image fusion not only considers activity level measurement and fusion rule separatively, but also becomes more and more complicated, a convolutional neural network as one of the deep learning techniques is proposed in this paper to obtain a multi-focus image fusion results with excellent performance through directly mapping between the blurred images and the focus map. Experimental results demonstrate the effectiveness of the proposed method.
-
장훈석(Hoon-Seok Jang) 한국정보기술학회 2021 한국정보기술학회논문지 Vol.19 No.8
Obtaining the 3D information such as 3D gaze is a fundamental purpose of research in computer vision. Shape from Focus(SFF) acquires 3D information by using 2D images with different focus levels. In each step of image acquisition, mechanical vibration, referred as jitter noise, occurs. Since the jitter noise changes the focus values of 2D images, it significantly degrades the performance of 3D information estimation. In this paper, a new filtering method which reduces the effects of jitter noise in 3D information estimation is proposed. At first, jitter noise and focus curves are modeled as Lé vy distribution and quadratic function, respectively. Next, Long Short-Term Memory (LSTM) network as one of the deep learning techniques is applied to the modeled focus curves for reducing the jitter noise. Experimental results demonstrate the effectiveness of the proposed method.
-
스마트팜 분야에서 고성능 삼차원 형상 복원을 위한 최적의 영상 획득 기술
장훈석(Hoon-Seok Jang) 한국정보기술학회 2021 Proceedings of KIIT Conference Vol.2021 No.6
스마트팜 분야에서 객체의 삼차원 형상을 정밀하게 복원하기 위해, Shape from Focus (SFF)는 효과적인 옵션들 중에 하나이다. SFF에서 다른 초점 레벨들을 가진 일련의 2D 영상들이 광축을 따라 일정한 스텝 사이즈로 획득된다. SFF를 위해 많은 2D 영상들을 사용하는 것은 정확한 형상 복원을 제공한다. 하지만, 2D 영상들을 획득하고 처리하는데 높은 계산 비용을 요구한다. 이 문제를 해결하기 위해 기존의 SFF 방법들보다 더 적은 2D 영상들을 획득하여 사용하는 가우시안 프로세스 회귀 (GPR) 방법이 제안된다. 이 방법은 광축을 따라 존재하는 픽셀들의 초점 값들에 최적의 커브들을 적합하기 위해 적용된다. 실험 결과들은 제안된 방법의 효과성을 증명한다. In order to accurately recover the 3D shape of an object in the field of smart farm, Shape from Focus (SFF) is one of the effective options. In SFF, a series of 2D images with different focus levels are obtained in the constant step size along the optical axis. Using many 2D images for SFF results in accurate shape recovery. However, it requires high computational cost in acquiring and processing the 2D images. To solve the problem, a Gaussian process regression (GPR) method that uses fewer 2D images is proposed. It is applied to fit optimal curves to focus values of pixels along the optical axis. Experimental results demonstrate the effectiveness of the proposed method.
-
-
장훈석(Hoon-Seok Jang) 한국정보기술학회 2022 한국정보기술학회논문지 Vol.20 No.7
In order to implement the digital twin, obtaining the 3D image of an object is one of the essential elements. In order to utilize SFF, which is one of the acquisition techniques of 3D information, a sequence of 2D images with various focus levels should be achieved through movement in the direction of the optical axis of an object on the translational stage. Since the acquisition interval of 2D images affects the performance of 3D shape recovery, it is important to set an appropriate interval. In this manuscript, a Gaussian process regression(GPR) method is proposed to obtain the 3D shape of an object with the small amount of 2D images. It is applied to fit the initial focus curves, which are obtained by using one of focus measure operators. Experimental results demonstrate the effectiveness of the proposed method.
-
-
-
국소 가중 비모수 회귀 기술을 사용한 스마트팜 영상 센서 정보 수집
장훈석(Hoon-Seok Jang) 한국정보기술학회 2022 Proceedings of KIIT Conference Vol.2022 No.6
스마트팜 영상 센서 정보를 수집하여 시각화하는 것은 디지털 트윈 구현에 있어서 중요한 요소들 중 하나이다. 시각화 방법들 중 하나인 Shape from Focus (SFF)는 광축을 따라 객체를 일정한 스텝 사이즈로 이동시킴으로써 다른 초점 정도를 가지는 2D 영상들을 사용하여 삼차원 형상을 복원하는 기술이다. 획득된 이미지 시퀀스에서 영상 품질을 계산하기 위해 초점 측정 연산자 적용이 필요하다. 그러나, 초점 측정 연산자는 영상 프레임들 사이에 정보 손실 때문에 정확한 초점값을 제공하지 않는다. 본 논문에서는 초점 측정 연산자를 통해 획득된 초점 커브들을 피팅하는 국소 가중 비모수 회귀 기술을 제안한다. 실험 결과들을 통해 제안된 방법의 효율성을 증명한다. Collecting and visualizing the information of image sensor for smart farm is one of the important factors in implementing a digital twin. Shape from Focus (SFF), one of the visualization methods, is 3D shape reconstruction technology using 2D images with different focus levels obtained by moving the object at a constant step size along the optical axis. Focus measure operators are needed to compute the focus quality in the acquired image sequence. However, focus measure operators don’t provide the accurate focus values due to information loss between consecutive image frames. In this paper, locally weighted non-parametric regression (LOREG), which fits the focus curves obtained by one of the focus measure operators, is proposed. Experimental results demonstrate the effectiveness of the proposed method.
DBpia연관 검색어 추천
이 검색어로 많이 본 자료
활용도 높은 자료
