• Journal of the Korean Institute of Illuminating and Electrical Installation Engineers
• /
• v.12 no.3
• /
• pp.7-13
• /
• 1998

In this paper, the control technique that ID controller are autotuned according to system dynamics, driving out sample in the changeable limits of system dynamics and learning neural network, is presented. In order to lean neural network, the backpropagation learning algorithm is used and the controller parameters obtained by rule-base are used as teacher's values. When load changes, the auto-tuning of PID controller proper to system dynamics is conformed by simulation.

• Journal of Broadcast Engineering
• /
• v.7 no.2
• /
• pp.114-125
• /
• 2002

Although there have been a lot of researches on a video abstraction algorithm, the video abstraction tool is necessary to correctly pick up some important shots in a video because the important shots might vary on the person's subjectivity. This paper proposes a video abstraction and search system for making a video abstraction automatically or manually. Proposed system has two main components, server and client. The server generates the Metadata and makes a video abstraction file. The description for video abstraction file is based on MPEG-7. Since proposed system has a function to manually correct the results of each step, the author could produce a good video abstraction easily and efficiently. The proposed system could be used to build a digital video library.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.17 no.7
• /
• pp.930-938
• /
• 2007

Captions which appear in images include information that relates to the images. In order to obtain the information carried by captions, the methods for text extraction from images have been developed. However, most existing methods can be applied to captions with fixed height of stroke's width. We propose a method which can be applied to various caption size. Our method is based on connected components. And then the edge pixels are detected and grouped into connected components. We analyze the properties of connected components and build a neural network which discriminates connected components which include captions from ones which do not. Experimental data is collected from broadcast programs such as news, documentaries, and show programs which include various height caption. Experimental result is evaluated by two criteria : recall and precision. Recall is the ratio of the identified captions in all the captions in images and the precision is the ratio of the captions in the objects identified as captions. The experiment shows that the proposed method can efficiently extract captions various in size.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2008.10a
• /
• pp.532-535
• /
• 2008

The requirement of Digital Culture Content(Movie, Music, Animation, Digital TV, Exhibition and etc.) is increasing so variety and quantity of content is also increasing. The Movie what majority of the digital Content is developing of technology and data. In the result, the efficient retrieval service has required and user want to use a recommendation engine and semantic retrieval methods through the recommendation system. Therefore, this paper will suggest analysing trait element of digital content data, building of retrieval technology, analysing and retrieval technology base on KANSEI vocabulary and etc. For the these, we made a extraction technology of trait element based on semantics and KANSEI processing algorithm based on color information.

• The KIPS Transactions:PartD
• /
• v.12D no.1 s.97
• /
• pp.31-42
• /
• 2005

Generally, conventional video searching or classification methods are based on its meta-data. However, it is almost Impossible to represent the precise information of a video data by its meta-data. Therefore, a processing method of video data that is based on its meta-data has a limitation to be efficiently applied in application fields. In this paper, for efficient classification of video data, a classification method of video data that is based on its low-level data is proposed. The proposed method extracts the characteristics of video data from the given video data by clustering process, and makes the profile of the video data. Subsequently. the similarity between the profile and video data to be classified is computed by a comparing process of the profile and the video data. Based on the similarity. the video data is classified properly. Furthermore, in order to improve the performance of the comparing process, generating and comparing techniques of integrated profile are presented. A comparing technique based on a differentiated weight to improve a result of a comparing Process Is also Presented. Finally, the performance of the proposed method is verified through a series of experiments using various video data.

• Journal of the Society of Naval Architects of Korea
• /
• v.29 no.2
• /
• pp.140-149
• /
• 1992

The new method, termed the substructural eigensensitivity synthesis method, utilizes the computational merits of the component mode synthesis technique and of sensitivity analysis for the design sensitivities of the dynamic characteristics of substructurally combined structures. It is shown that the eigensensitivities of the entire structure can be obtained by synthesizing the substructural eigensolution and the sensitivities of the eigensolution for the design variables of the modifiable substructure. The sensitivities of the eigenvalues and eigenvectors obtained by the new method are compared to exact eigensolutions in terms of accuracy and computational efficiency. The small errors in eigensensitivity due to the truncation of higher modes remain within a manageable and permissible range for further analysis. The advantage of the newly proposed method as compared to the direct application of sensitivity analysis of the whole structure is demonstrated through examples.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 1998.06a
• /
• pp.131-134
• /
• 1998

본 논문에서는 시공간 적응적 기반영역 (Adaptive Spatio-Temporal Support Region : ASTSR)을 바탕으로 하는 일반화된 Auto-Regressive(AT)모델을 제안한다. 시공간 적응적 기반 영역은 영상 내 경계선의 특성과 동영상에서의 시간적 불연속 (temporal discontinuity) 개념을 이용하여 구성되어질 수 있다. 설정된 시공간 적응적 기반영역은 기존의 AR 모델에 적용되어지는 직사각형 형태의 기반영역에 비하여 보다 정상상태(stationarity)의 특성을 가지며 이로 인해 더 정확한 모델 파라미터들을 추출해 낼 수 있을 뿐 아니라 데이터의 처리량에서도 큰 이득을 얻을 수 있다. 제안된 방법은 손상된 동영상 데이터를 복원(motion picture restoration)하는 측면에 응용되어 실험되어졌으며 기존의 모델과 비교하여 우수한 성능을 보여주었다.

• Proceedings of the IEEK Conference
• /
• 2000.09a
• /
• pp.359-362
• /
• 2000

본 논문에서는 비디오 영상에서 지역적 움직임 특성을 표현할 수 있는 지역적 움직임 활동(motion activity)에 관한 기술자(descriptor)를 제안한다. 제안된 방법은 화면 전체에 대해 지역적으로 높은 움직임 활동 정도를 갖는 영역에 대한 공간적 정보를 기술하고, 카메라 움직임에 무관하게 물체의 움직임 활동 특성을 정확히 표현하기 위해 움직임 벡터의 통계적 특성과 화면 분할을 이용한다 본 논문에서 제안하는 움직임 활동의 공간적 특성을 이용하면 동영상에서 화면의 일부에서 일어나는 움직임을 이용한 검색이 가능하고, 물체 추적, 감시 시스템에서도 활용이 가능하다. 실험으로 제안한 방법을 이용해서 움직임 활동이 높은 영역의 추출과정을 보이고, 이를 이용한 검색 결과를 보인다.

• Proceedings of the Computational Structural Engineering Institute Conference
• /
• 2009.04a
• /
• pp.200-204
• /
• 2009

구조물의 건전도를 평가하기 위하여 계측된 데이터로부터 구조물의 동특성 변화를 분석하여 손상정도를 추정하는 방법이 많이 사용되고 있다. 최근, 다점 측정된 가속도 데이터로부터 구조물의 고유진동수 및 모드형상을 추출하고 이를 초기값과 비교하여 손상탐지를 실시함으로써 손상위치 및 손상정도를 추정하는 기법이 활발히 연구되고 있다. 그러나 이러한 방법을 실제 적용하기 위해서는 계측시스템 구축에 많은 비용이 소요되며, 손상탐지를 위한 해석과정이 복잡하기 때문에 실시간에 가깝게 유용한 정보를 거주자에게 제공하는데 한계가 있다. 따라서 본 연구에서는 실용적인 계측유지관리 시스템을 구축할 수 있도록 구조물의 손상도에 따른 동적특성의 변화를 사전에 예측하여 실제 계측된 동적특성에 대한 관리 한계치를 제공하는 방안을 제시하고자 한다.

• Proceedings of the Computational Structural Engineering Institute Conference
• /
• 2010.04a
• /
• pp.352-355
• /
• 2010

본 논문에서는 내부 및 외부 긴장재로 보강된 PSC 거더의 가속도 응답 특성을 이용한 구조건전성 모니터링을 위해 실험 및 수치해석 결과를 비교 분석하는 연구를 수행하였다. 첫 번째로, 내부 및 외부 긴장재로 보강된 모형 PSC 거더를 제작하였다. 두 번째로, 모형 PSC 거더의 형상, 재료 및 경계조건과 긴장재의 배치를 고려하여 초기 유한요소모델을 설계하였다. 세 번째로, 다수의 내부 및 외부 긴장력 조건하의 모형 PSC 거더에 대한 동특성 추출 실험 및 수치해석을 수행하였다. 마지막으로, 실험결과와 수치해석 결과를 비교 분석하여 가속도 응답 특성을 이용한 내부 및 외부 긴장재로 보강된 PSC 거더의 구조건전성 모니터링에 대한 적용성을 검토하였다.