• Proceedings of the Korea Information Processing Society Conference
• /
• 2008.05a
• /
• pp.789-792
• /
• 2008

범죄 예방과 자원 보호를 위해 CCTV 카메라를 이용하는 방범 시스템의 필요성은 점차 커지고 있다. 아날로그 형식에서부터 디지털 형식으로 발전된 형태의 방범 시스템이 개발되고 사용 중이지만, 비용이 높고 효율성이 떨어지는 문제가 있다. 본 논문에서는 웹 서비스 기반의 서버에 적외선 카메라를 연결하고 사용자가 사전 인지 없이도 클라이언트에서 실시간으로 침입을 탐지하여 적절하게 대처할 수 있는 방범 시스템을 설계하고 구현한다.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.13 no.11
• /
• pp.2335-2340
• /
• 2009

This paper presents how to implement Markerless Augmented Reality and how to create and apply reference data sets. There are three parts related with implementation: setting camera, creation of reference data set, and tracking. To create effective reference data sets, we need a 3D model such as CAD model. It is also required to create reference data sets from various viewpoints. We extract the feature points from the mode1 image and then extract 3D positions corresponding to the feature points using ray tracking. These 2D/3D correspondence point sets constitute a reference data set of the model. Reference data sets are constructed for various viewpoints of the model. Fast tracking can be done using a reference data set the most frequently matched with feature points of the present frame and model data near the reference data set.

• The Journal of the Convergence on Culture Technology
• /
• v.7 no.4
• /
• pp.711-719
• /
• 2021

EBS has been producing numerous educational contents with traditional virtual studio production systems since the early 2000s and applied AR video production system in October 2020, twenty-years after. Although the basic concept of synthesizing graphic elements and actual image in real time by tracking camera movement and lens information is similar to the previous one but the newly applied AR video production system contains some of advanced technologies that are improved over the previous ones. Marker tracking technology that enables camera movement free and position tracking has been applied that can track the location stably, and the operating software has been applied with Unreal Engine, one of the representative graphic engines used in computer game production, therefore the system's rendering burden has been reduced, enabling high-quality and real-time graphic effects. This system is installed on a crane camera that is mainly used in a crane shot at the live broadcasting studio and applied for live broadcasting programs for children and some of the videos such as program introductions and quiz events that used to be expressed in 2D graphics were converted to 3D AR videos which has been enhanced. This paper covers the effect of introduction and application of the AR video production system on EBS content production and the future development direction and possibility.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2021.11a
• /
• pp.1184-1187
• /
• 2021

키오스크의 카메라에 촬영된 이미지를 바탕으로 사용자를 인식하여 사용자의 눈 이동을 탐지한다. 사용자가 현재 바라보고 있는 화면의 위치를 선택할 수 있도록 하여, 키오스크에 접촉하지 않음으로써 위생문제를 해결하고 직관적인 서비스를 제공한다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2021.11a
• /
• pp.1180-1183
• /
• 2021

카메라에 촬영된 이미지를 바탕으로 Mediapipe를 활용해 손을 식별하여 키오스크를 직접 만지지 않고 제어가 가능하게 함으로써, Covid-19 Pandemic 상황에 감염원을 제거하여 사용자의 코로나 감염 확률을 최소화 시킬 수 있다.

• Journal of Advanced Navigation Technology
• /
• v.9 no.1
• /
• pp.48-55
• /
• 2005

In this paper, we propose an efficient moving object tracking technique for multi-camera surveillance system. Color CCD cameras used in this system are network cameras with their own IP addresses. Input image is transmitted to the media server through wireless connection among server, bridge, and Access Point (AP). The tracking system sends the received images through the network to the tracking module, and it tracks moving objects in real-time using color matching method. We compose two sets of cameras, and when the object is out of field of view (FOV), we accomplish hand-over to be able to continue tracking the object. When hand-over is performed, we use MHI(Motion History Information) based on color information and M-bin histogram for an exact tracking. By utilizing MHI, we can calculate direction and velocity of the object, and those information helps to predict next location of the object. Therefore, we obtain a better result in speed and stability than using template matching based on only M-bin histogram, and we verified this result by an experiment.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2005.05a
• /
• pp.919-922
• /
• 2005

유비쿼터스 환경에서의 사용자의 위치인식 및 행동인식은 매우 중요하다. 인식을 하기위해 카메라를 쓰는 것은 센서를 이용하는 것에 비하여 여러 가지 장점들이 존재한다. 본 논문은 여러 대의 네트워크 카메라를 이용한 실내에서의 사용자의 위치인식 및 행동인식을 위한 시스템을 제안한다. 사용자의 위치인식, 행동인식을 위하여 시스템에서는 영상처리기법들이 사용된다. 또한 행동인식에서는 추가적으로 SVM을 이용한 학습 및 예측 방법이 사용된다.

• Journal of IKEEE
• /
• v.22 no.2
• /
• pp.281-288
• /
• 2018

In this paper, we propose an intelligent face recognition and tracking system that distributes GPU resources using CUDA. The proposed system consists of five steps such as GPU allocation algorithm that distributes GPU resources in optimal state, face area detection and face recognition using deep learning, real time face tracking, and PTZ camera control. The GPU allocation algorithm that distributes multi-GPU resources optimally distributes the GPU resources flexibly according to the activation level of the GPU, unlike the method of allocating the GPU to the thread fixedly. Thus, there is a feature that enables stable and efficient use of multiple GPUs. In order to evaluate the performance of the proposed system, we compared the proposed system with the non - distributed system. As a result, the system which did not allocate the resource showed unstable operation, but the proposed system showed stable resource utilization because it was operated stably. Thus, the utility of the proposed system has been demonstrated.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2009.10a
• /
• pp.204-207
• /
• 2009

This paper presents how to implement Markerless Augmented Reality and how to create and apply reference data sets. There are three parts related with implementation: setting camera, creation of reference data set, and tracking. To create effective reference data sets, we need a 3D model such as CAD model. It is also required to create reference data sets from various viewpoints. We extract the feature points from the model image and then extract 3D positions corresponding to the feature points using ray tracking. These 2D/3D correspondence point sets constitute a reference data set of the model. Reference data sets are constructed for various viewpoints of the model. Fast tracking can be done using a reference data set the most frequently matched with feature points of the present frame and model data near the reference data set.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2015.10a
• /
• pp.1563-1566
• /
• 2015

보안과 관련된 기술이 발전하여 대규모의 장소에 적합한 보안시스템들이 많이 개발되었다. 특히 CCTV를 이용한 감시카메라의 형태도 다양화되었다. 스마트폰의 어플리케이션이나 웹을 통해서 어디서든 감시할 수도 있어, 이를 통해 보안사고 시에 빠른 대처가 가능하다. 하지만 대규모 시스템이 아닌 경우에는 침입자 발견이 늦고, 뒤늦은 대처로 인해 큰 피해가 발생할 수 있다. 라즈베리파이, 실드 보드 등 기타 하드웨어들을 통하여 침입자를 스스로 감지하여 사용자에게 즉시 알림을 전송함으로써 보안사고에 대한 대처를 빠르고 효율적으로 할 수 있는 보안카메라를 구현하였다. 본 보안 시스템은 소리의 방향을 계산하고 정확한 방향으로의 보정을 통하여 최초 침입자를 인식한다. 이후 이미지트래킹을 통하여 침입자를 추적한다. 무선 네트워크를 이용하기 때문에 네트워크가 지원되는 어느 장소에서든지 사용이 가능하다. 대규모 보안시스템을 설치할 여건이 되기 어려운 작은 공장, 상가, 사무실 등에서 보안시스템으로 사용되면 유용할 것이다. 자세한 개발 내용은 본문에 기술한다.