• Journal of KIISE
• /
• v.43 no.7
• /
• pp.795-800
• /
• 2016

In image classification, recent CNNs compete with human performance. However, there are limitations in more general recognition. Herein we deal with indoor images that contain too much information to be directly processed and require information reduction before recognition. To reduce the amount of data processing, typically variational inference or variational Bayesian methods are suggested for object detection. However, these methods suffer from the difficulty of marginalizing over the given space. In this study, we propose an image-text integrated recognition system using active vision based on Spatial Transformer Networks. The system attempts to efficiently sample a partial region of a given image for a given language information. Our experimental results demonstrate a significant improvement over traditional approaches. We also discuss the results of qualitative analysis of sampled images, model characteristics, and its limitations.

• Proceedings of the Korean Association of Geographic Inforamtion Studies Conference
• /
• 2010.09a
• /
• pp.42-49
• /
• 2010

현재 대규모 실내공간이 증가하고 있고, 이에 따른 화재 등의 사고가 발생함에 따라 화재대피시스템과 같은 실시간 실내 응용에 대한 관심이 증대되고 있다. 그러나 학문적 연구 또는 상업용 시스템에서 사용되는 화재모델이내 보행자 모델은 대부분 2D 기반의 CAD와 같은 파일구조를 기반으로 하고 있으며 가상데이터를 이용한 시뮬레이션에 초점을 두고 있다. 따라서 이들을 실내 센서 등을 이용한 실시간 시스템으로 구축하기 위해서는 몇 가지 문제점들이 해결되어야 한다. 우선, 실내공간의 의미 있는 관계정보를 포함하는 위상적인 3D 모델이 필요하다. 또한, 실내 센서들에 의해 감지된 보행자의 이동을 저장하고, 저장된 데이터를 이용하여 실시간 안내를 위해서는 실내데이터 구축에 공간 DBMS를 이용해야 한다. 본 연구에서는 실내보행자 모델에서 두 가지 개선점을 제시하고자 한다. 첫째는, 간단한 3D 실내 모델구축과 공간 DBMS와 연계된 보행자 시뮬레이터를 구현하는 과정을 제시한다. 둘째는, 보행자의 가시성(visibility)에 대한 영향이 반영된, 개선된 floor field 모델을 제안한다. 이와 같은 과정을 캠퍼스 건물에 적용하고 시뮬레이션을 수행하는 과정을 예시하였다.

• Proceedings of the Korean Association of Geographic Inforamtion Studies Conference
• /
• 2006.05a
• /
• pp.187-192
• /
• 2006

도시의 과밀화로 인해 건물들은 고층화, 대형화되고 복잡한 형태를 이루고 있어 건물 이용자들에게 2차원의 정보뿐만 아니라 3차원 공간정보의 필요성이 증가하고 있으며 이를 해결하기 위해 3차원 GIS 모델의 활용이 요구되고 있다. 그러나 현재까지 연구되거나 응용되어 온 3차원 모델은 주로 건물 외부형태의 시각화를 위한 것으로 3차원 공간분석에 응용되기에는 한계가 존재해 왔다. 이에 본 연구에서는 3차원 모델을 공간분석에 적용하는 하나의 방안으로서 건물의 내부공간에서 경로탐색을 구현하기 위한 방법을 제시하였다. 이를 위해 건물 내부의 각 실들과 연결통로 및 기타 시설들을 각각 오브젝트로 분리하여 3차원으로 모델링하였다. 2차원 GIS데이터와 3차원 모델에 각각 벡터기반의 네트워크 모델을 생성하고 DB를 이용하여 두 모델을 연동함으로써 3차원 모델에서 네트워크기반의 경로분석과 탐지기능을 가능하게 하였다.

• Proceedings of the Korean Association of Geographic Inforamtion Studies Conference
• /
• 2008.06a
• /
• pp.554-558
• /
• 2008

점차 가속화 되는 현대화에 따라 복잡한 구조의 대형 시설 및 교통수단이 증가 되고 있으며 이는 화재 등과 같은 긴급 재난 상황 발생 시의 대형 피해를 초래하고 있다. 본 연구에서는 이와 같은 위험 상황에서의 피해를 최소화하기 위하여 사용되는 실시간 대피경로 안내시스템 구현을 위한 초기 과정으로 실시간으로 이동객체의 위치 정보를 획득하고 이를 건물 데이터 위해 표현하는 실시간 맵핑 시스템을 구현하였다.

• Spatial Information Research
• /
• v.21 no.6
• /
• pp.43-55
• /
• 2013

Recently, the research to visualize and to reproduce evacuation situations such as terrorism, the disaster and fire indoor space has been come into the spotlight and designing a model for interior space and reliable analysis through safety evaluation of the life is required. Therefore, this paper aims to develop simulation model which is able to suggest evacuation route guidance and safety analysis by considering the major risk factor of fire in actual building. First of all, we designed 3D-based fire and evacuation model at a subway station building in Incheon and performed fire risk analysis through thermal parameters on the basis of interior materials supplied by Incheon Transit Corporation. In order to evaluate safety of a life, ASET (Available Safe Egress Time), which is the time for occupants to endure without damage, and RSET (Required Safe Egress Time) are calculated through evacuation simulation by Fire Dynamics Simulator. Finally, we can come to the conclusion that a more realistic safety assessment is carried out through indoor space model based on 3-dimension building information and simulation analysis applied by safety guideline for measurement of fire and evacuation risk.

• Journal of Korean Society for Geospatial Information Science
• /
• v.24 no.1
• /
• pp.69-80
• /
• 2016

Some studies were conducted for the purpose of minimizing total clearance time for rapid evacuation from the indoor spaces when disaster occurs. Most studies took a long time to calculate the optimal evacuation route that derived minimum evacuation time. For this reason, this study proposes an evacuation route assignment algorithm that can shorten the total clearance time in a short operational time. When lots of exits are in the building, this algorithm can shorten the total clearance time by assigning the appropriate pedestrian traffic volume to each exit and balances each exit-load. The graph theory and greedy algorithm were utilized to assign pedestrian traffic volume to each exit in this study. To verify this algorithm, study used a cellular automata-based evacuation simulator and experimented various occupants distribution in a building structure. As a result, the total clearance time is reduced by using this algorithm, compared to the case of evacuating occupants to the exit within shortest distance. And it was confirmed that the operation takes a short time In a large building structure.

• Spatial Information Research
• /
• v.20 no.2
• /
• pp.117-127
• /
• 2012

The purpose of this study is to develop the auto-building algorithm of the Geometric Network Model(GNM), a topology model including geometric information because of the need to reflect the features' geometric characteristic into the topology model, which is for development of indoor 3D virtual model enabling queries. As the critical algorithm, the Straight Medial Axis Transformation(SMAT) algorithm is proposed in order to automatically extract the medial axis of features. The SMAT algorithm is generalized from the existing S-MAT algorithm and a range of target features where applicable is extended from simple polygons to weakly simple polygons which mean the polygons containing the inner ring inside. The GNM built automatically is finally printed out as the .csv file for easy access and w ide application in other systems. This auto-building algorithm of the GNM is available for plenty of cases such as finding a shortest path, guiding a route in emergency situation, and semantic analysis.

• Journal of Korean Society for Geospatial Information Science
• /
• v.21 no.2
• /
• pp.117-123
• /
• 2013

This study focuses on the determination of optimal photographing angle and overlapping ratio for automatic image stitching system using a non-metric camera module with motorized head. Photographing overlap was calculated considering the angle of view on different kinds of lenses and moving angle of motorized head per each option, and optimal photographing angle and overlapping ratio could be determined through the experimental result using the operating time, data volume and performance of image stitching. Through this experiment, we could find that it was effective to take a picture with $36^{\circ}$ of interval(33.82% of overlap) in vertical direction and $24^{\circ}$ or $30^{\circ}$ of interval(36.51% or 20.63% of overlap) in horizontal direction using 35mm lens for automatic image stitching system.

• 제어로봇시스템학회:학술대회논문집
• /
• 2005.06a
• /
• pp.1771-1774
• /
• 2005

Recently, RFID has become popular in the field of remote sensing applications. Location awareness is one of the most important keys to deploying RFID for advanced object tracking. Generally, multiple reference RF stations or additional sensors are used for the location sensing with RFID, but, particularly in indoor environments, spatial layout and cost problems limit the applicability of those approaches. In this paper, we propose a novel method for location sensing with active RFID systems not requiring the need for reference stations or additional sensors. The system triangulates the position of RF signal source using the signal pattern of the loop antenna connected to the power detector. The power detector consists of a signal strength detector and a signal analysis unit. The signal analysis unit indicates the signal strength and serial number using the signal from the strength detector, and provides the direction of the signal to the application target. We designed three different signal analysis units depending on the threshold type. The developed system can sense the direction to the transponder located over 10 m away within the maximum error of $5^{\circ}$. It falls within a reasonable range in our normal office environment.

• Journal of IKEEE
• /
• v.27 no.1
• /
• pp.71-77
• /
• 2023

In this paper, a dual-band stopband frequency selective surface that can be applied to Wi-Fi 6E systems is designed to block external interfering signals with adjacent operating frequency spectrum in indoor wireless LAN environments. The proposed frequency selective surface structure has frequency blocking characteristics in the 2.4GHz and 6GHz bands, and is realized through a modified crossed dipole structure and an interlocking puzzle form between unit structures. The proposed structure is designed to have stable frequency response characteristics with respect to incident angle and polarization, and the experimental results show good agreement with the simulation results for incident waves from 0° to 45°.