• 정보과학회 논문지
• /
• 제43권8호
• /
• pp.878-882
• /
• 2016

이미지의 내용을 설명하는 캡션을 자동으로 생성하는 기술은 이미지 인식과 자연어처리 기술을 필요로 하는 매우 어려운 기술이지만, 유아 교육이나 이미지 검색, 맹인들을 위한 네비게이션 등에 사용될 수 있는 중요한 기술이다. 본 논문에서는 이미지 캡션 생성을 위해 Convolutional Neural Network(CNN)으로 인코딩된 이미지 정보를 입력으로 갖는 이미지 캡션 생성에 최적화된 Recurrent Neural Network(RNN) 모델을 제안하고, 실험을 통해 본 논문에서 제안한 모델이 Flickr 8K와 Flickr 30K, MS COCO 데이터 셋에서 기존의 연구들보다 높은 성능을 얻음을 보인다.

• 한국컴퓨터정보학회:학술대회논문집
• /
• 한국컴퓨터정보학회 2008년도 제38차 하계학술발표논문집 16권1호
• /
• pp.205-209
• /
• 2008

본 논문은 입력되는 영상에서 특정 객체를 찾기 위하여 특징 검출 및 매칭 결과를 분석하여 기술한다. 영상의 특징을 추출하는 방법 중 코너를 특징으로 하는 방법인 해리스 코너 검출(Harris corner detection)을 이용하여 코너를 추출하였으며, 추출한 특징을 이용하여 다양한 크기의 템플릿을 만들어 입력된 영상과 상관계수를 구해 최대값을 가지는 위치를 찾아 입력된 영상과 객체를 매칭 시킨 결과를 분석하였다. 본 논문의 연구 결과들은 객체의 탐지 등과 같은 영상 분석 기반 기술에 활용될 수 있으리라 기대된다.

• 한국항만학회지
• /
• 제13권2호
• /
• pp.381-388
• /
• 1999

To improve the productivity in the harbor, successful development of an UCT(Unmanned Container Transporter) is needed. Well-designed steering and velocity control systems are the key factor for the development of the UCT. In this paper, a research concerning the achievement of the steering control is introduced. To get an information on the guide line that the UCT should track, the vision system is applied. By using neural network, proper steering angle is gotten fast with less influence of the image disturbance. A simulation based on the UCT kinematics is performed with the measured steering angle, and it shows satisfactory results.

• 전기학회논문지
• /
• 제65권12호
• /
• pp.2138-2144
• /
• 2016

People with visual disabilities suffer environmentally, socially, and technologically. Navigating through places and recognizing objects are already a big challenge for them who require assistance. This study aimed to develop an android-based assistive application for the visually impaired. Specifically, the study aimed to create a system that could aid visually impaired individuals performs significant tasks through object recognition and identifying locations through GPS and Google Maps. In this study, the researchers used an android phone allowing a visually impaired individual to go from one place to another with the aid of the application. Google Maps is integrated to utilize GPS in identifying locations and giving distance directions and the system has a cloud server used for storing pinned locations. Furthermore, Haar-like features were used in object recognition.

• 한국컴퓨터그래픽스학회논문지
• /
• 제6권4호
• /
• pp.15-22
• /
• 2000

영상기반 모델링 및 렌더링의 주된 흐름중의 하나는 plenoptic function을 어떻게 구현할 것인가에 집중되고 있다. 본 논문은 임의의 위치에 배치된 파노라마 군을 이용하여 plenoptic function을 구현하는 새로운 기법을 제안한다. 이 기법은 우선 간단한 컴퓨터 비전 기술을 사용하여 주변 환경에 대한 개괄적인 깊이 정보를 추출한 뒤, 이를 이용하여 파노라마 영상 데이터를 보정/보간하여 새로운 시점에서의 렌더링을 수행한다. 본 기법을 사용한 결과 실시간에 부드러운 네비게이션을 구현할 수 있었다.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 2001년도 하계학술대회 논문집 D
• /
• pp.2729-2731
• /
• 2001

In this paper, we studied on the real-time environment map building for the mobile robot navigation using the stereo camera system. Distance measurement are necessary to build the environment map. We used a area-based stereo matching for the distance measurement with the stereo camera system. To reduce the computation time, we used DSP processor on the vision board, took a suitable area size for stereo matching, and used hierarchical search method. Using the fast acquired distance values, the environment map was built.

• 한국항만학회지
• /
• 제14권2호
• /
• pp.199-207
• /
• 2000

In this study, as the preliminary step for developing an unmanned vehicle to deliver a container-box, we designed and implemented Automatic Guided Vehicle(AGV) Simulator for the purpose of Port Facilities Automation. It is preferable to research the intelligent AGV for delivery all day long. For complementing AGV simulator driving, we used multiple-sensor systems with vision, ultrasonic, IR and adapted the high-speed wireless LAN that satisfies the IEEE 802.11 Standard for bi-directional communication between main processor in AGV and Host computer. Here, we mounted on bottom frame in AGV Pentium-III processor, which combine and compute the information from each sensor system and control the AGV driving, and used the 80C196KC micro-controller to control the actuating and steering motors.

• 대한의용생체공학회:학술대회논문집
• /
• 대한의용생체공학회 1996년도 추계학술대회
• /
• pp.95-98
• /
• 1996

In this paper, The vehicle's path construction method for motorized wheelchair's autonomous navigation in a building through analysis of a corridor image using vision system has been proposed and We detected lines of vertical axis through camera distortion parameter, which was measured by camera calibration in a corridor image. Then we got the feature points in the lines. We analyzed the distance of feature points and what is feature points. we reconstructed corridor image to vehicle's path.

• 한국군사과학기술학회지
• /
• 제11권4호
• /
• pp.107-115
• /
• 2008

Accurate extraction of traversable region is a critical issue for autonomous navigation of unmanned ground vehicle(UGV). This paper introduces enhanced extraction of traversable region by combining scene clustering with 3D world modeling using CCD(Charge-Coupled Device)/IR(Infra Red) image. Scene clustering is developed with K-means algorithm based on CCD and IR image. 3D world modeling is developed by fusing CCD and IR stereo image. Enhanced extraction of traversable regions is obtained by combining feature of extraction with a clustering method and a geometric characteristic of terrain derived by 3D world modeling.

• 한국군사과학기술학회지
• /
• 제13권4호
• /
• pp.683-688
• /
• 2010

The ability to navigate autonomously in off-road terrain is the most critical technology needed for Unmanned Ground Vehicles(UGV). In this paper, we present a method for vision-based terrain cover classification using DCT features. To classify the terrain, we acquire image from a CCD sensor, then the image is divided into fixed size of blocks. And each block transformed into DCT image then extracts features which reflect frequency band characteristics. Neural network classifier is used to classify the features. The proposed method is validated and verified through many experiments and we compare it with wavelet feature based method. The results show that the proposed method is more efficiently classify the terrain-cover than wavelet feature based one.