• 제어로봇시스템학회논문지
• /
• 제20권11호
• /
• pp.1098-1102
• /
• 2014

Localization of aerial vehicles and map building of flight environments are key technologies for the autonomous flight of small UAVs. In outdoor environments, an unmanned aircraft can easily use a GPS (Global Positioning System) for its localization with acceptable accuracy. However, as the GPS is not available for use in indoor environments, the development of a SLAM (Simultaneous Localization and Mapping) system that is suitable for small UAVs is therefore needed. In this paper, we suggest a vision-based SLAM system that uses vision sensors and an AHRS (Attitude Heading Reference System) sensor. Feature points in images captured from the vision sensor are obtained by using GPU (Graphics Process Unit) based SIFT (Scale-invariant Feature Transform) algorithm. Those feature points are then combined with attitude information obtained from the AHRS to estimate the position of the small UAV. Based on the location information and color distribution, a Gaussian process model is generated, which could be a map. The experimental results show that the position of a small unmanned aircraft is estimated properly and the map of the environment is constructed by using the proposed method. Finally, the reliability of the proposed method is verified by comparing the difference between the estimated values and the actual values.

• 한국멀티미디어학회논문지
• /
• 제13권12호
• /
• pp.1778-1785
• /
• 2010

This paper proposes a new video-based method of analyzing human gait which is a highly variable dynamic process. It captures a human gait of varying directions as a trajectory in the phase space. The proposed method includes two options of a stochastic process model and a self-organizing feature map as the tool of feature space representation and analysis. Test results show that the model is highly intuitive and we believe it can contribute to our understanding of human activity as well as gait behavior.

• Journal of Computational Design and Engineering
• /
• 제1권1호
• /
• pp.48-54
• /
• 2014

In this paper, the authors propose a system for assisting mold designers of plastic parts. With a CAD model of a part, the system automatically determines the optimal ejecting direction of the part with minimum undercuts. Since plastic parts are generally very thin, many rib features are placed on the inner side of the part to give sufficient structural strength. Our system extracts the rib features from the CAD model of the part, and determines the possible ejecting directions based on the geometric properties of the features. The system then selects the optimal direction with minimum undercuts. Possible ejecting directions are represented as discrete points on a Gauss map. Our new point distribution method for the Gauss map is based on the concept of the architectural geodesic dome. A hierarchical structure is also introduced in the point distribution, with a higher level "rough" Gauss map with rather sparse point distribution and another lower level "fine" Gauss map with much denser point distribution. A system is implemented and computational experiments are performed. Our system requires less than 10 seconds to determine the optimal ejecting direction of a CAD model with more than 1 million polygons.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 1997년도 한국자동제어학술회의논문집; 한국전력공사 서울연수원; 17-18 Oct. 1997
• /
• pp.1589-1592
• /
• 1997

Among the prerequisite abilities (perception of environment, path planning and position estimation) of an autonomous mobile robot, position estimation has been seldom studied by mobile robot researchers. In most cases, conventional positioin estimation has been performed by placing landmarks or giving the entrire environmental information in advance. Unlikely to the conventional ones, the study addresses a new method that the robot itself can select distinctive features in the environment and save them as landmarks without any a priori knowledge, which can maximize the autonomous behavior of the robot. First, an orjentaion probaility model is applied to construct a lcoal map of robot's surrounding. The feature of the object in the map is then extracted and the map is saved as landmark. Also, presented is the position estimation method that utilizes the correspondence between landmarks and current local map. In dong this, the uncertainty of the robot's current positioin is estimated in order to select the corresponding landmark stored in the previous steps. The usefulness of all these approaches are illustrated with the results porduced by a real robot equipped with ultrasonic sensors.

• 대한전자공학회논문지SP
• /
• 제42권6호
• /
• pp.43-54
• /
• 2005

이본 논문에서는 Bayesian 영상 분할법과 SOM(Self Organization feature Map)을 이용한 텍스쳐(Texture) 분할 방법을 제안한다. SOM의 입력으로 다중 스케일에서의 웨이블릿 계수를 사용하고, 훈련된 SOM으로부터 관측 데이터에 대한 우도(尤度, likelihood)와 사후확률을 구하는 방법을 제시한다. 훈련된 SOM들로부터 구한 사후확률과 MAP(Maximum A Posterior) 분류법을 이용하여 텍스쳐 분할을 얻는다. 그리고 문맥 정보를 이용하여 텍스쳐 분할 결과를 개선하였다. 제안 방법은 HMT(Hidden Markov Tree)을 이용한 텍스쳐 분할보다 더 우수한 결과를 보여준다. 또한 SOM과 HMTseg라고 불리는 다중스케일 Bayesian 영상 분할 기법을 이용한 텍스쳐 분할 결과는 HMT와 HMTseg을 이용한 결과보다 더 우수한 성능을 보여준다.

• 한국정밀공학회지
• /
• 제21권12호
• /
• pp.100-108
• /
• 2004

In reverse engineering, data acquisition methodology can generally be categorized into contacting and non-contacting types. Recently, researches on hybrid or sensor fusion of the two types have been increasing. In addition, efficient construction of a geometric model from the measurement data is required, where considerable amount of user interaction to classify and localize regions of interest is inevitable. Our research focuses on the classification of each bounded region into a pre-defined feature shape fer a hybrid measuring scheme, where the overall procedures are described as fellows. Firstly, the physical model is digitized by a non-contacting laser scanner which rapidly provides cloud-of-points data. Secondly, the overall digitized data are approximated to a z-map model. Each bounding curve of a region of interest (featured area) can be 1.aced out based on our previous research. Then each confined area is systematically classified into one of the pre-defined feature types such as floor, wall, strip or volume, followed by a more accurate measuring step via a contacting probe. Assigned to each feature is a specific digitizing path topology which may reflect its own geometric character. The research can play an important role in minimizing user interaction at the stage of digitizing path planning.

• 스마트미디어저널
• /
• 제12권2호
• /
• pp.9-14
• /
• 2023

이 논문에서는 gradCAM를 활용한 적은 데이터로 얼굴 전체 또는 더 다양한 feature을 사용하여 얼굴인식을 할 수 있는 새로운 앙상블 방법론을 제안하였다. 인공지능 모델의 판단 근거는 gradCAM을 통하여 saliency map으로 표현될 수 있다. 따라서 본 논문에서는 학습된 얼굴인식 모델이 어느 부분에 편향적으로 관찰하여 판단했는지 gradCAM으로 시각화한다. 계산된 saliency map에서 일정 수치 이상의 돌출된 부분을 추가 모델이 학습에 사용할 수 없도록 노이즈를 추가해 데이터를 생산한다. 노이즈를 추가해서 만든 데이터로 학습할 경우 노이즈 부분을 활용하여 학습을 할 수 없으므로 새로운 얼굴 부분을 사용하여 얼굴인식 네트워크를 학습하게 된다. 기본 데이터로 학습한 네트워크와 돌출 부분에 노이즈를 추가해서 학습한 모델은 얼굴의 서로 다른 얼굴 feature을 사용할 수밖에 없고, 앙상블로 결합했을 때 얼굴의 좀 더 다양한 부분들을 사용한 임베딩 feature를 만들 수 있다. 이 논문에서 제안하는 앙상블 기법은 일반적인 앙상블 모델보다 정확도는 1.79% 상승하였고 equal error rate (EER)은 0.01788 감소하였다.

• 한국정밀공학회:학술대회논문집
• /
• 한국정밀공학회 2004년도 추계학술대회 논문집
• /
• pp.1496-1499
• /
• 2004

The paper persents an efficient method of extracting line segment in a grid map. The grid map is composed of 2-D grids that have both the occupancy and orientation probabilities based on the simplified Bayesian updating model. The probabilities and orientations of cells in the grid map are continuously updated while the robot explorers to their values. The line segments are, then, extracted from the clusters using Hough transform methods. The eng points of a line segment are evaluated from the cells in each cluster, which is simple and efficient comparing to existing methods. The proposed methods are illustrated by sets of experiments in an indoor environment.

• 제어로봇시스템학회논문지
• /
• 제19권4호
• /
• pp.290-298
• /
• 2013

Geomagnetic is refracted by building's wall and pillar. Therefore refracted geomagnetic is able to be used as feature point. In a specific space, a mobile device that is equipped with magnetic sensor array measures 3-axis magnetic field for each point. Magnetic field map is acquired by collecting the every sample point in the magnetic field. The measured magnetic field must be calibrated, because each magnetic sensor has a distortion. For this reason, sensor distortion model and sensor calibration method are proposed in this paper. Magnetic field that is measured by mobile device matches magnetic field map. Result of the matching is used for position estimation. This paper implements hardware system for position estimation method using magnetic field map.

• 전자공학회논문지
• /
• 제49권9호
• /
• pp.113-127
• /
• 2012

영상처리 기법을 이용한 얼굴검출에 관한 많은 다양한 방법들이 제시되어 왔다. 일반적으로 가장 많이 쓰이는 얼굴 검출 방식은 Viola와 Jones이 제안한 Adaboost 방식이다. 이 방식은 Haar-like feature을 이용하여 얼굴영상을 선행 학습하고, 검출 성능은 학습된 DB에 의존한다. 이는 일정 거리 범위 안의 학습된 얼굴 크기에서는 얼굴 검출을 잘 수행하지만, 카메라에서 객체(얼굴)의 거리가 멀어지면 얼굴 크기가 작아져 기존에 학습한 Haar-like feature로 얼굴 검출을 하지 못하는 경우가 발생한다. 이에 본 논문에서는 생물학 기반의 선택적 주의집중 기반의 Haar-like feature 정보를 이용한 Adaboost 모델과 사용자의 시선 응시 점 정보를 이용하여, 사용자의 관심영역 확장을 통한 원거리 얼굴 검출 모델을 제안한다. 생물학적 기반의 선택적 주의 집중 모델인 돌출맵(Saliency map) 정보를 이용하여 입력 영상에 대하여 얼굴 후보 영역을 검출하고, 검출된 얼굴 후보 영역 중에서 선행 학습된 Haar-like feature 정보로 Adaboost 알고리즘을 이용하여 최종 얼굴 영상을 검출한다. 그리고 사용자의 시선 응시 점 정보는 관심영역을 선택 하는데 이용된다. 피 실험자가, 카메라로부터 멀리 거리 떨어져 얼굴의 크기가 얼굴검출이 힘들더라도 사용자 시선 응시 점 영역을 선형 보간법으로 확대하여 입력영상으로 재사용함으로써 얼굴 검출 성능을 높일 수 있다. 제안된 방법이 기존의 Adaboost 방법보다 얼굴 검출 성능과 수행시간 면에서 우수함을 실험을 통해 확인하였다.