• 제어로봇시스템학회논문지
• /
• 제22권9호
• /
• pp.738-744
• /
• 2016

In this paper, a robust visual odometry system has been proposed and implemented in an environment with dynamic illumination. Visual odometry is based on stereo images to estimate the distance to an object. It is very difficult to realize a highly accurate and stable estimation because image quality is highly dependent on the illumination, which is a major disadvantage of visual odometry. Therefore, in order to solve the problem of low performance during the feature detection phase that is caused by illumination variations, it is suggested to determine an optimal threshold value in the image binarization and to use an adaptive threshold value for feature detection. A feature point direction and a magnitude of the motion vector that is not uniform are utilized as the features. The performance of feature detection has been improved by the RANSAC algorithm. As a result, the position of a mobile robot has been estimated using the feature points. The experimental results demonstrated that the proposed approach has superior performance against illumination variations.

• 한국측량학회:학술대회논문집
• /
• 한국측량학회 2006년도 춘계학술발표회 논문집
• /
• pp.463-467
• /
• 2006

Several segmentation algorithms have been proposed for DTM generation or building modeling from airborne LiDAR data. Three components are important for accurate segmentation: (i) the adjacent relationship of n-nearest points or mesh, etc. (ii) the effective decision parameters of height, slope, curvature, and plane condition, (iii) grouping methods. In this paper, we created the topology of point cloud data using the contour tree and implemented the region-growing Terrain and non-terrain points were classified correctly in the segmented data, which can be used also for feature classification.

• 한국정밀공학회:학술대회논문집
• /
• 한국정밀공학회 1991년도 춘계학술대회 논문집
• /
• pp.309-315
• /
• 1991

This paper presents a method for automatic part feature recognition from the database of AutoCAD system for automatic process planning input. The parts considered in this study are primastic parts composed of faces perpendicular to the X, Y, Z axes and the types of features considered are through steps, blind steps, through slots, blind slots, and pockets. Features are recognized by using the concept of convex points and concave points. The software program is coded by using Turbo Pascal on the IBM PC/AT.

• 한국정보처리학회논문지
• /
• 제1권4호
• /
• pp.551-558
• /
• 1994

본 논문은 윤곽선으로 표현된 목적물 형상의 특징추출, 특징점 다각형 형상 표현, 유사도 측정 등에 관하여 연구하였다. 특징값은 윤곽선에 근사한 곡선을 만들어 주는 제어점들로 하였다. 유사도 측정실험으로 본 연구의 방법이 간편하게 형상 식별 처리 를 할 수 있음을 보였다.

• 한국정밀공학회지
• /
• 제16권7호
• /
• pp.125-131
• /
• 1999

In this paper, the algorithms which can track the two dimensional moving circular object using simple vision system are described. In order to track the moving object, the process of finding the object feature points - such as centroid of the object, corner points, area - is indispensable. With the assumption of two-dimensional circular moving object, the centroid of the circular object is computed from three points on the object circumference. Different kinds of algorithms for computing three edge points - simple x directional detection method, stick method. T-shape method are suggested. Through the computer simulation and experiments, three algorithms are compared from the viewpoint of detection accuracy and computational time efficiency.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2008년도 하계종합학술대회
• /
• pp.1013-1014
• /
• 2008

In this paper, we propose a new method of generating 3D face by using single frontal face image and 3D generic face model. By using active appearance model (AAM), the control points among facial feature points were localized in the 2D input face image. Then, the transform parameters of 3D generic face model were found to minimize the error between the 2D control points and the corresponding 2D points projected from 3D facial model. Finally, by using the obtained model parameters, 3D face was generated. We applied this 3D face to 3D game framework and found that the proposed method could make a realistic 3D face of game user.

• 한국통신학회논문지
• /
• 제22권7호
• /
• pp.1523-1533
• /
• 1997

Image registration is concerned with the establishment of correspondence between images of the same scene with translational, rotational, and scaling differences. The estimated transformation parameters between images are very important information in the field of many applications. In this paper, we propose a shape matching scheme for finding correspondence points for images with various differences, Tranditional solutions to this area are unreliable for the rotational and schaling changes between images, and the feature extraction of partially occluded scene. To solve those problems, dominant points on digital curves are detected by scale-space filtering, and initial matching is performed by similarity measure of cumulative curvatures for dominant points. For initial matching segments pairs, optimal matching points are calculated using dynamic programming.Finally, transformation parameters are estimated.

• 해양환경안전학회지
• /
• 제4권1호
• /
• pp.13-20
• /
• 1998

Since the digital chart consists of a large number of points, the effective method for the coastline data compression(CDC), storing the data compactly and reproducting the coastline feature accurately, is important. In the CDC, the key technique is to determine the optimal positions as node points in given coastlines. In this paper, a new CDC method, selecting node points with conspicuous coast positions in the view point on navigation and adopting spline interpolation to the nodes partly, is proposed. Using the northern part of KEOJE-DO coastline in Korean chart No.204, CDC experiments are carrie out with various compression ratio. The results fro the influence of coastline shape according to various CDC methods are discussed and presented.

• Korean Journal of Hydrosciences
• /
• 제5권
• /
• pp.45-55
• /
• 1994

A new method to estimate the mean areal precipitation using kriging is developed. Urlike the conventional approach, points for double and quadruple numerical integrations in the kriging equation are selected randomly, given the boundary of area of interest. This feature eliminates the conventional approach's necessity of dividing the area into subareas and calculating the center of each subarea, which in turn makes the developed method more powerful in the case of complex boundaries. The algorithm to select random points within an arbitrary boundary, based on the theory of complex variables, is described. The results of Monte Carlo simulation showed that the error associated with estimation using randomly selected points is inversely proportional to the square root of the number of sampling points.

• 인터넷정보학회논문지
• /
• 제16권3호
• /
• pp.21-32
• /
• 2015

동작인식 연구에서 비디오를 표현하는 시공간 부분 특징이 모델 없는 상향식 방식의 주요 주제가 되면서 동작 특징을 검출하고 표현하는 방법이 여러 연구를 통해 다양하게 제안되고 있다. 그 중에서 BoF(bag of features)방식은 가장 일관성 있는 인식 결과를 보여주고 있다. 비디오의 동작을 BoF로 나타내기 위해서는 어떻게 동작의 역동적 정보를 표현할 것인가가 가장 중요한 부분이다. 그래서 기존 연구에서는 비디오를 시공간 볼륨으로 간주하고 3D 동작 특징점 주변의 볼륨 패치를 복잡하게 설명하는 것이 가장 일반적인 방법이다. 본 연구에서는 기존 3D 기반 방식을 간략화하여 비디오의 동작을 BoF로 표현할 때 비디오에서 2D 특징점을 직접 수집하는 방식을 제안한다. 제안 방식의 기본 아이디어는 일반적 공간프레임의 2D xy 평면뿐만 아니라 시공간 프레임으로 불리는 시간축 평면에서 동작 특징점을 추출하여 표현하는 것으로 특징점이 비디오에서 역동적 동작 정보를 포착하기 때문에 동작 표현 특징 설명자를 3D로 확장할 필요 없이 2D 설명자만으로 간단하게 동작인식이 가능하다. SIFT, SURF 특징 표현 설명자로 표현하는 시공간 BoF 방식을 주요 동작인식 데이터에 적용하여 우수한 동작 인식율을 보였다. 3D기반의 HoG/HoF 설명자와 비교한 경우에도 제안 방식이 더 계산하기 쉽고 단순하게 이해할 수 있다.