• The Journal of the Korea institute of electronic communication sciences
• /
• v.9 no.6
• /
• pp.649-656
• /
• 2014

In this paper, we address the problem of center lane detection algorithm for autonomous driving. Color information for center lane is gathered by analyzing a row line color distribution of road in front of a vehicle. The candidate pixels for center lane are extracted from the histogram of road colors. Morphological filtering and clustering process are applied to the candidate pixels to extract the exact center lane. We predict a expected area of center lane and search only the regions in subsequent frames, that reduces the time required for center lane detection.

• Proceedings of the KSPS conference
• /
• 2003.05a
• /
• pp.127-130
• /
• 2003

This paper describes an efficient method for unsupervised speaker adaptation. This method is based on selecting a subset of speakers who are acoustically close to a test speaker, and calculating adapted model parameters according to the previously stored sufficient HMM statistics of the selected speakers' data. In this method, only a few unsupervised test speaker's data are required for the adaptation. Also, by using the sufficient HMM statistics of the selected speakers' data, a quick adaptation can be done. Compared with a pre-clustering method, the proposed method can obtain a more optimal speaker cluster because the clustering result is determined according to test speaker's data on-line. Experiment results show that the proposed method attains better improvement than MLLR from the speaker independent model. Moreover the proposed method utilizes only one unsupervised sentence utterance, while MLLR usually utilizes more than ten supervised sentence utterances.

• Journal of IKEEE
• /
• v.16 no.2
• /
• pp.95-101
• /
• 2012

A novel approach for the reconstruction of 3D building model from aerial image data is proposed in this paper. In this approach, a Centroid Neural Network (CNN) with a metric of line segments is proposed for connecting low-level linear structures. After the straight lines are extracted from an edge image using the CNN, rectangular boundaries are then found by using an edge-based grouping approach. In order to avoid producing unrealistic building models from grouping lined segments, a hierarchical grouping method is proposed in this paper. The proposed hierarchical grouping method is evaluated with a set of aerial image data in the experiment. The results show that the proposed method can be successfully applied for the reconstruction of 3D building model from satellite images.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.67 no.6
• /
• pp.773-781
• /
• 2018

In recent years, a study of power-line inspection using an unmanned aerial vehicle (UAV) has been actively conducted. However, relevant studies have been conducting power-line inspection with an UAV operated by manual control, and they have developed just power-line detection algorithm on aerial images. To overcome limitations of existing research, we propose a drone-based power-line tracking system in this paper. The main contributions of this paper are to operate developed system under configured environment and to develop a power-line detection algorithm in real-time. Developed system is composed of the power-line detection and the image-based tracking control. To detect a power-line in real-time, a region of interest (ROI) image is extracted. Furthermore, clustering algorithm is used in order to discriminate the power-line from background. Finally, the power-line is detected by using the Hough transform, and a center position and a tilt angle are estimated by using the Kalman filter to control a drone smoothly. We design a position controller and an attitude controller for image-based tracking control, and both controllers are designed based on the proportional-derivative (PD) control method. The interaction between the position controller and the attitude controller makes the drone track the power-line. Several experiments were carried out in environments where conditions are similar to actual environments, which demonstrates the superiority of the developed system.

• Journal of Institute of Control, Robotics and Systems
• /
• v.4 no.1
• /
• pp.123-133
• /
• 1998

In this paper, an algorithm of fault detection and diagnosis during operation of induction motors under the condition of various loads and rates is investigated. For this purpose, the spectrum pattern of input currents is used in monitoring the state of induction motors, and by clustering the spectrum pattern of input currents, the newly occurrence of spectrum patterns caused by faults are detected. For the diagnosis of the fault detected, a fuzzy fault tree is designed, and the fuzzy relation equation representing the relation between an induction motor fault and each fault type, is solved. The solution of the fuzzy relation equation shows the possibility of occurence of each fault. The results obtained are summarized as follows : (1) Using clustering algorithm by unsupervised learning, an on-line fault detection method unaffected by the characteristics of loads and rates is implemented, and the degree of dependency for experts during fault detection is reduced. (2) With the fuzzy fault tree, the fault diagnosis process become systematic and expandable to the whole system, and the diagnosis for sub-systems can be made as an object-oriented module.

• Journal of KIISE:Computer Systems and Theory
• /
• v.34 no.12
• /
• pp.665-682
• /
• 2007

Line drawing is a widely used style in non-photorealistic rendering because it generates expressive descriptions of object shapes with a set of strokes. Although various techniques for line drawing of 3D objects have been developed, line drawing of 2D images has attracted little attention despite interesting applications, such as image stylization. This paper presents a robust and effective technique for generating line drawings from 2D images. The algorithm consists of three parts; filtering, linking, and stylization. In the filtering process, it constructs a likelihood function that estimates possible positions of lines in an image. In the linking process, line strokes are extracted from the likelihood function using clustering and graph search algorithms. In the stylization process, it generates various kinds of line drawings by applying curve fitting and texture mapping to the extracted line strokes. Experimental results demonstrate that the proposed technique can be applied to the various kinds of line drawings from 2D images with detail control.

• Tunnel and Underground Space
• /
• v.13 no.1
• /
• pp.33-43
• /
• 2003

Field investigations of the orientations of discontinuity planes inside the borehole for designing the underground rock structures have been depend solely on the borehole image-taking techniques. But, borehole image-taking has to be processed after the completion of drilling operation and also requires the handling of highly expensive apparatus so that practical application is very restricted. In this study Discontinuity Orientation Measurement (DOM) drilling system and discontinuity analysis model RoSA-DOM are developed to acquire the reliable information of rock structure by analyzing the characteristics of joint distribution. DOM drilling system retrieves the rock core on which the reference line of pre-fixed drilling orientation is engraved. Coordinates of three arbitrary points on the joint surface relative to the position of reference line are assessed to determine the orientation of joint plane. The position of joint plane is also allocated by calculating the location of core axis at which joint plane is intersected. Then, the formation of joint set is analyzed by utilizing the clustering algorithm. Total and set spacings are calculated by considering the borehole axis as the scanline. Engineering applicability of in-situ rock mass around the borehole is also estimated by calculating the total and regional RQDs along the borehole axis.

• The Journal of the Korea Contents Association
• /
• v.6 no.4
• /
• pp.46-54
• /
• 2006

We propose an optimization method for PCB assembly lines including multiple surface mounters. To increase the productivity of PCB assembly line, the component allocation, feeder assignment, and assembly sequence of each surface mounter should be optimized. The optimization Problem is formulated as an integer programming problem. We divide the overall problem into two hierarchical sub-problems: forward-path problem and backward-path problem. The clustering algorithm and branch-and-bound algorithm are applied to solve the forward-path problem. The assignment algorithm and connection algorithm are applied to solve the backward-path problem. Simulation results are presented to verify the usefulness of the proposed method.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2013.07a
• /
• pp.341-342
• /
• 2013

본 연구에서는 배전선로의 내부 부식을 진단하는 방법이 미흡하여 예기치 못한 사고가 빈번히 발생하는 현실을 감안하여 이에 대한 대책으로 와류탐상법을 이용하여 배전선로의 내부 부식 신호를 검출한 후, 와류 탐상 신호가 매우 민감한 신호임을 고려하여 내부 부식에 대한 보다 정확한 분석을 할 수 있도록 패턴 해석 방법인 군집화기법을 이용하여 와류 탐상 신호에 대한 시뮬레이션을 수행하였으며, 배전선로의 열화 정도를 제시하기 위해 물성 변화 및 인장력을 각각 조사하였다.

• The KIPS Transactions:PartB
• /
• v.13B no.4 s.107
• /
• pp.401-408
• /
• 2006

In this paper, we propose a new class of segmentation technique for feature extraction based on the statistical and regional classification at each vertical or horizontal line of digital image data. Data is processed and clustered at each line, different from the point or space process. They are designed to segment gray-scale sectional images using a horizontal and vertical line process due to their statistical and property differences, and to extract the feature. The techniques presented here show efficient results in case of the gray level overlap and not having threshold image. Such images are also not easy to be segmented by the global or local threshold methods. Line pixels inform us the sectionable data, and can be set according to cluster quality due to the differences of histogram and statistical data. The total segmentation on line clusters can be obtained by adaptive extension onto the horizontal axis. Each processed region has its own pixel value, resulting in feature extraction. The advantage and effectiveness of the line-cluster approach are both shown theoretically and demonstrated through the region-segmental carotid artery medical image processing.