• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2009.05a
• /
• pp.407-410
• /
• 2009

본 논문에서는 개선된 mean shift를 이용한 컬러 영상 분할을 소개한다. Mean shift는 Yizong Cheng에 의해 재조명되고 Dorin Comaniciu 등에 의해 정리되어 영상 필터링(image filtering), 영상 분할(image segmentation), 물체 추적(object tracking) 등 여러 응용 분야에 널리 활용되고 있다. 커널을 이용해 밀도를 추정하고 밀도가 가장 높은 점으로 커널을 연속적으로 이동함으로써 지역적으로 주요한 위치로 데이터 값을 갱신시킨다. 그러나 영상에 포함된 모든 화소에 대해 mean shift를 수행해야하기 때문에 연산 시간이 많이 소요되는 단점이 있다. 본 논문에서는 mean shift 필터링 과정을 분석하고 참조수렴방법과 강제수렴방법을 이용해 소요 시간을 단축시켰다. 모든 점에 대해 mean shift를 수행하는 대신 특정 조건을 만족하는 픽셀은 이웃 픽셀의 수렴 값을 참조하고, mean shift 과정에 진동 또는 미미한 이동을 계속하는 픽셀은 강제 수렴을 실시하였다. 개선된 방법과 기존의 mean shift 방식을 적용하여 영상 필터링과 영상 분할에 적용한 실험에서 결과 영상에는 차이가 적고 기존의 방법에 비해 수행 시간이 24% 정도 소요됨을 확인하였다.

• Annual Conference on Human and Language Technology
• /
• 2021.10a
• /
• pp.221-226
• /
• 2021

우리는 범주 불균형 분류 문제를 해결하기 위해 학습 과정 중 범주 크기 기반 배치 샘플링 방법 전환을 위한 스케줄링 방법을 제안한다. 범주별 샘플링 확률로 범주 크기의 역수(LWRS-Reciporcal)와 범주 비율의 반수(LWRS-Ratio)를 적용하여 각각 실험을 진행하였고, LWRS-Reciporcal 방법이 F1 성능 개선에 더 효과적인 것을 확인하였다. 더하여 고정된 샘플링 확률값으로 인해 발생할 수 있는 또 다른 편향 문제를 완화하기 위해 학습 과정 중 샘플링 방법을 전환하는 스케줄링 방법을 설계하였다. 결과적으로 검증 성능의 갱신 유무로 샘플링 방법을 전환하였을 때 naver shopping 데이터셋과 KLUE-TC에 대하여 f1 score와 accuracy의 성능 합이 베이스라인보다 각각 0.7%, 0.8% 향상된 가장 이상적인 성능을 보임을 확인하였다.

• Journal of the Institute of Electronics Engineers of Korea TC
• /
• v.46 no.9
• /
• pp.39-45
• /
• 2009

When Mobile IPv6 is deployed in commercial network, a mobile node needs AAA services for an authentication, authorization and accounting. AAA and Mobile IPv6 are protocols which are operated independently. Then schemes which merge these protocols have been emerged. These schemes can enable a mobile node to establish a security association between the mobile node and a home agent and to perform a binding update for the home agent using AAA authentication request. But these schemes introduce many signal messages and long handover latency during the handover, since Route Optimization mode for Mobile Ipv6 is performed using Return Routability procedure. To solve this problem, we propose a scheme for Route Optimization mode that the home agent performs the binding update for a correspondent node via the AAA infrastructure between the home agent and the correspondent node instead of Return Routability procedure. For performance evaluation, we analyze signal message transmission costs and handover latencies during handover. We show performance improvement of the proposed scheme which reduces handover latency as 61% compared with the existing scheme.

• Journal of the Korean Society for Aeronautical & Space Sciences
• /
• v.35 no.2
• /
• pp.87-93
• /
• 2007

Generally, the analysis using Computational Fluid Dynamics(CFD) is necessary for aircraft design. However, the analysis using CFD, it requires a lot of computational time and cost. But we can reduce grid reconstruction time of analyzing the various models if we use dynamic mesh. In addition, dynamic mesh can be an efficient technique in aeroelastic analysis and design optimization problem because these problems need grid reconstruction process frequently.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.15 no.3
• /
• pp.53-64
• /
• 2005

In this paper, proposes Kerberos certification mechanism that improve certification service of PKINIT base that announce in IETF CAT Working Croup. Also proposed Authentication Mechanism for reusability of Ticket that after Ticket's Lifetime is ended, message exchange that Local Client receives Remote Server's service. Since my suggestion to regional services are not described in Kerberos, authentication between regions can be performed via PKINIT(Public Key Cryptography for Initial Authentication) presented by IETF(Internet Engineering Task Force) CAT working group. The new protocol is better than the authentication mechanism proposed by IETF CAT Working group in terms of communication complexity and mechanism according to simplified Ticket issue processing.

• Journal of the Korean Society of Marine Environment & Safety
• /
• v.15 no.4
• /
• pp.355-362
• /
• 2009

In this paper, the author carried out the study to develop oil spill response training courses for competent responders examining IMO guidelines and response training courses of developed maritime countries as well as Korea. According to a result of study, fundamental training courses are to be established on the basis of IMO guideline, and also to prepare pre-arranged brief course like Basics of Spill Response for beginner. The courses should be step-by-step related with mutual continuous contents, and more times are given to on-site training for basic course and to classroom lecture with desk-top exercise for advanced course. Besides, SCAT shall be added to shoreline cleanup course, finally it is advisable that total 9 types of courses are established with additional refresh course.

• Proceedings of KOSOMES biannual meeting
• /
• 2009.06a
• /
• pp.17-24
• /
• 2009

In this paper, the authors carried out the study to develop oil spill response training courses for competent responders examining IMO guidelines, response training courses of maritime development countries and Korea response. According to a result of study, fundamental training courses are to be established on the basis of IMO guideline, and also to prepare pre-arranged brief course like Basics of Spill Response for beginner. The courses should be step-by-step related with mutual continuous contents, and more times are given to on-site training for basic course and to classroom lecture with desk-top exercise for advanced course. Besides, SCAT shall be added to shoreline cleanup course, finally it is advisable that total 9 types of courses are established with additional refresh course.

• The Journal of the Acoustical Society of Korea
• /
• v.22 no.7
• /
• pp.603-611
• /
• 2003

This paper proposes an effective feature compensation scheme based on speech model for achieving robust speech recognition. Conventional model-based method requires off-line training with noisy speech database and is not suitable for online adaptation. In the proposed scheme, we can relax the off-line training with noisy speech database by employing the parallel model combination technique for estimation of correction factors. Applying the model combination process over to the mixture model alone as opposed to entire HMM makes the online model combination possible. Exploiting the availability of noise model from off-line sources, we accomplish the online adaptation via MAP (Maximum A Posteriori) estimation. In addition, the online channel estimation procedure is induced within the proposed framework. For more efficient implementation, we propose a selective model combination which leads to reduction or the computational complexities. The representative experimental results indicate that the suggested algorithm is effective in realizing robust speech recognition under the combined adverse conditions of additive background noise and channel distortion.

• Journal of KIISE:Information Networking
• /
• v.35 no.1
• /
• pp.46-55
• /
• 2008

As a mobility support for IP have studied, Internet Engineering Task Force(IETF) standardized the Mobile IPv6(MIPv6) protocol. When a mobile node moves between subnets, MIPv6 maintains connectivity to network and supports seamless communication, and these processes are called a Handover. Whenever the mobile node moves between subnets, the Handover is performed. The mobile node can not communicate during the Handover. This period is Galled Handover latency. To reduce this latency, mipshop working group standardizes Fast Handovers for Mobile IPv6(FMIPv6), but latency which the mobile node registers its new care-of address to a home agent and a correspondent node is still long. To solve this problem, we propose a scheme that the mobile node registers the new care-of address to the home agent and initiates Return Routability procedure in advance during layer 2 handover, based on FMIPv6 and IEEE 802.11. We analyze MIPv6, FMIPv6 and the proposed scheme in term of packet transmission cost during the Handover. Compared to MIPv6 the proposed scheme gains 79% improvement, while it gains 31% improvement compared to FMIPv6.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.41 no.5
• /
• pp.23-34
• /
• 2004

In this paper we propose a feature point tracking algorithm using optical flow under non-prior taming active feature model (NPT-AFM). The proposed algorithm mainly focuses on analysis non-rigid objects[1], and provides real-time, robust tracking by NPT-AFM. NPT-AFM algorithm can be divided into two steps: (i) localization of an object-of-interest and (ii) prediction and correction of the object position by utilizing the inter-frame information. The localization step was realized by using a modified Shi-Tomasi's feature tracking algoriam[2] after motion-based segmentation. In the prediction-correction step, given feature points are continuously tracked by using optical flow method[3] and if a feature point cannot be properly tracked, temporal and spatial prediction schemes can be employed for that point until it becomes uncovered again. Feature points inside an object are estimated instead of its shape boundary, and are updated an element of the training set for AFH Experimental results, show that the proposed NPT-AFM-based algerian can robustly track non-rigid objects in real-time.