• 방송공학회논문지
• /
• 제24권2호
• /
• pp.217-226
• /
• 2019

본 논문에서는 가이디드 영상 필터 (guided image filter: GIF)를 이용하여 컨볼루션 신경망 (convolutional neural network; CNN)을 이용한 역 톤 매핑 (inverse tone-mapping) 기법의 결과를 향상시킬 수 있는 필터링 기법을 제안한다. 저동적범위 (low dynamic range; LDR) 영상을 고동적범위 (high dynamic range; HDR) 디스플레이에서 표현할 수 있도록 변환하는 역 톤 매핑 기법은 지속적으로 제안되어왔다. 최근 들어 컨볼루션 신경망을 이용하여 단일 LDR 영상을 HDR 영상으로 변환하는 알고리듬이 많이 연구되었다. 그 중엔 제한된 동적범위 (dynamic range)로 인해 화소가 포화되어 기존 화소 정보가 손실되는데 이를 학습된 컨볼루션 신경망을 이용해서 복원하는 알고리듬이 존재한다. 해당 알고리듬은 비포화 영역의 잡음까지는 억제하지 못하며 포화 영역의 디테일까지는 복원하지 못한다. 제안한 알고리듬은 입력 영상에 가중된 가이디드 영상 필터 (weighted guided image filter; WGIF)를 사용해서 비포화 영역의 잡음을 억제하고 포화 영역의 디테일을 복원시킨 다음 컨볼루션 신경망에 인가하여 최종 결과 영상의 품질을 개선하였다. 제안하는 알고리듬은 HDR 정량적 화질평가 지표를 측정하였을 때 기존의 알고리듬에 비해 높은 화질평가 지수를 나타내었다.

• 전력전자학회:학술대회논문집
• /
• 전력전자학회 2007년도 하계학술대회 논문집
• /
• pp.26-28
• /
• 2007

Design equations of II-SEPIC (Isolated Inverse-SEPIC) are derived and its dynamic model is also made using state-space averaging method in this paper. II-SEPIC is comprehensively investigated through the frequency response analysis, compared with those of I-SEPIC (Isolated SEPIC).

• Structural Engineering and Mechanics
• /
• 제37권4호
• /
• pp.351-365
• /
• 2011

This study proposes an inverse estimation method for the input forces of a fixed beam structural system. The estimator includes the fuzzy Kalman Filter (FKF) technology and the fuzzy weighted recursive least square method (FWRLSM). In the estimation method, the effective estimator are accelerated and weighted by the fuzzy accelerating and weighting factors proposed based on the fuzzy logic inference system. By directly synthesizing the robust filter technology with the estimator, this study presents an efficient robust forgetting zone, which is capable of providing a reasonable trade-off between the tracking capability and the flexibility against noises. The period input of the fixed beam structure system can be effectively estimated by using this method to promote the reliability of the dynamic performance analysis. The simulation results are compared by alternating between the constant and adaptive and fuzzy weighting factors. The results demonstrate that the application of the presented method to the fixed beam structure system is successful.

• 한국지진공학회:학술대회논문집
• /
• 한국지진공학회 2002년도 춘계 학술발표회 논문집
• /
• pp.125-132
• /
• 2002

In this study, a numerical method is developed for nonlinear analysis for soil-pile-structure interaction system in time domain. Finite elements considering material nonlinearity are used for the near field and boundary elements for the far field. In the near field, frame elements are used for modeling a pile and plane-strain elements for surrounding soil and superstructure. In. the far field, boundary element formulation using the dynamic fundamental solution is adopted and coupled with the near field. Transformation of stiffness matrices of boundary elements into time domain is performed by inverse FFT. Stiffness matrices in the near field and far field are coupled. Newmark direct time integration method is applied. Developed soil-pile-structure interaction analysis method is verified with available literature and commercial code. Also, parametric studies by developed numerical method are performed. And seismic response analysis is performed using actual earthquake records.

• 방송공학회논문지
• /
• 제20권5호
• /
• pp.756-769
• /
• 2015

본 논문에서는 SHVC (Scalable High Efficiency Video Coding)의 계층적 부호화 구조를 기반으로 HDR (High Dynamic Range) 영상과 SDR (Standard Dynamic Range) 영상의 부호화 방법을 제안한다. 본 논문의 제안하는 방법은 SHVC 계층적 부호화 구조 중 기본 계층에서 SDR 영상을 부호화 및 복호화하여, 기존의 디스플레이 장치에 대한 하위 호환성을 제공하고, 향상 계층에서 HDR 영상에 대한 부호화 및 복호화를 수행한다. 또한, 기본 계층의 복원 영상에 전역 역 톤 맵핑 (Global inverse tone mapping)을 수행하여 향상 계층의 참조 영상으로 사용함으로써 HDR 영상의 부호화 효율을 향상시켰다. 본 논문에서 제안하는 방법을 통해 기존의 SHVC 레퍼런스 소프트웨어인 SHM7.0 대비 약 향상 계층에서 평균 43%, 최대 76.3%의 BD-Bitrate 감소를 얻을 수 있었으며, 기본 계층과 향상 계층 전체에서 평균 15.7%, 최대 31%의 BD-Bitrate 감소를 보였다.

• Journal of Mechanical Science and Technology
• /
• 제18권9호
• /
• pp.1519-1528
• /
• 2004

Atmospheric Re-Entry guidance is divided as longitudinal and lateral. This paper proposes a longitudinal reference trajectory and control law using the inverse dynamics method with pseudospectral Legendre method. Application of this method into Re-Entry problem forces a power of calculation time-reduction due to unnecessary of integration or any iteration as well as sufficient accuracy convergence. The used guidance scheme is time-to-go.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 2001년도 ICCAS
• /
• pp.33.1-33
• /
• 2001

We address a novel approach to identify a nonlinear dynamic system for Wiener models, which are composed of a linear dynamic system part followed by a nonlinear static part. The aim of system identification here is to provide the optimal mathematical model of both the linear dynamic and the nonlinear static parts in some appropriate sense. Assuming the nonlinear static part is invertible, we approximate the inverse function by a piecewise linear function. We estimate the piecewise linear inverse function by using the evolutionary computation approach such as genetic algorithm (GA) and evolution strategies (ES), while we estimate the linear dynamic system part by the least squares method. The results of numerical simulation studies indicate the usefulness of proposed approach to the Wiener model identification.

• 한국전산유체공학회:학술대회논문집
• /
• 한국전산유체공학회 2002년도 추계 학술대회논문집
• /
• pp.35-40
• /
• 2002

This paper deals with a thrust generation on flapping-airfoil by dynamic stall. Dynamic stall refers to a series of complicated aerodynamic phenomena accompanied by a stall delay in unsteady motion. In most cases, once it occurs, the dynamic stall may lead to an abrupt fluctuation of aerodynamic forces. An inverse $k\acute{a}rm\acute{a}n$ vortex has been considered as a main reason for a thrust generation. In this paper, however, we have found out that a thrust is closely related to reduced frequency and leading edge vortex in addition to inverse Karman vortex. In order to certify our opinion, picking and plunging motions were calculated with the parameter of amplitude and frequency by using the unsteady, incompressible Navier-Stokes flow solver with a two-equation turbulence model. For more efficient computation, it is parallelized by MPI programming method.

• Structural Engineering and Mechanics
• /
• 제15권6호
• /
• pp.735-752
• /
• 2003

The quasi-static and dynamic responses of a linear viscoelastic circular beam on Winkler foundation are studied numerically by using the mixed finite element method in transformed Laplace-Carson space. This element VCR12 has 12 independent variables. The solution is obtained in transformed space and Schapery, Dubner, Durbin and Maximum Degree of Precision (MDOP) transform techniques are employed for numerical inversion. The performance of the method is presented by several quasi-static and dynamic example problems.

• Journal of Mechanical Science and Technology
• /
• 제17권4호
• /
• pp.538-544
• /
• 2003

Although many methodologies exist for determining the constrained equations of motion, most of these methods depend on numerical approaches such as the Lagrange multiplier's method expressed in differential/algebraic systems. In 1992, Udwadia and Kalaba proposed explicit equations of motion for constrained systems based on Gauss's principle and elementary linear algebra without any multipliers or complicated intermediate processes. The generalized inverse method was the first work to present explicit equations of motion for constrained systems. However, numerical integration results of the equation of motion gradually veer away from the constraint equations with time. Thus, an objective of this study is to provide a numerical integration scheme, which modifies the generalized inverse method to reduce the errors. The modified equations of motion for constrained systems include the position constraints of index 3 systems and their first derivatives with respect to time in addition to their second derivatives with respect to time. The effectiveness of the proposed method is illustrated by numerical examples.