• 한국광학회지
• /
• 제20권1호
• /
• pp.16-22
• /
• 2009

본 논문에서는 2차원 동영상을 normalized cut과 optical flow를 이용하여 3차원 동영상으로 변환하는 방법을 제안하였다. 이를 통해 특정 디스플레이 장치와 특정 동영상 포맷에 국한되지 않는 2차원 동영상의 3차원 동영상 변환 방법을 제안하였다. 본 연구에서는 2차원 동영상의 3차원 변환을 위하여 먼저 영상을 객체로 분할하고, 분할된 객체의 깊이를 추정하는 방법을 사용하였다. Normalized cut은 영상분할의 한 방법으로, 본 연구에서는 연산속도 향상을 위하여 기존 방법에 watershed 알고리즘을 적용하였고, 정확도 향상을 위하여 가중치에 optical flow를 추가하였다. Normalized cut을 이용하여 분할된 영상의 깊이 정보를 추정하기 위하여 optical flow를 이용하였다. Optical flow의 차이를 통해 정의할 수 있는 가려진 영역의 분할 영상 변화를 통해 순서적 깊이 정보를 추정한다. 추정된 순서적 깊이를 보정하기 위해 optical flow의 절대적 크기를 이용해 운동시차로 상대적 깊이를 추정하였다. 최종적으로 추정된 깊이 정보는 순서적 깊이와 상대적 깊이의 곱을 평균 optical flow로 나누어, 순서적 깊이의 차이를 보정하였다. 제안한 방법의 검증을 위하여 2차원 동영상을 3차원 동영상으로 변환하여 깊이 정보가 추정됨을 확인하였다.

• Structural Engineering and Mechanics
• /
• 제33권5호
• /
• pp.543-560
• /
• 2009

The rotating Rayleigh-Timoshenko beam element based on B-spline wavelet on the interval (BSWI) is constructed to discrete short shaft and stiffness disc. The crack is represented by non-dimensional linear spring using linear fracture mechanics theory. The wavelet-based finite element model of rotor system is constructed to solve the first three natural frequencies functions of normalized crack location and depth. The normalized crack location, normalized crack depth and the first three natural frequencies are then employed as the training samples to achieve the neural networks for crack diagnosis. Measured natural frequencies are served as inputs of the trained neural networks and the normalized crack location and depth can be identified. The experimental results of fatigue crack in short shaft is also given.

• 대한원격탐사학회지
• /
• 제34권4호
• /
• pp.625-638
• /
• 2018

In this paper, we derive i) a function to estimate snow cover fraction (SCF) from a MODIS satellite image that has a wide observational area and short re-visit period and ii) a function to determine snow depth from the estimated SCF map. The SCF equation is important for estimating the snow depth from optical images. The proposed SCF equation is defined using the Gaussian function. We found that the Gaussian function was a better model than the linear equation for explaining the relationship between the normalized difference snow index (NDSI) and the normalized difference vegetation index (NDVI), and SCF. An accuracy test was performed using 38 MODIS images, and the achieved root mean square error (RMSE) was improved by approximately 7.7 % compared to that of the linear equation. After the SCF maps were created using the SCF equation from the MODIS images, a relation function between in-situ snow depth and MODIS-derived SCF was defined. The RMSE of the MODIS-derived snow depth was approximately 3.55 cm when compared to the in-situ data. This is a somewhat large error range in the Republic of Korea, which generally has less than 10 cm of snowfall. Therefore, in this study, we corrected the calculated snow depth using the relationship between the measured and calculated values for each single image unit. The corrected snow depth was finally recorded and had an RMSE of approximately 2.98 cm, which was an improvement. In future, the accuracy of the algorithm can be improved by considering more varied variables at the same time.

• 한국통신학회논문지
• /
• 제35권12C호
• /
• pp.984-994
• /
• 2010

Pose variation is a critical problem in face recognition. Three-dimensional(3D) face recognition techniques have been proposed, as 3D data contains depth information that may allow problems of pose variation to be handled more effectively than with 2D face recognition methods. This paper proposes a pose-normalized 3D face modeling method that translates and rotates any pose angle to a frontal pose using a plane fitting method by Singular Value Decomposition(SVD). First, we reconstruct 3D face data with stereo vision method. Second, nose peak point is estimated by depth information and then the angle of pose is estimated by a facial plane fitting algorithm using four facial features. Next, using the estimated pose angle, the 3D face is translated and rotated to a frontal pose. To demonstrate the effectiveness of the proposed method, we designed 2D and 3D face recognition experiments. The experimental results show that the performance of the normalized 3D face recognition method is superior to that of an un-normalized 3D face recognition method for overcoming the problems of pose variation.

• 대한기계학회논문집B
• /
• 제23권3호
• /
• pp.398-407
• /
• 1999

In the present image processing technique, the concept of the gradient indicator(GI) has been introduced to find out the depth-of-field in sizing large particles ranging from $30{\mu}m$ to $30{\mu}m$ where using of the concept of the normalized contrast value(VC) is not appropriate. The gradient indicator is defined as the ratio of the local value to the maximum possible value of the gray-level gradient in an image frame. The gradient indicator decreases with the increases of the particle size and the distance from the exact focal plane. A particle is considered to be in focus when the value of the gradient indicator at its image boundary stays above a critical value. This critical gradient indicator($GI_{critical}$) is defined as the maximum gradient indicator($GI_{max}$) subtracted by a constant ${\Delta}GI$ which is to account for the particle-size effect. In the present ca.so, the value of ${\Delta}GI$ was set to 0.28 to keep the standard deviation of the measured particles mostly within 0.1. It was also confirmed that, to find the depth-of-field for small particles(${\leq}30{\mu}m$) with the same measurement accuracy, tho concept of the critical normalized contrast($VC_{critical}$) is applicable with 85% of the maximum normalized contrast value($VC_{max}$). Finally, the depth-of-field was checked for the size range between $10{\mu}m$ and $300{\mu}m$ when the both in-focus criteria ($GI_{critical}$ and $VC_{critical}$) were adopted. The change of the depth-of-field with the particle size shows good linearity in both the VC-applicable and the GI-applicable ranges with a reasonable accuracy.

• 대한기계학회논문집B
• /
• 제24권4호
• /
• pp.599-607
• /
• 2000

In this study, experiments have been performed to examine the effects of background gray-level on the depth-of-field and on the in-focus criteria. The normalized value of contrast(VC) and the gradient indicator(GI) were used as the in-focus criteria for the small and the large size-ranges of particles, respectively. The slightly larger number of pixels were detected with the brighter background. The maximum of the normalized value of contrast(VCmax) is decreased with the brighter background and its deviation from that with the background gray-level of 160 turned out to be about $pm$15% when the background gray-level changes from 100 to 200. However, the maximum gradient indicator(GImax) changes with the background gray-level within only $pm$5%. The depth-of-field for the VC-applicable particle-size range is largely dependent on the background gray-level. On the other hand, the depth-of-field for the GI-applicable particle-size range changes only slightly with the background gray-level. To keep the normalized standard deviation of the particle size within 0.1, the background gray-level should be set 160$pm$20 for both the VC-applicable and GI-applicable ranges which cover the particle size between $10{\mu}m$ and $300{\mu}m$.

• 대한원격탐사학회지
• /
• 제33권4호
• /
• pp.401-410
• /
• 2017

Estimation of snow depth using optical image is conducted by using correlation with Snow Cover Fraction (SCF). Various algorithms have been proposed for the estimation of snow cover fraction based on Normalized Difference Snow Index (NDSI). In this study we tested linear, quadratic, and exponential equations for the generation of snow cover fraction maps using data from the Moderate Resolution Imaging Spectroradiometer (MODIS) Aqua satellite in order to evaluate their applicability to the complex terrain of South Korea and to search for improvements to the estimation of snow depth on this landscape. The results were validated by comparison with in-situ snowfall data from weather stations, with Root Mean Square Error (RMSE) calculated as 3.43, 2.37, and 3.99 cm for the linear, quadratic, and exponential approaches, respectively. Although quadratic results showed the best RMSE, this was due to the limitations of the data used in the study; there are few number of in-situ data recorded on the station at the time of image acquisition and even the data is mostly recorded on low snowfall. So, we conclude that linear-based algorithms are better suited for use in South Korea. However, in the case of using the linear equation, the SCF with a negative value can be calculated, so it should be corrected. Since the coefficients of the equation are not optimized for this area, further regression analysis is needed. In addition, if more variables such as Normalized Difference Vegetation Index (NDVI), land cover, etc. are considered, it could be possible that estimation of national-scale snow depth with higher accuracy.

• 디지털융복합연구
• /
• 제10권4호
• /
• pp.191-197
• /
• 2012

본 논문에서는 단일 프레임 영상에서 초점을 이용하여 초기 깊이정보를 추출한 후 입체 영상을 생성하는 방법을 제안하였다. 단일 프레임 영상에서 깊이를 추정하기 위해 원본 영상과 가우시안 필터를 중첩 적용하여 생성된 영상의 비교를 통해 영상의 초점 값을 추출하고 추출된 값을 기반으로 초기 깊이정보를 생성하도록 하였다. 생성된 초기 깊이정보를 Normalized cut을 이용한 객체 분할 결과에 할당하고 각 객체의 깊이를 객체 내 깊이 정보의 평균값으로 보정하여 동일 객체가 같은 깊이 값을 갖도록 하였다. 객체를 제외한 배경 영역은 객체를 제외한 배경 영역의 에지 정보를 이용하여 깊이를 생성하였다. 생성된 깊이를 DIBR(Depth Image Based Rendering)을 이용하여 입체 영상으로 변환하였고 기존 알고리즘을 통해 생성된 영상과 비교 분석하였다.

• 한국방송∙미디어공학회:학술대회논문집
• /
• 한국방송∙미디어공학회 2022년도 추계학술대회
• /
• pp.99-102
• /
• 2022

실재하는 물체나 장소를 디지털 카메라나 휴대폰 카메라로 여러 장 촬영하여 얻은 2차원 이미지 데이터셋으로부터 3차원 영상을 얻기 위해서 이미지를 이루는 각 pixel의 depth 정보를 얻는 것은 필수적인 과정이다. 주어진 이미지에서 depth 정보를 얻기 위해 Shuhan Shen은 PatchMatch 알고리즘을 활용하는 것을 제안하였다. 그 이후 PatchMatch 기반의 알고리즘은 널리 사용되며 우수한 성능을 보이고 있다. PatchMatch 기반의 알고리즘을 사용해 depth를 추정하는 과정에서 depth와 법선 벡터를 Zero-mean Normalized Cross Correlation(ZNCC)를 사용해 평가한다. 하지만, ZNCC는 depth를 평가하려는 pixel의 주변 pixel들의 밝기 값 혹은 색상 값의 분포를 사용하기 때문에 밝기 값이나 색상 값의 변화가 적은 texture-less region에서는 신뢰성이 떨어진다. 본 논문에서는 이 문제를 epipolar geometry를 활용한 기하학적 정보를 이용하여 개선하고자 한다.

• 전자공학회논문지A
• /
• 제33A권6호
• /
• pp.136-143
• /
• 1996

A simple analytic model for the planar junctions with multiple foating field limiting rings(FLR) is presented which yields analytic expressions for the breakdown voltage and optimum ring spacings. the normalized potential of each ring is derived as a function of the normalized depletion width and the ring spacing. Based on the assumption that the breakdwon occurs simulataneously at cylindrical junctions of FLR structure where the peak sruface electric fields are equal, the optimum ring spacings are determined. The resutls are in good agreement with the simulations obtained from two dimensional device simulation program MEDICI and with the experimental data reported. The normalized experessions allow a calculation of breakdown voltage and optimum spacing over a broad range of junction depth and background doping levels.