• Journal of the Korean Statistical Society
• /
• 제28권2호
• /
• pp.265-277
• /
• 1999

Classical and inverse estimation methods are two well known methods in statistical calibration problems. When there are outliers, both methods have large MSE's and could not estimate the input value correctly. We suggest median calibration estimation based on the LD-statistics. To investigate the robust performances, the influence function of the median calibration estimator is calculated and compared with other methods. When there are outliers in the response variables, the influence function is found to be bounded. In simulation studies, the MSE's for each calibration methods are compared. The estimated inputs as well as the performance of the influence functions are calculated.

• Journal of Mechanical Science and Technology
• /
• 제14권1호
• /
• pp.113-120
• /
• 2000

This study has developed a new liquid crystal calibration technique using Neural networks with median filtering and applied this technique to heat transfer measurements. To verify the validity of this new measurement technique, the local Nusselt numbers on a flat plate surface subjected to an axisymmetric impinging jet were measured and compared with the results by the conventional Hue-temperature calibration technique under the same conditions. Because the Neural networks predict the non-linear relations between temperatures and corresponding R, G, B values, Neural networks-median filtering calibration technique can utilize a much wider color band in the experiment than the Hue-temperature calibration technique, resulting in a significant reduction in the experimental time.

• 제어로봇시스템학회논문지
• /
• 제13권3호
• /
• pp.224-229
• /
• 2007

This paper presents a camera calibration method using several images for three dimensional measurement applications such as stereo systems, mobile robots, and visual inspection systems in factories. Conventional calibration methods that use single image suffer from errors related to reference point extraction in image, lens distortion, and numerical analysis of nonlinear optimization. The camera parameter values obtained from images of same camera is not same even though we use same calibration method. The camera parameters that are obtained from several images of different view for a calibration target is usaully not same with large error values and we can not assume a special probabilistic distribution when we estimate the parameter values. In this paper, the median value of camera parameters from several images is used to improve estimation of the camera values in an iterative step with nonlinear optimization. The proposed method is proved by experiments using real images.

• 한국정보통신학회논문지
• /
• 제21권9호
• /
• pp.1711-1717
• /
• 2017

본 논문에서는 적절한 패턴의 입력 영상과 고속 중간 값 필터를 이용하여 집적 영상에서 빠르고 정확한 렌즈 배열의 격자 검출 방법을 제안한다. 렌즈 배열의 격자를 검출하기 위해서 수직, 수평 방향의 에지 영상이 필요하다. 이를 위해, 이전의 잘 알려진 격자 검출 방법은 1차원 중간 값 필터를 사용한다. 하지만 이 방법은 속도가 느리고 중간 값 필터 크기를 결정하는데 어려움이 발생한다. 이를 극복하기 위해, 본 논문에서는 binary counting방법으로 중간 값을 구함으로써 속도를 개선하고자 한다. 또, 에지를 잘 검출 할 수 있는 캘리브레이션 패턴 영상을 제안하여 정확도를 향상 시킨다. 실험을 통해 제안하는 방법이 기존의 방법보다 집적 영상에서 렌즈 배열의 격자를 검출할 때 우수하게 적용될 수 있음을 보여주었다.

• Journal of the Korean Data and Information Science Society
• /
• 제19권3호
• /
• pp.995-1006
• /
• 2008

This article deals with the one-sided hypothesis testing problem in inverse Gaussian distribution. We propose Bayesian hypothesis testing procedures for the one-sided hypotheses of the shape parameter under the noninformative prior. The noninformative prior is usually improper which yields a calibration problem that makes the Bayes factor to be defined up to a multiplicative constant. So we propose the objective Bayesian hypothesis testing procedures based on the fractional Bayes factor, the median intrinsic Bayes factor and the encompassing intrinsic Bayes factor under the reference prior. Simulation study and a real data example are provided.

• 농업과학연구
• /
• 제39권3호
• /
• pp.413-420
• /
• 2012

Spectroscopic measurement method based on visible and near-infrared wavelengths was prominent technology for rapid and non-destructive evaluation of internal quality of fruits. Reflectance measurement was performed to evaluate firmness, soluble solid content, and acid content of truss tomatoes by hyperspectral reflectance imaging system. The Vis/NIR reflectance spectra was acquired from truss tomatoes sorted by 6 ripening stages. The multivariable analysis based on partial least square (PLS) was used to develop regression models with several preporcessing methods, such as smoothing, normalization, multiplicative scatter correction (MSC), and standard normal variate (SNV). The best model was selected in terms of coefficient of determination of calibration ($R_c^2$) and full cross validation ($R_{cv}^2$), and root mean standard error of calibration (RMSEC) and full cross validation (RMSECV). The results of selected models were 0.8976 ($R_p^2$), 6.0207 kgf (RMSEP) with gaussian filter of smoothing, 0.8379 ($R_p^2$), $0.2674^{\circ}Bx$ (RMSEP) with the mean of normalization, and 0.7779 ($R_p^2$), 0.1033% (RMSEP) with median filter of smoothing for firmness, soluble solid content (SSC), and acid content, respectively. Results show that Vis / NIR hyperspectral reflectance imaging technique has good potential for the measurement of internal quality of truss tomato.

• 한국산학기술학회논문지
• /
• 제18권3호
• /
• pp.249-257
• /
• 2017

신뢰성 있는 통행시간 예측을 위해 DSRC로부터 수집된 통행시간에서의 이상치(outlier) 필터링은 필수이다. 통행시간 예측을 위해 사용되는 보편적 기법인 TRANSGUIDE는 특정 분석 시간동안 통행시간의 변동이 크게 발생하는 조건에서 수집데이터의 이상치 제거를 효율적으로 처리하지 못하는 문제점이 존재한다. 이에 본 연구에서는 TRANSGUIDE의 한계점 을 보완할 수 있는 알고리즘을 제안하고자 한다. TRANSGUIDE가 특정 분석 시간대 충분한 데이터 관측이 어려울 경우 Median Absolute Deviation(MAD)를 이용하여 이상치 제거를 위한 새로운 유효 분석 영역을 설정하였다. 새로운 분석 영역 설정 후 특정 시간대 교통 조건하에서 최대 허용 가능한 이상치를 고려한 변수 ${\alpha}$, ${\beta}$를 제안하였다. 변수 ${\alpha}$, ${\beta}$를 추정하기 위해 과거 데이터와 도로 구간의 특성을 반영하였다. 개발된 알고리즘은 수도권 일반국도 3호선, 2013년 1월 1달간 DSRC 데이터가 존재하는 다차로 일반국도에 적용하였다. 누적상대도수를 이용하여 모형의 정산 수행 후 성능에 대해 정량적 평가를 수행하였다. 개발된 알고리즘은 기존의 TRANSGUIDE가 특정 조건, 즉 일정 분석 시간동안 교통 조건이 급하게 변동되는 구간에서 이상치 제거에 취한한 점을 보완하는 것으로 판단되었다. TRANSGUDIDE가 특정 조건에서 통행시간 예측이 어려울 경우 본 개발 알고리즘은 활용될 것으로 판단한다.

• 한국컴퓨터그래픽스학회논문지
• /
• 제23권3호
• /
• pp.31-38
• /
• 2017

본 논문에서는 도심 영상에 대해 맨하탄 좌표계를 추정하는 합성곱 신경망(Convolutional Neural Network) 기반의 시스템을 제안한다. 도심 영상에서 맨하탄 좌표계를 추정하는 것은 영상 조정, 3차원 장면 복원 등 컴퓨터 그래픽스 및 비전 문제 해결의 기본이 된다. 제안하는 합성곱 신경망은 GoogLeNet[1]을 기반으로 구성한다. 합성곱 신경망을 훈련하기 위해 구글 스트리트 뷰 API로 영상을 수집하고 기존 캘리브레이션 방법으로 맨하탄 좌표계를 계산하여 데이터셋을 생성한다. 장면마다 새롭게 합성곱 신경망을 학습해야하는 PoseNet[2]과 달리, 본 논문에서 제안하는 시스템은 장면의 구조를 학습하여 맨하탄 좌표계를 추정하기 때문에 학습되지 않은 새로운 장면에 대해서도 맨하탄 좌표계를 추정한다. 제안하는 방법은 학습에 참여하지 않은 구글 스트리트 뷰 영상을 검증 데이터로 테스트하였을 때 $3.157^{\circ}$의 중간 오차로 맨하탄 좌표계를 추정하였다. 또한, 동일 검증 데이터에 대해 제안하는 방법이 기존 맨하탄 좌표계 추정 알고리즘[3]보다 더 낮은 중간 오차를 보이는 것을 확인하였다.

• Radiation Oncology Journal
• /
• 제36권2호
• /
• pp.139-146
• /
• 2018

Purpose: IBTR! 2.0 nomogram is web-based nomogram that predicts ipsilateral breast tumor recurrence (IBTR). We aimed to validate the IBTR! 2.0 using an external data set. Materials and Methods: The cohort consisted of 2,206 patients, who received breast conserving surgery and radiation therapy from 1992 to 2012 at our institution, where wide surgical excision is been routinely performed. Discrimination and calibration were used for assessing model performance. Patients with predicted 10-year IBTR risk based on an IBTR! 2.0 nomogram score of <3%, 3%-5%, 5%-10%, and >10% were assigned to groups 1, 2, 3, and 4, respectively. We also plotted calibration values to observe the actual IBTR rate against the nomogram-derived 10-year IBTR probabilities. Results: The median follow-up period was 73 months (range, 6 to 277 months). The area under the receiver operating characteristic curve was 0.607, showing poor accordance between the estimated and observed recurrence rate. Calibration plot confirmed that the IBTR! 2.0 nomogram predicted the 10-year IBTR risk higher than the observed IBTR rates in all groups. High discrepancies between nomogram IBTR predictions and observed IBTR rates were observed in overall risk groups. Compared with the original development dataset, our patients had fewer high grade tumors, less margin positivity, and less lymphovascular invasion, and more use of modern systemic therapies. Conclusions: IBTR! 2.0 nomogram seems to have the moderate discriminative ability with a tendency to over-estimating risk rate. Continued efforts are needed to ensure external applicability of published nomograms by validating the program using an external patient population.

• 한국물환경학회지
• /
• 제35권6호
• /
• pp.520-538
• /
• 2019

In this study, it was proposed that a method of setting the target water quality for TOC using the watershed model and the load duration curves to manage non-biodegradable organics in the total water load management system. To simulate runoff and water quality of the watershed, the HSPF model is used which is appropriate for urban and rural areas. Additionally, the load duration curve is used to reflect the variable water quality correlated with various river flow rates in preparing the TMDL plans in the U.S. First, the model was constructed by inputting the loads calculated from the pollutant sources in 2015. After the calibration and verification process, the water quality by flow conditions was analyzed from the BOD and TOC simulation results. When the BOD achieved the target water quality by inputting the target year loads for 2020, the median and average values of TOC were proposed for the target water quality. The provisional method of TOC target water quality for the management of non-biodegradable organics, which is one of the challenges of the total water load management system, was considered. In the future, it is expected to be used as basic data for the conversion of BOD into TOC in the total water load management system.