• Journal of Sensor Science and Technology
• /
• v.28 no.3
• /
• pp.182-186
• /
• 2019

When a camera is employed for 3D sensing, accurate camera calibration is vital as it is a prerequisite for the subsequent steps of the sensing process. Camera calibration is usually performed by complex mathematical modeling and geometric analysis. On the other contrary, data learning using an artificial neural network can establish a transformation relation between the 3D space and the 2D camera image without explicit camera modeling. However, a neural network requires a large amount of accurate data for its learning. A significantly large amount of time and work using a precise system setup is needed to collect extensive data accurately in practice. In this study, we propose a two-step neural calibration method that is effective when only a small amount of learning data is available. In the first step, the camera projection transformation matrix is determined using the limited available data. In the second step, the transformation matrix is used for generating a large amount of synthetic data, and the neural network is trained using the generated data. Results of simulation study have shown that the proposed method as valid and effective.

• 제어로봇시스템학회:학술대회논문집
• /
• 2003.10a
• /
• pp.1933-1938
• /
• 2003

A covariance matrix is a tool that expresses odometry uncertainty of the wheeled mobile robot. The covariance matrix is a key factor in various localization algorithms such as Kalman filter, topological matching and so on. However it is not easy to acquire an accurate covariance matrix because we do not know the real states of the robot. Up to the authors knowledge, there seems to be no established result on the covariance matrix estimation for the odometry. In this paper, we propose a new method which can estimate the covariance matrix from empirical data. It is based on the PC-method and shows a good estimation ability. The experimental results validate the performance of the proposed method.

• MALSORI
• /
• no.43
• /
• pp.127-136
• /
• 2002

The speaker adaptation technique is generally used to reduce the speaker difference in speech recognition. In this work, we focus on the features fitted to a linear regression-based speaker adaptation. These are obtained by feature transformation based on independent component analysis (ICA), and the transformation matrix is learned from a speaker independent training data. When the amount of data is small, however, it is necessary to adjust the ICA-based transformation matrix estimated from a new speaker utterance. To cope with this problem, we propose a smoothing method: through a linear interpolation between the speaker-independent (SI) feature transformation matrix and the speaker-dependent (SD) feature transformation matrix. We observed that the proposed technique is effective to adaptation performance.

• Proceedings of the Computational Structural Engineering Institute Conference
• /
• 2004.04a
• /
• pp.49-56
• /
• 2004

Model updating is a very active research field, in which significant efforts has been invested in recent years. Model updating methodologies are invariably successful when used on noise-free simulated data, but tend to be unpredictable when presented with real experimental data that are-unavoidably-corrupted with uncorrected noise content. In this paper, Reanalysis using frequency response functions for correlating and updating dynamic systems is presented. A transformation matrix is obtained from the relationship between the complex and the normal frequency response functions of a structure. The transformation matrix is employed to calculate the modified damping matrix of the system. The modified mass and stiffness matrices are identified from the normal frequency response functions by using the least squares method. One simulated system is employed to illustrate the applicability of the proposed method. The result indicate that the damping matrix of correlated finite element model can be identified accurately by the proposed method. In addition, the robustness of the new approach uniformly distributed measurement noise Is also addressed.

• Journal of the Korean Data and Information Science Society
• /
• v.27 no.2
• /
• pp.539-548
• /
• 2016

This paper is a study on the multivariate CUSUM control charts using three different control statistics for monitoring covariance matrix. We get control limits and ARLs of the proposed multivariate CUSUM control charts using three different control statistics by using computer simulations. The performances of these proposed multivariate CUSUM control charts have been investigated by comparing ARLs. The purpose of control charts is to detect assignable causes of variation so that these causes can be found and eliminated from process, variability will be reduced and the process will be improved. We show that the charts based on three different control statistics are very effective in detecting shifts, especially shifts in covariances when the variables are highly correlated. When variables are highly correlated, our overall recommendation is to use the multivariate CUSUM control charts using trace for detecting changes in covariance matrix.

• Asian Journal of Innovation and Policy
• /
• v.5 no.3
• /
• pp.251-274
• /
• 2016

Practical methodology for categorizing collaborative disciplines or research in a quantitative manner is presented by developing a Correlation Matrix of Major Disciplines (CMMD) using bibliometric data collected between 2009 and 2014. First, 21 major disciplines in science and engineering are defined based on journal publication frequency. Second, major disciplines using a comparing discipline correlation matrix is created and correlation score using CMMD is calculated based on an analyzer function that is given to the matrix elements. Third, a correlation between the major disciplines and 14 research fields using CMMD is calculated for validation. Collaborative researches are classified into three groups by partially accepting the definition of pluri-discipline from peer review manual, European Science Foundation, inner-discipline, inter-discipline and cross-discipline. Applying simple categorization criteria identifies three groups of collaborative research and also those results can be visualized. Overall, the proposed methodology supports the categorization for each research field.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2009.01a
• /
• pp.737-741
• /
• 2009

This paper addresses the factorization method to estimate the projective structure of a scene from feature (points) correspondences over images with occlusions. We propose both a column and a row space approaches to estimate the depth parameter using the subspace constraints. The projective depth parameters are estimated by maximizing projection onto the subspace based either on the Joint Projection matrix (JPM) or on the the Joint Structure matrix (JSM). We perform the maximization over significant observation and employ Tardif's Camera Basis Constraints (CBC) method for the matrix factorization, thus the missing data problem can be overcome. The depth estimation and the matrix factorization alternate until convergence is reached. Result of Experiments on both real and synthetic image sequences has confirmed the effectiveness of our proposed method.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.23 no.7
• /
• pp.1689-1698
• /
• 1998

In this paper, we present a new method for estimating the parameters of transient-type signal in additive white Gaussian noise. This method makes use of the truncated singular value decomposition of an extended-order auto-correlation-like matrix based on the linear-prediction model. The method is tested on data consisting of two exponentially dampled sinusoidal signals with the same damping factor and different damping factor. Simulation results are illustrated to demonstrate the better performance of the method applied to the auto-correlation-like matrix than that applied to the data matrix.

• Journal of the Korean Data and Information Science Society
• /
• v.26 no.6
• /
• pp.1593-1600
• /
• 2015

A control chart is very useful in monitoring various production process. There are many situations in which the simultaneous control of two or more related quality variables is necessary. We construct bivariate Shewhart control charts based on the trace of the product of the estimated variance-covariance matrix and the inverse of the in-control matrix and investigate the properties of bivariate Shewart control charts with VSI procedure for monitoring covariance matrix in term of ATS (Average time to signal) and ANSW (Average number of switch) and probability of switch, ASI (Average sampling interval). Numerical results show that ATS is smaller than ARL. From examining the properties of switching in changing covariances and variances in ${\Sigma}$, ANSW values show that it does not switch frequently and does not matter to use VSI procedure.

• Proceedings of the Korea Concrete Institute Conference
• /
• 2004.11a
• /
• pp.421-424
• /
• 2004

Model updating is a very active research field, in which significant efforts has been invested in recent years. Model updating methodologies are invariably successful when used on noise-free simulated data, but tend to be unpredictable when presented with real experimental data that are-unavoidably-corrupted with uncorrelated noise content. In this paper, Reanalysis using frequency response functions for correlating and updating dynamic systems is presented. A transformation matrix is obtained from the relationship between the complex and the normal frequency response functions of a structure. The transformation matrix is employed to calculate the modified damping matrix of the system. The modified mass and stiffness matrices are identified from the normal frequency response functions by using the least squares method. Full scale pseudo dynamic pier test is employed to illustrate the applicability of the proposed method. The result indicate that the damping matrix of correlated finite element model can be identified accurately by the proposed method. In addition, the robustness of the new approach uniformly distributed measurement noise is also addressed.