• Journal of the Korean Institute of Telematics and Electronics S
• /
• v.36S no.12
• /
• pp.51-57
• /
• 1999

In this paper, an adaptive thresholding of the nonuniformly contrasted images is proposed through using the contrast pre-enhancement of the local regions and the bilinear interpolation between the local threshold values. The nonuniformly contrasted image is decomposed into 9${\times}$9 sized local regions, and the contrast is enhanced by intensifying the gray level difference of each low contrasted or blurred region. Optimal threshold values are obtained by iterative method from the gray level distribution of each contrast-enhanced local region. Discontinuities are reduced at the region of interest or at the characters by using bilinear interpolation between the neighboring threshold surfaces. Character recognition experiments are conducted using backpropagation neural network on the characters extracted from the nonuniformly contrasted document, PCB, and wafer images binarized through using the proposed thresholding and the conventional thresholding methods, and the results prove the relative effectiveness of the proposed scheme.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.38A no.7
• /
• pp.560-570
• /
• 2013

In this paper, we propose transceiver design strategies for the two-cell multiple-input multiple-output (MIMO) interfering broadcast channel where inter-cell interference (ICI) exists in addition to inter-user interference (IUI). We first formulate the generalized zero-forcing interference alignment (ZF-IA) method based on the alignment of IUI and ICI in multi-dimensional subspace. We then devise a minimum weighted-mean-square-error (WMSE) method based on "regularizing" the precoders and decoders of the generalized ZF-IA scheme. In contrast to the existing weighted-sum-rate-maximizing transceiver, our method does not require an iterative calculation of the optimal weights. Because of this, the proposed scheme, while not designed specially to maximize the sum-rate, is computationally efficient and achieves a faster convergence compared to the known weighed-sum-rate maximizing scheme. Through analysis and simulation, we show the effectiveness of the proposed regularized ZF-IA scheme.

• Journal of Power Electronics
• /
• v.17 no.4
• /
• pp.1014-1026
• /
• 2017

The traditional fault diagnosis method for photovoltaic (PV) inverters has a difficult time meeting the requirements of the current complex systems. Its main weakness lies in the study of nonlinear systems. In addition, its diagnosis time is long and its accuracy is low. To solve these problems, a hidden Markov model (HMM) is used that has unique advantages in terms of its training model and its recognition for diagnosing faults. However, the initial value of the HMM has a great influence on the model, and it is possible to achieve a local minimum in the training process. Therefore, a genetic algorithm is used to optimize the initial value and to achieve global optimization. In this paper, the HMM is combined with a genetic algorithm (GHMM) for PV inverter fault diagnosis. First Matlab is used to implement the genetic algorithm and to determine the optimal HMM initial value. Then a Baum-Welch algorithm is used for iterative training. Finally, a Viterbi algorithm is used for fault identification. Experimental results show that the correct PV inverter fault recognition rate by the HMM is about 10% higher than that of traditional methods. Using the GHMM, the correct recognition rate is further increased by approximately 13%, and the diagnosis time is greatly reduced. Therefore, the GHMM is faster and more accurate in diagnosing PV inverter faults.

• Journal of the Korea Society of Computer and Information
• /
• v.25 no.4
• /
• pp.37-45
• /
• 2020

The development process of deep learning is an iterative task that requires a lot of manual work. Among the steps in the development process, pre-processing of learning data is a very costly task, and is a step that significantly affects the learning results. In the early days of AI's algorithm research, learning data in the form of public DB provided mainly by data scientists were used. The learning data collected in the real environment is mostly the operational data of the sensors and inevitably contains various noises. Accordingly, various data cleaning frameworks and methods for removing noises have been studied. In this paper, we proposed a method for detecting and removing noises from time-series data, such as sensor data, that can occur in the IoT environment. In this method, the linear regression method is used so that the system repeatedly finds noises and provides data that can replace them to clean the learning data. In order to verify the effectiveness of the proposed method, a simulation method was proposed, and a method of determining factors for obtaining optimal cleaning results was proposed.

• Journal of the Korea Society of Computer and Information
• /
• v.18 no.2
• /
• pp.19-30
• /
• 2013

We study sub-Peres functions that are defined recursively as Peres function for random number generation. Instead of using two parameter functions as in Peres function, the sub-Peres functions uses only one parameter function. Naturally, these functions produce less random bits, hence are not asymptotically optimal. However, the sub-Peres functions runs in linear time, i.e., in O(n) time rather than O(n logn) as in Peres's case. Moreover, the implementation is even simpler than Peres function not only because they use only one parameter function but because they are tail recursive, hence run in a simple iterative manner rather than by a recursion, eliminating the usage of stack and thus further reducing the memory requirement of Peres's method. And yet, the output rate of the sub-Peres function is more than twice as much as that of von Neumann's method which is widely known linear-time method. So, these methods can be used, instead of von Neumann's method, in an environment with limited computational resources like mobile devices. We report the analyses of the sub-Peres functions regarding their running time and the exact output rates in comparison with Peres function and other known methods for random number generation. Also, we discuss how these sub-Peres function can be implemented.

• Journal of KIISE:Computing Practices and Letters
• /
• v.8 no.2
• /
• pp.195-205
• /
• 2002

Crew scheduling is the problem of pairing crews with each of the vehicles in operation during a certain period of time. A typical procedure of crew schedule optimization consists of enumerating all possible pairings and then selecting the subset which can cover all the operating vehicles, with the goal of minimizing the number of pairings in the subset. The linear programming approach popularly adopted for optimal selection of pairings, however, is not applicable when the objective function cannot be expressed in a linear form. This paper proposes a method of integrating integer programming and heuristic search to solve difficult crew scheduling problems in which the objective function cannot be expressed in linear form and at the same time the number of crews available is limited. The role of heuristic search is to improve the incomplete solution generated by integer programming through iterative repair. Experimental results show that our method outperforms human experts in terms of both solution quality and execution time when applied to real world crew scheduling Problems which can hardly be solved by traditional methods.

• Journal of Korea Water Resources Association
• /
• v.44 no.9
• /
• pp.711-720
• /
• 2011

SWAT (Soil and Water Assessment Tool) model was calibrated for the flow rate of the Deachung lake with a large area of 3108.29 $km^2$. Application of SWAT model requires significant number of input data and is prone to result in uncertainties due to errors in input data, model structure and model parameters. The SUFI-2 (Sequential Uncertainty Fitting Ver. 2) program and GLUE (Generalized Likelihood Uncertainty Estimation) program in SWAT-CUP (SWAT-Calibration and Uncertainty Program) are used to select the best parameters for SWAT model. Optimal combination of parameter values was determined through 2,000 iterative SWAT model runs. The Nash-Sutcliffe values and $R^2$ values were 0.87 and 0.89 respectively indicating both methods show good agreements with observed data successfully. RMSE and MSE values also showed similar results for both programs. It seems the SWAT-CUP has a great practical appeal for parameter optimization especially for large basin area and it also can be used for less experienced SWAT model users.

• Geophysics and Geophysical Exploration
• /
• v.7 no.3
• /
• pp.207-212
• /
• 2004

This article reviews recent developments in three-dimensional (3-D) magntotelluric (MT) imaging. The inversion of MT data is fundamentally ill-posed, and therefore the resultant solution is non-unique. A regularizing scheme must be involved to reduce the non-uniqueness while retaining certain a priori information in the solution. The standard approach to nonlinear inversion in geophysis has been the Gauss-Newton method, which solves a sequence of linearized inverse problems. When running to convergence, the algorithm minimizes an objective function over the space of models and in the sense produces an optimal solution of the inverse problem. The general usefulness of iterative, linearized inversion algorithms, however is greatly limited in 3-D MT applications by the requirement of computing the Jacobian(partial derivative, sensitivity) matrix of the forward problem. The difficulty may be relaxed using conjugate gradients(CG) methods. A linear CG technique is used to solve each step of Gauss-Newton iterations incompletely, while the method of nonlinear CG is applied directly to the minimization of the objective function. These CG techniques replace computation of jacobian matrix and solution of a large linear system with computations equivalent to only three forward problems per inversion iteration. Consequently, the algorithms are efficient in computational speed and memory requirement, making 3-D inversion feasible.

• Korean Journal of Radiology
• /
• v.22 no.4
• /
• pp.624-633
• /
• 2021

Objective: To evaluate the reliability of CT measurements of muscle quantity and quality using variable CT parameters. Materials and Methods: A phantom, simulating the L2-4 vertebral levels, was used for this study. CT images were repeatedly acquired with modulation of tube voltage, tube current, slice thickness, and the image reconstruction algorithm. Reference standard muscle compartments were obtained from the reference maps of the phantom. Cross-sectional area based on the Hounsfield unit (HU) thresholds of muscle and its components, and the mean density of the reference standard muscle compartment, were used to measure the muscle quantity and quality using different CT protocols. Signal-to-noise ratios (SNRs) were calculated in the images acquired with different settings. Results: The skeletal muscle area (threshold, -29 to 150 HU) was constant, regardless of the protocol, occupying at least 91.7% of the reference standard muscle compartment. Conversely, normal attenuation muscle area (30-150 HU) was not constant in the different protocols, varying between 59.7% and 81.7% of the reference standard muscle compartment. The mean density was lower than the target density stated by the manufacturer (45 HU) in all cases (range, 39.0-44.9 HU). The SNR decreased with low tube voltage, low tube current, and in sections with thin slices, whereas it increased when the iterative reconstruction algorithm was used. Conclusion: Measurement of muscle quantity using HU threshold was reliable, regardless of the CT protocol used. Conversely, the measurement of muscle quality using the mean density and narrow HU thresholds were inconsistent and inaccurate across different CT protocols. Therefore, further studies are warranted in future to determine the optimal CT protocols for reliable measurements of muscle quality.

• Nuclear Medicine and Molecular Imaging
• /
• v.41 no.3
• /
• pp.234-240
• /
• 2007

Purpose: In this study we propose a block-iterative method for reconstructing Compton scattered data. This study shows that the well-known expectation maximization (EM) approach along with its accelerated version based on the ordered subsets principle can be applied to the problem of image reconstruction for Compton camera. This study also compares several methods of constructing subsets for optimal performance of our algorithms. Materials and Methods: Three reconstruction algorithms were implemented; simple backprojection (SBP), EM, and ordered subset EM (OSEM). For OSEM, the projection data were grouped into subsets in a predefined order. Three different schemes for choosing nonoverlapping subsets were considered; scatter angle-based subsets, detector position-based subsets, and both scatter angle- and detector position-based subsets. EM and OSEM with 16 subsets were performed with 64 and 4 iterations, respectively. The performance of each algorithm was evaluated in terms of computation time and normalized mean-squared error. Results: Both EM and OSEM clearly outperformed SBP in all aspects of accuracy. The OSEM with 16 subsets and 4 iterations, which is equivalent to the standard EM with 64 iterations, was approximately 14 times faster in computation time than the standard EM. In OSEM, all of the three schemes for choosing subsets yielded similar results in computation time as well as normalized mean-squared error. Conclusion: Our results show that the OSEM algorithm, which have proven useful in emission tomography, can also be applied to the problem of image reconstruction for Compton camera. With properly chosen subset construction methods and moderate numbers of subsets, our OSEM algorithm significantly improves the computational efficiency while keeping the original quality of the standard EM reconstruction. The OSEM algorithm with scatter angle- and detector position-based subsets is most available.