• Journal of Korean Society for Quality Management
• /
• v.25 no.4
• /
• pp.99-105
• /
• 1997

We consider a nonparametric estimation of the renewal function. In this paper, we suggest modified methods for Frees's estimator to enhance the efficiency. The methods are based on a piecewise linearization and on the fact that the bounded monotonic functions converging pointwise to the bounded monotonic continuous function converge uniformly. In a simulation study, we show that the modified methods have the better efficiency than that introduced by Frees.

• Proceedings of the KIEE Conference
• /
• 2007.10a
• /
• pp.105-106
• /
• 2007

We present a fast reconstruction technique for elliptic objects, which can be applied to real-time computer tomography (CT) for simple geometric objects. It will be also shown that only 3 projections are needed to reconstruct an ellipse. A piecewise quadratic model is also proposed for more efficient Kalman filter based support estimation, which is used for the fast reconstruction technique. The performance of the piecewise quadratic model is compared with that of the existing piecewise linear model. Simulation results for the fast reconstruction are also presented.

• Journal of the Korean Society of Safety
• /
• v.8 no.2
• /
• pp.39-43
• /
• 1993

The flash points of flammable liquids are a fundamental and important property relative to fire and explosion hazards. A new estimation method, based on statistics (mutiple regression analysis), is being developed for the prediction of flash points of pure flammable liquids by means of computer simulation. This method has been applied to alcohol liquids. The proposed method has proved to be the general method for predicting the flash points of alcohol liquids.

• Annual Conference of KIPS
• /
• 2020.05a
• /
• pp.483-486
• /
• 2020

Age estimation and gender classification for human is a classic problem in computer vision. Almost research focus just only one task and the models are too heavy to run on low-cost system. In our research, we aim to apply multitasking learning to perform both task on a lightweight model which can achieve good precision on embedded system in the real time.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2009.01a
• /
• pp.254-257
• /
• 2009

In this paper, we propose a learning-based super-resolution algorithm. In the proposed algorithm, a multi-resolution wavelet approach is adopted to perform the synthesis of local high-frequency features. To obtain a high-resolution image, wavelet coefficients of two dominant LH- and HL-bands are estimated based on wavelet frames. In order to prepare more efficient training sets, the proposed algorithm utilizes the LH-band and transposed HL-band. The training sets are then used for the estimation of wavelet coefficients for both LH- and HL-bands. Using the estimated high frequency bands, a high resolution image is reconstructed via the wavelet transform. Experimental results demonstrate that the proposed scheme can synthesize high-quality images.

• Journal of Institute of Control, Robotics and Systems
• /
• v.22 no.10
• /
• pp.833-838
• /
• 2016

This paper presents a performance evaluation of various feature-initialization algorithms for forward-viewing mono-camera based simultaneous localization and mapping (SLAM), specifically in indoor environments. For mono-camera based SLAM, the position of feature points cannot be known from a single view; therefore, it should be estimated from a feature initialization method using multiple viewpoint measurements. The accuracy of the feature initialization method directly affects the accuracy of the SLAM system. In this study, four different feature initialization algorithms are evaluated in simulations, including linear triangulation; depth parameterized, linear triangulation; weighted nearest point triangulation; and particle filter based depth estimation algorithms. In the simulation, the virtual feature positions are estimated when the virtual robot, containing a virtual forward-viewing mono-camera, moves forward. The results show that the linear triangulation method provides the best results in terms of feature-position estimation accuracy and computational speed.

• Journal of the Korea Society of Computer and Information
• /
• v.25 no.5
• /
• pp.65-72
• /
• 2020

To improve the performance of D2D power control algorithm proposed in the previous work, three modified D2D power control algorithms are proposed to compensate the channel estimation error. Then, we evaluate the performance of three modified D2D power control algorithms in the channel estimation error environment. In real channel environment, the channel estimation is not perfect. To that end, the impact of imperfect channel estimation on the D2D power control algorithm, which was developed with the assumption of perfect channel estimation, has been analyzed in the previous work. Three modified D2D power control algorithms are based on 1) Retransmission, 2) SIR Margin, and 3) Retransmission and SIR Margin. Simulation results show that the Retransmission and SIR Margin approach shows best performance in the sense of the transmit power consumption and the latency.

• Journal of the Korea Computer Graphics Society
• /
• v.28 no.3
• /
• pp.45-54
• /
• 2022

Image depth estimation is a technology that is the basis of various image analysis. As analysis methods using deep learning models emerge, studies using deep learning in image depth estimation are being actively conducted. Currently, most deep learning-based depth estimation models are being trained with clean and ideal images. However, due to the lack of data on adverse conditions such as haze or fog, the depth estimation may not work well in such an environment. It is hard to sufficiently secure an image in these environments, and in particular, obtaining non-homogeneous haze data is a very difficult problem. In order to solve this problem, in this study, we propose a method of synthesizing non-homogeneous haze images and a learning method for a monocular depth estimation deep learning model using this method. Considering that haze mainly occurs outdoors, datasets mainly containing outdoor images are constructed. Experiment results show that the model with the proposed method is good at estimating depth in both synthesized and real haze data.

• Journal of Institute of Control, Robotics and Systems
• /
• v.22 no.9
• /
• pp.716-722
• /
• 2016

Coordinate estimation is an essential function for autonomous navigation of a mobile robot. The optical mouse sensor is convenient and cost-effective for the coordinate estimation problem. It is possible to overcome the position estimation error caused by the slip and the model mismatch of robot's motion equation using the optical mouse sensor. One of the simple methods for the position estimation using the optical mouse sensor is integration of the velocity data from the sensor with time. However, the unavoidable noise in the sensor data may deteriorate the position estimation in case of the simple integration method. In general, a mobile robot has ready-to-use motion information from the encoder sensors of driving motors. By combining the velocity data from the optical mouse sensor and the motion information of a mobile robot, it is possible to improve the coordinate estimation performance. In this paper, a coordinate estimation algorithm for an autonomous mobile robot is presented based on the well-known Kalman filter that is useful to combine the different types of sensors. Computer simulation results show the performance of the proposed localization algorithm for several types of trajectories in comparison with the simple integration method.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.10 no.7
• /
• pp.3050-3065
• /
• 2016

Neural networks (NNs) are extensively used in applications requiring signal classification and regression analysis. In this paper, a NN based threshold selection algorithm for 60 GHz millimeter wave (MMW) time of arrival (TOA) estimation using an energy detector (ED) is proposed which is based on the skewness, kurtosis, and curl of the received energy block values. The best normalized threshold for a given signal-to-noise ratio (SNR) is determined, and the influence of the integration period and channel on the performance is investigated. Results are presented which show that the proposed NN based algorithm provides superior precision and better robustness than other ED based algorithms over a wide range of SNR values. Further, it is independent of the integration period and channel model.