• Journal of Broadcast Engineering
• /
• v.27 no.1
• /
• pp.104-123
• /
• 2022

Style transfer based on neural network provides very high quality results by reflecting the high level structural characteristics of images, and thereby has recently attracted great attention. This paper deals with the problem of resolution limitation due to GPU memory in performing such neural style transfer. We can expect that the gradient operation for style transfer based on partial image, with the aid of the fixed size of receptive field, can produce the same result as the gradient operation using the entire image. Based on this idea, each component of the style transfer loss function is analyzed in this paper to obtain the necessary conditions for partitioning and padding, and to identify, among the information required for gradient calculation, the one that depends on the entire input. By structuring such information for using it as auxiliary constant input for partition-based gradient calculation, this paper develops a recursive algorithm for super high-resolution image style transfer. Since the proposed method performs style transfer by partitioning input image into the size that a GPU can handle, it can perform style transfer without the limit of the input image resolution accompanied by the GPU memory size. With the aid of such super high-resolution support, the proposed method can provide a unique style characteristics of detailed area which can only be appreciated in super high-resolution style transfer.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.30 no.5A
• /
• pp.436-444
• /
• 2005

Unitary matrix modulation (UMM) is investigated in multiple antennas system that is called unitary space-time modulation (USTM). In an OFDM, the diagonal components of UMM with splitting over the coherence bandwidth (UMM-S/OFDM) have been proposed. Recently LDPC code is strongly attended and studied due to simple decoding property with good error correction property. In this paper, we propose LDPC coded UMM-S/OFDM for increasing the system performance. Our proposed system can obtain frequency diversity using UMM-S/OFDM like USTM/OFDM, and large coding gain using LDPC code. The superior characteristics of the proposed UMM-S/OFDM are demonstrated by extensive computer simulations in multi-path Rayleigh fading channel.

• Transactions of the Korean Society of Mechanical Engineers B
• /
• v.37 no.8
• /
• pp.789-798
• /
• 2013

The procedures and relevant programs developed for analyzing mechanical load data of wind turbine generator systems, which are obtained through type certification tests, are verified. The following issues according to IEC 61400-13 are covered in the developed programs: data validation, time series analysis, summary load statistics, generation of fatigue load spectra, and estimation of equivalent loads. A capture matrix for normal power production is generated to determine whether the collected data sets are sufficient to carry out fatigue analysis. Fatigue load spectra are obtained through the rainflow counting method using 50 load ranges; finally, equivalent loads are calculated using different S-N curve slopes, m, according to the relevant materials. Case studies are performed using aero-elastic simulation data of the NREL 5 MW baseline wind turbine with a monopile foundation.

• Journal of Digital Contents Society
• /
• v.13 no.1
• /
• pp.91-99
• /
• 2012

In this paper we introduce a quiz application program that digitizes the analogue quiz game. We digitize the quiz components such as quiz proceeding, participants recognition, problem presentation, volunteer recognition who raises his hand first, answer judgement, score addition, winner decision, etc, which are manually performed in the normal quiz game. For automation, we obtained the depth images from the kinect camera which comes into the spotlight recently, so that we located the quiz participants and recognized the user-friendly defined gestures. Analyzing the depth distribution, we detected and segmented the upper body parts and located the hands' areas. Also, we extracted hand features and designed the decision function that classified the hand pose into palm, fist or else, so that a participant can select the example that he wants among presented examples. The implemented quiz application program was tested in real time and showed very satisfactory gesture recognition results.

• Journal of the Korea Computer Graphics Society
• /
• v.8 no.3
• /
• pp.9-16
• /
• 2002

Recently, high dynamic range (HDR) compression has attracted much attention due to the wide availability of HDR images. In this paper, we present an HDR compression method using a progressive image, which is a multi-level image representation based on a progressive mesh. An HDR image can be decomposed into a base image and a sequence of details by conversion into a progressive image. This decomposition provides a good structure to highly compress the dynamic range while preserving image details. The base image and larger details are considerably scaled down but smaller details are slightly scaled down. Experimental results show that our method successfully generates HDR compressed images without halo artifacts by controlling two intuitive parameters.

• Transactions of the Korean Society of Mechanical Engineers
• /
• v.15 no.1
• /
• pp.127-138
• /
• 1991

A gear forming method by cold extrusion and an analytical method with its numerical solution program based on the upper bound method were developed. In the analysis the involute curve was as a shape of die and the upper bound method was used to calculate energy dissipation rate. By this method the power requirement and optimum conditions necessary for extruding helical(spur) gear were successfully calculated. These numerical solutions are in good agreement with experimental data. In the experiment, 4-6 class helical gear of KS standard for automobile transmission was successfully manufactured.

• The Journal of Information Technology and Database
• /
• v.2 no.2
• /
• pp.25-34
• /
• 1995

In this paper, we present design philosophy and implementation issues of a functionality testing tool for object-oriented database systems. A testing tool has been developed to validate UniSQL/X functionalities with C++ interface. A testing tool is designed under consideration of scaleability, simplicity and extendibility. The schema is deliberately constructed to verify the object-oriented functionalities such as abstraction, inheritance and aggregation. Each test item has been derived under various black box techniques such as equivalent partitioning and boundary-value analysis. The testing tool consists of six phases, namely, database creation, database population, construction of testindex, compilation and link, execution and result reporting, and final cleanup. The prototype provides more than 140 test items at 90 programs.

• KSCE Journal of Civil and Environmental Engineering Research
• /
• v.11 no.4
• /
• pp.45-54
• /
• 1991

The closed forms of mass matrix with rotational inertia matrix are developed for free vibration analysis in space structures containing linearing tapered members with cross section of thin-walled tube. The exact displacement functions are used for formulating mass matrix. The very small slopes of the tapered member are used in usual practice, such that the series expansion forms of these are also developed to avoid numerical failure in vibration analysis. Significant improvements of accuracy and efficiency of free vibration analysis are achieved by using the mass matrices developed in this study. Frequencies of free vibration of tapered members are compared with solutions based upon stepped representation of beam element.

• Journal of the Korea Computer Graphics Society
• /
• v.7 no.3
• /
• pp.1-7
• /
• 2001

We present an efficient and robust algorithm to parametrize a perspective silhouette curve of a canal surface. We also detect each connected component of the silhouette. A canal surface is an envelope surface of a moving sphere with varying radii, which is defined by the center trajectory C(t) and radius function r(t) of the moving sphere. A canal surface can be decomposed to a set of characteristic circles. We derive the equations for the silhouette points on each characteristic circle, then parameterize the silhouette curve by using the equation. The sphere $S(t_*)$ with center point $C(t_*)$ and radius $r(t_*)$ intersects with the canal surface at a characteristic circle $K(t_*)$. The perspective silhouette of the sphere $S(t_*)$ from given view point consists of a circle $Q(t_*)$. By finding the values of t at which K(t) and Q(t) tangentially intersect, we detect each connected component of the silhouette curve.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.19 no.12
• /
• pp.2924-2932
• /
• 2015

Image binarization is a process to divide the image into objects and backgrounds, widely applied to the fields of image analysis and its recognition. In the existing method of binarization, there is some uncertainty when there is insufficient brightness gap between objects and backgrounds in setting threshold. The method of fuzzy binarization has improved the features of objects efficiently. However, since this method sets ${\alpha}$-cut value statically, there remain some problems that important features of objects can be lost during binarization. Therefore, in this paper, we propose a binarization method which does not set ${\alpha}$-cut value statically. The proposed method uses fuzzy membership functions calculated by thresholds of mean, iterative, and Otsu binarization. Experiment results show the proposed method binaries various images with less loss than the existing methods.