• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.12 no.5
• /
• pp.2253-2272
• /
• 2018

We propose a deep learning method for multi-focus image fusion. Unlike most existing pixel-level fusion methods, either in spatial domain or in transform domain, our method directly learns an end-to-end fully convolutional two-stream network. The framework maps a pair of different focus images to a clean version, with a chain of convolutional layers, fusion layer and deconvolutional layers. Our deep fusion model has advantages of efficiency and robustness, yet demonstrates state-of-art fusion quality. We explore different parameter settings to achieve trade-offs between performance and speed. Moreover, the experiment results on our training dataset show that our network can achieve good performance with subjective visual perception and objective assessment metrics.

• 한국연소학회:학술대회논문집
• /
• 2014.11a
• /
• pp.1-4
• /
• 2014

The present work focuses on the analysis of the pulverized coal combustion aerodynamics of the dual swirl burner by the control of the swirl-modes such as the outer swirl intensity (OSI). The detailed structure of pulverized coal swirling flames with swirl-mode was studied experimentally by particle image velocimetry and local flame colors based on $OH^*$, $CH^*$, and ${C_2}^*$ radicals. For all co-swirling conditions, the internal recirculation zone (IRZ) was observed near the inner shear layer with respect to the processing vortex core structure. Furthermore, a co-rotating vortex in the outer shear layer and the exhaust tube vortex (ETV) along the central axis were observed. The intensity of $CH^*$ signal was higher with small coal particle size, conversely, the size of the distribution of the $CH^*$ signal becomes larger. Therefore, the control of the aerodynamics with changing swirl intensities may play an important role in improving both environmental and combustion performances.

• Journal of the Korea Fashion and Costume Design Association
• /
• v.11 no.2
• /
• pp.33-42
• /
• 2009

The purpose of this study was to research the development of jacquard fabrics by the Bi-Dan patterns of Dan-Chung as one of the traditional korean symbols. For performing this design study, it designed a surface design of the geometric images in Bi-Dan patterns using by EAT Jacquard Textile CAD system and simulated to fabrics for the clothing & bags. To consider the modern image of Bi-Dan patterns in the Dan-Chung motives, the theoretical background was made on the basic elements, compositions and symbolic meanings of Dan-Chung patterns. To merchandise the fashion items and see the visual image, these jacquard fabric simulations was processed by two-dimension modeling through YoungWoo CNI CAD. The applied weave structure was based on a single layer of 5 or 10 points-satin and a double layer of 8 points-satin. Among the fabric simulations of Bi-Dan patterns, the six simulated fabrics were woven into the real jacquard fabrics under the electronic Stabuli jacquard loom in $120{\sim}130$ picks per inch and 171 ends per inch. These developed jacquard fabrics of Bi-Dan patterns were appeared a contemporary yet ethnic feeling so that they could make use of the unique korean cultural products by the further commercialization.

• Journal of Korea Multimedia Society
• /
• v.19 no.5
• /
• pp.837-847
• /
• 2016

Tone mapping for High Dynamic Range(HDR) image provides matching human visual perception between real world scene and displayable devices. Recently, a tone mapping algorithm based on localized gamma correction is proposed. This algorithm is using human visual properties of contrast and colorfulness with background intensity, generating a weight map for gamma correction. However, this method have limitations of controlling enhancement region as well as generating halo artifacts caused by the weight map construction. To overcome aforementioned limitations, proposed algorithm in this paper modifies previous weight map, considering base layer intensity of input luminance channel. By determining enhancement region locally and globally based on base layer intensity, gamma values are corrected accordingly. Therefore, proposed algorithm selectively enhances local brightness and controls strength of edges. Subjective evaluation using z-score shows that our proposed algorithm outperforms the conventional methods.

• Journal of computational fluids engineering
• /
• v.16 no.1
• /
• pp.30-35
• /
• 2011

In this paper, our effort to apply 3-D Virtual Reality system for stereoscopic visualization of mesh data on the web is briefly described. This study is an extension of our previous and on-going research efforts to develop an automatic grid generation program specialized for wing mesh, named as eGWing. The program is developed by using JAVA programming language, and it can be used either as an application program on a local computer or as an applet in the network environment. In this research advancing layer method(ALM) augmented by elliptic smoothing method is used for the structured grid generation. And to achieve a stereoscopic viewing capability, two graphic windows are used to render its own viewing image for the left and right eye respectively. These two windows are merged into one image using 3D monitor and the viewers can see the mesh data visualization results with stereoscopic depth effects by using polarizing glasses. In this paper three dimensional mesh data visualization with stereoscopic technique combined with 3D monitor is demonstrated, and the current achievement would be a good start-up for further development of low-cost high-quality stereoscopic mesh data visualization system which can be shared by many users through the web.

• Journal of Korea Multimedia Society
• /
• v.17 no.7
• /
• pp.778-786
• /
• 2014

Buyer-seller watermarking protocol is one of the copyright protection techniques which combine a cryptographic protocol used in electronic commerce with a digital wetermarking scheme aiming at proving the ownership of multimedia contents and preventing the illegal reproduction and redistribution of digital contents. In this paper, it is proposed a new watermarking scheme in an encrypted domain in an image that is based on the block-DCT framework. In order to implement watermarking scheme in a public-key cryptosystem, it is divided that frequency coefficients exist as real number into integer and decimal layer. And the decimal layer is modified integer type through integral-processing. Also, for robustness and invisibility required in watermarking scheme, it is designed a direction-adaptive watermarking scheme based on locally edge-properties of each block in an image through analyzing distribution property of the frequency coefficients in a block using JND threshold.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.31B no.10
• /
• pp.140-148
• /
• 1994

In this paepr, the modified neocognitron, which has capability of recognizing multi-patterns in input image in one pass, is proposed. It is the hierarchical neural network composed of S and V layer which are able to extract features and of C layer with little effect from deformation, changes in size, shifts in position. S and V cells extract the features of all patterns in input image by applying DCC(don't care condition) to those cells. S and C cells also have position informations of extracted patterns. Position information is used in extracting good features without extracted features beting interfered one another. The proposed method is shorter in recognition time than the selective attention method with backward connection, because of recognizing multi-patterns in one passe. The modified neocognitron can recognizze attached multi-patterns because of using DCC and position informations.

• Journal of information and communication convergence engineering
• /
• v.8 no.6
• /
• pp.690-696
• /
• 2010

Web-based virtual collaboration is increasingly gaining popularity in almost every area in our society due to the fact that it can bridge the gap imposed by time and geographical constraints. However, in medical field, such collaboration has been less popular than other fields. Some of the reasons were timeliness, security, and preciseness of the information they are dealing with. In this paper, we are proposing a web-based distributed medical collaboration system called Virtual Collaboration System for Medicine (VCSM) for medical doctors that meet the needs. The proposed system consists of two parts - multimedia presentation and recordable virtual collaboration. The former supports synchronized multimedia presentation using Synchronous Multimedia Integration Language (SMIL.) It allows synchronization of the contents of a PowerPoint presentation file and a video file. The presentation may be provided to the participants before the discussion begins. Next, in the virtual collaboration stage, participants can use text along with associated symbols during the discussion over the presented medical images. The symbols such as arrows or polygons can be set or removed dynamically to represent areas of interest in digital images using so called layered architecture that separates image layer from annotation layer. XML files are used to record participants' opinions along with the symbols over some particular images

• Journal of the Korean Society of Fisheries and Ocean Technology
• /
• v.53 no.4
• /
• pp.404-412
• /
• 2017

A centrifugal cyclone dust collecting apparatus includes a hydro cyclone dust collecting apparatus for separating solid or liquid using liquid or suspension as a medium. In this study, the formation mechanism and improvement of air core and inner air layer were confirmed through Particle Image Velocimetry. These results showed that the modified experimental model was designed in the conventional method suitable for the separation of juvenile fish and eggs. The inlet speed of the multi-stage hydrocyclone dust collector, which can increase the inlet velocity and minimize floatage in the turbulence chamber, was increased from 0.15 to 0.30 m/s. As a result, the air core was stably formed, the inner air layer was increased with increasing speed. In addition, the dust collecting efficiency of egg and juvenile fish was 97.8% on average, It can infer that this system confirmed the ability to efficiently collect particles of $40{\mu}m$ or more.

• Journal of IKEEE
• /
• v.20 no.3
• /
• pp.303-306
• /
• 2016

In this paper, we propose a method to accelerate convolutional neural network by utilizing a GPGPU. Convolutional neural network is a sort of the neural network learning features of images. Convolutional neural network is suitable for the image processing required to learn a lot of data such as images. The convolutional layer of the conventional CNN required a large number of multiplications and it is difficult to operate in the real-time on the embedded environment. In this paper, we reduce the number of multiplications through Winograd convolution operation and perform parallel processing of the convolution by utilizing SIMT-based GPGPU. The experiment was conducted using ModelSim and TestDrive, and the experimental results showed that the processing time was improved by about 17%, compared to the conventional convolution.