• Journal of Internet Computing and Services
• /
• v.21 no.4
• /
• pp.9-16
• /
• 2020

In the Image processing and computer vision, the problem of reconstructing from one image to another or generating a new image has been steadily drawing attention as hardware advances. However, the problem of computer-generated images also continues to emerge when viewed with human eyes because it is not natural. Due to the recent active research in deep learning, image generating and improvement problem using it are also actively being studied, and among them, the network called Generative Adversarial Network(GAN) is doing well in the image generating. Various models of GAN have been presented since the proposed GAN, allowing for the generation of more natural images compared to the results of research in the image generating. Among them, pix2pix is a conditional GAN model, which is a general-purpose network that shows good performance in various datasets. pix2pix is based on U-Net, but there are many networks that show better performance among U-Net based networks. Therefore, in this study, images are generated by applying various networks to U-Net of pix2pix, and the results are compared and evaluated. The images generated through each network confirm that the pix2pix model with Attention, R2, and Attention-R2 networks shows better performance than the existing pix2pix model using U-Net, and check the limitations of the most powerful network. It is suggested as a future study.

• Korean Journal of Remote Sensing
• /
• v.35 no.5_1
• /
• pp.715-726
• /
• 2019

With the increasing prevalence of high-resolution satellite images, the need for technology to generate accurate 3D information from the satellite images is emphasized. In order to create a digital terrain model (DTM) that is widely used in applications such as change detection and object extraction, it is necessary to extract trees, buildings, etc. that exist in the digital surface model (DSM) and estimate the height of the ground. This paper presents a method for automatically generating DTM from DSM extracted from KOMPSAT-3A stereo images. The technique was developed to detect the non-ground area and estimate the height value of the ground by using the previously constructed low-resolution topographic data. The average vertical accuracy of DTMs generated in the four experimental sites with various topographical characteristics, such as mountainous terrain, densely built area, flat topography, and complex terrain was about 5.8 meters. The proposed technique would be useful to produce high-quality DTMs that represent precise features of the bare-earth's surface.

• Asian Journal for Public Opinion Research
• /
• v.1 no.2
• /
• pp.90-102
• /
• 2014

By integrating useful insights from social cognitive theory and social capital theory, we aim to develop a model for better understanding people's behaviors related to the use of social networking sites (SNSs) and formalize the role of social capital in individuals' continuous SNS use. Propositions that emphasize the triadic interactive relationships among environmental, personal, and behavioral factors were highlighted in this study. After reviewing previous studies, in this paper we proposed the following: (1) the causation between SNS use and individuals' perceived social capital might be mutual; social capital may not only be the result of media selectivity, but could also be an essential stimulus initiating the start of using SNSs; (2) the influences of SNSs use on the generation of individuals' online social capital might be conditional upon particular patterns of use; (3) both the level of dependence on SNSs and the differentiated patterns of SNSs use vary according to individuals' perceived offline social capital and their personal characteristics, for instance, personality or self-construal, and social anxiety.

• Annual Conference on Human and Language Technology
• /
• 2022.10a
• /
• pp.277-283
• /
• 2022

사주 풀이란 주어진 사주에 대해서 그에 맞는 해석 글을 생성해주는 작업을 의미한다. 전통적으로 사주 풀이는 온전한 사람의 영역으로 인식되어왔으나, 우리는 본 연구를 통해 사주 풀이 영역도 인공지능으로 대체할 수 있을 것이라는 가능성을 탐구한다. 본 연구에서 우리는 최근 연구되고 있는 자연어 생성분야의 연구들에서 영감을 받아, 사주 유형과 사주 풀이 내에 포함할 명사 키워드를 기반으로 풀이글을 생성하는 인공지능 모델 SaJuTeller를 설계한다. 특히 이전 문맥을 고려하여 풀이글을 생성하는 모델과 단순 사주 유형 및 명사 키워드를 기반으로 풀이글을 생성하는 두가지 모델을 제안하며, 이들 각각의 성능을 분석함으로써 각 모델의 구체적인 활용 방안을 제안한다. 본 연구는 우리가 아는 한 최초의 인공지능 기반 사주풀이 연구이며, 우리는 이를 통해 사주풀이에 요구되는 전문인력의 노력을 경감시킴과 동시에, 다양한 표현을 가진 사주 풀이 글을 생성할 수 있음을 제안한다.

• Water for future
• /
• v.29 no.6
• /
• pp.179-188
• /
• 1996

The chronical sequences of daily precipitation are of great practical importance in the planning and operational processes of water resources system. A sequence of days with alternate dry day and wet day can be generated by two state Markov chain model that establish the subsequent daily state as wet or dry by previously calculated vconditional probabilities depending on the state of previous day. In this study, a synthetic generation model for obtaining the daily precipitation series is presented by classifying the precipitation amount in wet days into multi-states. To apply multi-state Markov chain model, the daily precipitation amounts for wet day are rearranged by grouping into thirty states with intervals for each state. Conditional probabilities as transition probability matrix are estimated from the computational scheme for stepping from the precipitation on one day to that on the following day. Statistical comparisons were made between the historical and synthesized chracteristics of daily precipitation series. From the results, it is shown that the proposed method is available to generate and simulate the daily precipitation series with fair accuracy and conserve the general statistical properties of historical precipitation series.

• International conference on construction engineering and project management
• /
• 2024.07a
• /
• pp.1294-1294
• /
• 2024

Flood risk maps are used in urban flooding to understand the spatial extent and depth of inundation damage. To construct these maps, hydrodynamic modeling capable of simulating flood waves is necessary. Flood waves are typically fast, and inundation patterns can significantly vary depending on the terrain, making it essential to accurately represent the terrain of the flood source in flood wave analysis. Recently, methods using UAVs for terrain data construction through Structure-from-Motion or LiDAR have been utilized. These methods are crucial for UAV operations, and thus, still require a lot of time and manpower, and are limited when UAV operations are not possible. Therefore, for efficient nationwide monitoring, this study developed a model that can automatically generate terrain data by estimating depth information from a single image using c-GAN (Conditional Generative Adversarial Networks) and BBDM (Brownian Bridge Diffusion Model). The training, utilization, and validation datasets employed images from the ISPRS (2018) and directly aerial photographed image sets from five locations in the territory of the Republic of Korea. Compared to the ground truth of the test data set, it is considered sufficiently usable as terrain data for flood wave analysis, capable of generating highly accurate and precise terrain data with high reproducibility.

• Journal of Energy Engineering
• /
• v.2 no.2
• /
• pp.200-207
• /
• 1993

A methanol reformer was designed and fabricated using a CuO-ZnO low temperature shift catalyst, and its operation characteristics have been studied for the phosphoric acid fuel cell (PAFC) power generation system. The type of reactor was annular Methanol was consumed both for heating and for reforming fuel. Contents of carbon monoxide produced from the reformer increased as the reaction temperatures increased, but decreased as the mole ratios of water to methanol(H$_2$O/CH$_3$OH) increased. At steady state operating conditional, temperature profile of the catalytic reactor of the reformer was well coincide with the model equation, and it took 50 minutes from start to the rated condition of the reformer. When the system was operated at 4/4 and 1/4 of load, thermal efficiencies of the system were 72.3% and 77%, respectively. When the PAFC system was operated with reformed gas in the range of 62 V-37.6 V and 0-147 A, the trend of I-V curve showed a typical fuel tell characteristic. At steady state condition, the flow rates of reforming and combustion methanol were 88.1 mol/h and 50.1 mol/h, respectively.

• Journal of Internet Computing and Services
• /
• v.21 no.6
• /
• pp.23-31
• /
• 2020

This study was carried out to generate various images of railroad surfaces with random defects as training data to be better at the detection of defects. Defects on the surface of railroads are caused by various factors such as friction between track binding devices and adjacent tracks and can cause accidents such as broken rails, so railroad maintenance for defects is necessary. Therefore, various researches on defect detection and inspection using image processing or machine learning on railway surface images have been conducted to automate railroad inspection and to reduce railroad maintenance costs. In general, the performance of the image processing analysis method and machine learning technology is affected by the quantity and quality of data. For this reason, some researches require specific devices or vehicles to acquire images of the track surface at regular intervals to obtain a database of various railway surface images. On the contrary, in this study, in order to reduce and improve the operating cost of image acquisition, we constructed the 'Defective Railroad Surface Regeneration Model' by applying the methods presented in the related studies of the Generative Adversarial Network (GAN). Thus, we aimed to detect defects on railroad surface even without a dedicated database. This constructed model is designed to learn to generate the railroad surface combining the different railroad surface textures and the original surface, considering the ground truth of the railroad defects. The generated images of the railroad surface were used as training data in defect detection network, which is based on Fully Convolutional Network (FCN). To validate its performance, we clustered and divided the railroad data into three subsets, one subset as original railroad texture images and the remaining two subsets as another railroad surface texture images. In the first experiment, we used only original texture images for training sets in the defect detection model. And in the second experiment, we trained the generated images that were generated by combining the original images with a few railroad textures of the other images. Each defect detection model was evaluated in terms of 'intersection of union(IoU)' and F1-score measures with ground truths. As a result, the scores increased by about 10~15% when the generated images were used, compared to the case that only the original images were used. This proves that it is possible to detect defects by using the existing data and a few different texture images, even for the railroad surface images in which dedicated training database is not constructed.