• Proceedings of the Korea Contents Association Conference
• /
• 2008.05a
• /
• pp.892-896
• /
• 2008

In construction work's process of a basic planning, we must carry out a topographical for construction reserved land and to the basis of this, production of profile is Indispensable factor for the purpose of every construction work such as road, rail way, canal and etc. From this research, the production of profile about construction reserved land, using topography information of numerical value topographical map produced by NGIS construction project, construct precision 3D topography model and from this, propose plans for utilizing by producing automatic profile. With the aim of this, abstract every layers of main facilities and altitude from numerical value topographical map and while producing 3D topography model by using this, we produce and product automatic profile from construct precision 3D topography. And about the produced profile's process, a topographical by mutual analysis was carried out to check whether the automatic produced profile's accuracy could be accepted at actual estimation and permissible accuracy. It is considered that the result of the research could be suggested as a new formation techniques concept which can reduce the designing time and expenses and increase the efficiency of planning in the basic design process for the construction.

• Proceedings of the Korean Vacuum Society Conference
• /
• 2013.08a
• /
• pp.99-99
• /
• 2013

In this talk, I will present two research works in progress, which are: i) mapping of piezoelectric polarization and associated charge density distribution in the heteroepitaxial InGaN/GaN multi-quantum well (MQW) structure of a light emitting diode (LED) by using inline electron holography and ii) in-situ observation of the polarization switching process of an ferroelectric Pb(Zr1-x,Tix)O3 (PZT) thin film capacitor under an applied electric field in transmission electron microscope (TEM). In the first part, I will show that strain as well as total charge density distributions can be mapped quantitatively across all the functional layers constituting a LED, including n-type GaN, InGaN/GaN MQWs, and p-type GaN with sub-nm spatial resolution (~0.8 nm) by using inline electron holography. The experimentally obtained strain maps were verified by comparison with finite element method simulations and confirmed that not only InGaN QWs (2.5 nm in thickness) but also GaN QBs (10 nm in thickness) in the MQW structure are strained complementary to accommodate the lattice misfit strain. Because of this complementary strain of GaN QBs, the strain gradient and also (piezoelectric) polarization gradient across the MQW changes more steeply than expected, resulting in more polarization charge density at the MQW interfaces than the typically expected value from the spontaneous polarization mismatch alone. By quantitative and comparative analysis of the total charge density map with the polarization charge map, we can clarify what extent of the polarization charges are compensated by the electrons supplied from the n-doped GaN QBs. Comparison with the simulated energy band diagrams with various screening parameters show that only 60% of the net polarization charges are compensated by the electrons from the GaN QBs, which results in the internal field of ~2.0 MV cm-1 across each pair of GaN/InGaN of the MQW structure. In the second part of my talk, I will present in-situ observations of the polarization switching process of a planar Ni/PZT/SrRuO3 capacitor using TEM. We observed the preferential, but asymmetric, nucleation and forward growth of switched c-domains at the PZT/electrode interfaces arising from the built-in electric field beneath each interface. The subsequent sideways growth was inhibited by the depolarization field due to the imperfect charge compensation at the counter electrode and preexisting a-domain walls, leading to asymmetric switching. It was found that the preexisting a-domains split into fine a- and c-domains constituting a $90^{\circ}$ stripe domain pattern during the $180^{\circ}$ polarization switching process, revealing that these domains also actively participated in the out-of-plane polarization switching. The real-time observations uncovered the origin of the switching asymmetry and further clarified the importance of charged domain walls and the interfaces with electrodes in the ferroelectric switching processes.

• The Journal of Engineering Geology
• /
• v.28 no.4
• /
• pp.727-740
• /
• 2018

The media in the undersaturated zone is defined as the uppermost layer of the water table at which the groundwater is unsaturated or saturated discontinuously. The properties of the unsaturated zone can affect the reduction of contaminants that flow from the lower part of soil to the water table. In recent, there have been problems in evaluating groundwater contaminations vulnerability because weighted value for permeability is given, regardless of anisotropy and heterogeneity in the unsaturated media. Geological media have various ranges of permeability. When applying the weighted value, representative of permeability for grain sizes standardized, to construction of contamination vulnerability, it will produce more exaggerated result than the case that considers unsaturated geological properties. In this study, we performed laboratory column tests considering two sets of the unsaturated layers in order to investigate the permeability in anisotropic unsaturated zone with anisotropy. On the basis of the tests, average permeability coefficients were calculated considering the properties of unsaturated media obtained from drill cores in the field. The final contamination vulnerability map constructed shows that the contamination vulnerability map applying the properties of geological media of the unsaturated zone coincides much better with the results measured in the field, compared to the case of contamination vulnerability considering the weighted value in the unsaturated zone.

• Journal of Intelligence and Information Systems
• /
• v.29 no.4
• /
• pp.15-30
• /
• 2023

Recently, various services using artificial intelligence(AI) are emerging in the media field as well However, most of the video editing, which involves finding an editing point and attaching the video, is carried out in a passive manner, requiring a lot of time and human resources. Therefore, this study proposes a methodology that can detect the edit points of video according to whether person in video are spoken by using Video Swin Transformer. First, facial keypoints are detected through face alignment. To this end, the proposed structure first detects facial keypoints through face alignment. Through this process, the temporal and spatial changes of the face are reflected from the input video data. And, through the Video Swin Transformer-based model proposed in this study, the behavior of the person in the video is classified. Specifically, after combining the feature map generated through Video Swin Transformer from video data and the facial keypoints detected through Face Alignment, utterance is classified through convolution layers. In conclusion, the performance of the image editing point detection model using facial keypoints proposed in this paper improved from 87.46% to 89.17% compared to the model without facial keypoints.

• Proceedings of the KOSOMBE Conference
• /
• v.1991 no.05
• /
• pp.65-70
• /
• 1991

A hybrid model which uses a probabilistic model and a MLP(multi layer perceptron) model for pattern recognition of EMG(electromyogram) signals is proposed in this paper. MLP model has problems which do not guarantee global minima of error due to learning method and have different approximation grade to bayesian probabilities due to different amounts and quality of training data, the number of hidden layers and hidden nodes, etc. Especially in the case of new test data which exclude design samples, the latter problem produces quite different results. The error probability of probabilistic model is closely related to the estimation error of the parameters used in the model and fidelity of assumtion. Generally, it is impossible to introduce the bayesian classifier to the probabilistic model of EMG signals because of unknown priori probabilities and is estimated by MLE(maximum likelihood estimate). In this paper we propose the method which get the MAP(maximum a posteriori probability) in the probabilistic model by estimating the priori probability distribution which minimize the error probability using the MLP. This method minimize the error probability of the probabilistic model as long as the realization of the MLP is optimal and approximate the minimum of error probability of each class of both models selectively. Alocating the reference coordinate of EMG signal to the outside of the body make it easy to suit to the applications which it is difficult to define and seperate using internal body coordinate. Simulation results show the benefit of the proposed model compared to use the MLP and the probabilistic model seperately.

• Journal of the Korean Association of Geographic Information Studies
• /
• v.2 no.2
• /
• pp.1-8
• /
• 1999

Information representing the thickness of the original lines from the thinning results can be used efficiently in order to implement the automated vectorizing system. This paper proposes a preprocessing scheme of thinning which can show the information of the original lines' thickness from the thinning result. In the proposed scheme, the depth of each pixel constructing the lines was calculated, which was represented by the number of layers composed of neighboring pixels surrounding the original pixel. And then the original lines' thickness could be recognized through the depth information of the skeleton from the thinning results. We implemented the proposed scheme and experimented on a contour map. Using the depth information of the skeleton, we could easily distinguish each line of the contour either an intermediate or an index contour.

• The Journal of the Acoustical Society of Korea
• /
• v.39 no.6
• /
• pp.592-599
• /
• 2020

This paper proposes a Convolutional Recurrent Neural Net (CRNN) structure that can simultaneously reflect both static and dynamic characteristics of seismic waveforms for various earthquake events classification. Addressing various earthquake events, including not only micro-earthquakes and artificial-earthquakes but also macro-earthquakes, requires both effective feature extraction and a classifier that can discriminate seismic waveform under noisy environment. First, we extract the static characteristics of seismic waveform through an attention-based convolution layer. Then, the extracted feature-map is sequentially injected as input to a multi-input single-output Long Short-Term Memory (LSTM) network structure to extract the dynamic characteristic for various seismic event classifications. Subsequently, we perform earthquake events classification through two fully connected layers and softmax function. Representative experimental results using domestic and foreign earthquake database show that the proposed model provides an effective structure for various earthquake events classification.

• Journal of Advanced Navigation Technology
• /
• v.19 no.6
• /
• pp.484-491
• /
• 2015

Aerial scene is greatly increased by the introduction and supply of the image due to the growth of digital optical imaging technology and development of the UAV. It has been used as the extraction of ground properties, classification, change detection, image fusion and mapping based on the aerial image. In particular, in the image analysis and utilization of deep learning algorithm it has shown a new paradigm to overcome the limitation of the field of pattern recognition. This paper presents the possibility to apply a more wide range and various fields through the segmentation and classification of aerial scene based on the Deep learning(ConvNet). We build 4-classes image database consists of Road, Building, Yard, Forest total 3000. Each of the classes has a certain pattern, the results with feature vector map come out differently. Our system consists of feature extraction, classification and training. Feature extraction is built up of two layers based on ConvNet. And then, it is classified by using the Multilayer perceptron and Logistic regression, the algorithm as a classification process.

• International Journal of Highway Engineering
• /
• v.7 no.2 s.24
• /
• pp.87-95
• /
• 2005

To prevent the lowering of structural capability due to freezing and thawing in cold winter, numerous researches on frost heaving have been performed. As the result, the freezing index contour map of the Korea peninsula has been made for the design of the anti-freezing layer of pavements. However, the validity of the anti-freezing layer needs to be evaluated because systematic investigations on the variation of freezing depth with the thickness and material types of pavement layers and the configuration of the ground have been rarely performed. The freezing index of the Korea Highway Corporation test road site was calculated and the freezing depths of the concrete and asphalt pavements of the test road were investigated using the ambient and pavement temperature and water content. In addition, the investigated freezing depths were compared to the values estimated by existing freezing depth models. This is the preliminary study on the freezing-related data measured at the test road. The results with higher reliability will be produced by the long-term accumulation of the data and the analysis on it.

• The Journal of the Korea Contents Association
• /
• v.6 no.4
• /
• pp.19-28
• /
• 2006

XML instances for purpose of information exchange are normally stored in the legacy relational database. Therefore, integrations with relational database are required for effective XML applications. To support these requirements, virtual decomposition storage or decomposition storage methods which save separates structures of instances to relational database have researched. However, these storage methods contain different information of schema structure and layers which has caused difficulties to process query during search operation as well as increased overheads due to duplicate savings for separate storages. Therefore, in this research, additional field of 'Eltype' has introduced to previous database schema structure to instance and schema structure, provide consistent level information and propose storage structure to map each field to schema field of relational database. As results, XML instance and structures can be stored together to minimize overheads and required storage-space. Also, synchronized storage layer structure provides easier processing of search query.