• Publications of The Korean Astronomical Society
• /
• v.30 no.2
• /
• pp.645-647
• /
• 2015

We have initiated a Very Long Baseline Interferometer (VLBI) monitoring project of 36 methanol maser sources at 6.7 GHz using the Japanese VLBI Network (JVN) and East-Asian VLBI Network (EAVN), starting in August 2010. The purpose of this project is to systematically reveal 3-dimensional (3-D) kine-matics of rotating disks around forming high-mass protostars. As an initial result, we present proper mo- tion detections for two methanol maser sources showing an elliptical spatial morphology, G 002.53+00.19 and G 006.79-00.25, which could be the best candidates associated with the disk. The detected proper motions indicate a simple rotation in G 002.53+00.19 and rotation with expansion in G 006.79-00.25, respectively, on the basis of disk model fits with rotating and expanding components. The expanding motions might be caused by the magnetic-centrifugal wind on the disk.

• Journal of Convergence for Information Technology
• /
• v.11 no.11
• /
• pp.57-65
• /
• 2021

We propose a novel method to detect abnormal data of specific symptoms using deep learning in air pollution measurement system. Existing methods generally detect abnomal data by classifying data showing unusual patterns different from the existing time series data. However, these approaches have limitations in detecting specific symptoms. In this paper, we use DeepLab V3+ model mainly used for foreground segmentation of images, whose structure has been changed to handle one-dimensional data. Instead of images, the model receives time-series data from multiple sensors and can detect data showing specific symptoms. In addition, we improve model's performance by reducing the complexity of noisy form time series data by using 'piecewise aggregation approximation'. Through the experimental results, it can be confirmed that anomaly data detection can be performed successfully.

• Korean Journal of Remote Sensing
• /
• v.34 no.5
• /
• pp.811-827
• /
• 2018

The purpose of this study is to compare machine learning algorithm and deep learning algorithm in crop classification using multi-temporal remote sensing data. For this, impacts of machine learning and deep learning algorithms on (a) hyper-parameter and (2) training sample size were compared and analyzed for Haenam-gun, Korea and Illinois State, USA. In the comparison experiment, support vector machine (SVM) was applied as machine learning algorithm and convolutional neural network (CNN) was applied as deep learning algorithm. In particular, 2D-CNN considering 2-dimensional spatial information and 3D-CNN with extended time dimension from 2D-CNN were applied as CNN. As a result of the experiment, it was found that the hyper-parameter values of CNN, considering various hyper-parameter, defined in the two study areas were similar compared with SVM. Based on this result, although it takes much time to optimize the model in CNN, it is considered that it is possible to apply transfer learning that can extend optimized CNN model to other regions. Then, in the experiment results with various training sample size, the impact of that on CNN was larger than SVM. In particular, this impact was exaggerated in Illinois State with heterogeneous spatial patterns. In addition, the lowest classification performance of 3D-CNN was presented in Illinois State, which is considered to be due to over-fitting as complexity of the model. That is, the classification performance was relatively degraded due to heterogeneous patterns and noise effect of input data, although the training accuracy of 3D-CNN model was high. This result simply that a proper classification algorithms should be selected considering spatial characteristics of study areas. Also, a large amount of training samples is necessary to guarantee higher classification performance in CNN, particularly in 3D-CNN.

• Journal of Broadcast Engineering
• /
• v.16 no.5
• /
• pp.750-761
• /
• 2011

Currently, 3DTV technologies are being developed as the future services of the HD digital broadcast environment. As one of the various research topics to apply 3DTV technologies to the conventional broadcasting network, methods to configure stereoscopic videos are being studied. In this paper, we proposed a method to broadcast high quality stereoscopic videos based on analysis of a method to add a stereoscopic descriptor to the PMT of MPEG-2 transport streams and a method to transmit stereoscopic videos by the expansion of MPEG-C part 3 which are from precedent studies. The proposed technique maintains compatibility with conventional MPEG-2 transport streams by showing only reference video for models that do not support 3D broadcasting. Therefore, the compatibility between conventional broadcasting and stereoscopic videos should make this method useful when activating 3D services in the communications and broadcasting area

• Journal of the Korean Association of Geographic Information Studies
• /
• v.19 no.3
• /
• pp.61-74
• /
• 2016

The purpose of this paper is to analyze the relationship between the location of the epicenter of a medium-sized earthquake(magnitude 4.8) that occurred on January 20, 2007 in the Odaesan area with lineament features using a shaded relief map(1/25,000 scale) and satellite images from LANDSAT-8 and KOMPSAT-2. Previous studies have analyzed lineament features in tectonic settings primarily by examining two-dimensional satellite images and shaded relief maps. These methods, however, limit the application of the visual interpretation of relief features long considered as the major component of lineament extraction. To overcome some existing limitations of two-dimensional images, this study examined three-dimensional images, produced from a Digital Elevation Model and drainage network map, for lineament extraction. This approach reduces mapping errors introduced by visual interpretation. In addition, spline interpolation was conducted to produce density maps of lineament frequency, intersection, and length required to estimate the density of lineament at the epicenter of the earthquake. An algorithm was developed to compute the Value of the Relative Density(VRD) representing the relative density of lineament from the map. The VRD is the lineament density of each map grid divided by the maximum density value from the map. As such, it is a quantified value that indicates the concentration level of the lineament density across the area impacted by the earthquake. Using this algorithm, the VRD calculated at the earthquake epicenter using the lineament's frequency, intersection, and length density maps ranged from approximately 0.60(min) to 0.90(max). However, because there were differences in mapped images such as those for solar altitude and azimuth, the mean of VRD was used rather than those categorized by the images. The results show that the average frequency of VRD was approximately 0.85, which was 21% higher than the intersection and length of VRD, demonstrating the close relationship that exists between lineament and the epicenter. Therefore, it is concluded that the density map analysis described in this study, based on lineament extraction, is valid and can be used as a primary data analysis tool for earthquake research in the future.

• Journal of the Korean Geotechnical Society
• /
• v.37 no.5
• /
• pp.47-63
• /
• 2021

Recently, the demand for three-dimensional (3D) underground maps from the perspective of digital twins and the demand for linkage utilization are increasing. However, the vastness of national geotechnical survey data and the uncertainty in applying geostatistical techniques pose challenges in modeling underground regional geotechnical characteristics. In this study, an optimal learning model based on multi-layer perceptron (MLP) was constructed for 3D subsurface lithological and geotechnical classification in Seoul, South Korea. First, the geotechnical layer and 3D spatial coordinates of each borehole dataset in the Seoul area were constructed as a geotechnical database according to a standardized format, and data pre-processing such as correction and normalization of missing values for machine learning was performed. An optimal fitting model was designed through hyperparameter optimization of the MLP model and model performance evaluation, such as precision and accuracy tests. Then, a 3D grid network locally assigning geotechnical layer classification was constructed by applying an MLP-based bet-fitting model for each unit lattice. The constructed 3D geotechnical layer map was evaluated by comparing the results of a geostatistical interpolation technique and the topsoil properties of the geological map.

• Journal of the Mineralogical Society of Korea
• /
• v.22 no.4
• /
• pp.313-324
• /
• 2009

We explore the effect of particle shape and size on 3-dimensional (3D) network and pore structure of porous earth materials composed of glass beads and silica gel using NMR micro-imaging in order to gain better insights into relationship between structure and the corresponding hydrologic and seismological properties. The 3D micro-imaging data for the model porous networks show that the specific surface area, porosity, and permeability range from 2.5 to $9.6\;mm^2/mm^3$, from 0.21 to 0.38, and from 11.6 to 892.3 D (Darcy), respectively, which are typical values for unconsolidated sands. The relationships among specific surface area, porosity, and permeability of the porous media are relatively well explained with the Kozeny equation. Cube counting fractal dimension analysis shows that fractal dimension increases from ~2.5－2.6 to 3.0 with increasing specific surface area from 2.5 to $9.6\;mm^2/mm^3$, with the data also suggesting the effect of porosity. Specific surface area, porosity, permeability, and cube counting fractal dimension for the natural mongolian sandstone are $0.33\;mm^2/mm^3$, 0.017, 30.9 mD, and 1.59, respectively. The current results highlight that NMR micro-imaging, together with detailed statistical analyses can be useful to characterize 3D pore structures of various porous earth materials and be potentially effective in accounting for transport properties and seismic wave velocity and attenuation of diverse porous media in earth crust and interiors.

• Journal of KIISE:Databases
• /
• v.33 no.3
• /
• pp.299-309
• /
• 2006

Due to the rapid development in mobile communication technologies, the usage of mobile devices such as cell phone or PDA becomes increasingly popular. As different devices require different applications, various new services are being developed to satisfy the needs. One of the popular services under heavy demand is the Location-based Service (LBS) that exploits the spatial information of moving objects per temporal changes. In order to support LBS efficiently, it is necessary to be able to index and query well a large amount of spatio-temporal information of moving objects. Therefore, in this paper, we investigate how such location information of moving objects can be efficiently stored and indexed. In particular, we propose a novel location encoding method based on hierarchical administrative district information. Our proposal is different from conventional approaches where moving objects are often expressed as geometric points in two dimensional space, (x,y). Instead, in ours, moving objects are encoded as one dimensional points by both administrative district as well as road information. Our method is especially useful for monitoring traffic situation or tracing location of moving objects through approximate spatial queries.

• Journal of Intelligence and Information Systems
• /
• v.24 no.1
• /
• pp.205-225
• /
• 2018

Convolutional Neural Network (ConvNet) is one class of the powerful Deep Neural Network that can analyze and learn hierarchies of visual features. Originally, first neural network (Neocognitron) was introduced in the 80s. At that time, the neural network was not broadly used in both industry and academic field by cause of large-scale dataset shortage and low computational power. However, after a few decades later in 2012, Krizhevsky made a breakthrough on ILSVRC-12 visual recognition competition using Convolutional Neural Network. That breakthrough revived people interest in the neural network. The success of Convolutional Neural Network is achieved with two main factors. First of them is the emergence of advanced hardware (GPUs) for sufficient parallel computation. Second is the availability of large-scale datasets such as ImageNet (ILSVRC) dataset for training. Unfortunately, many new domains are bottlenecked by these factors. For most domains, it is difficult and requires lots of effort to gather large-scale dataset to train a ConvNet. Moreover, even if we have a large-scale dataset, training ConvNet from scratch is required expensive resource and time-consuming. These two obstacles can be solved by using transfer learning. Transfer learning is a method for transferring the knowledge from a source domain to new domain. There are two major Transfer learning cases. First one is ConvNet as fixed feature extractor, and the second one is Fine-tune the ConvNet on a new dataset. In the first case, using pre-trained ConvNet (such as on ImageNet) to compute feed-forward activations of the image into the ConvNet and extract activation features from specific layers. In the second case, replacing and retraining the ConvNet classifier on the new dataset, then fine-tune the weights of the pre-trained network with the backpropagation. In this paper, we focus on using multiple ConvNet layers as a fixed feature extractor only. However, applying features with high dimensional complexity that is directly extracted from multiple ConvNet layers is still a challenging problem. We observe that features extracted from multiple ConvNet layers address the different characteristics of the image which means better representation could be obtained by finding the optimal combination of multiple ConvNet layers. Based on that observation, we propose to employ multiple ConvNet layer representations for transfer learning instead of a single ConvNet layer representation. Overall, our primary pipeline has three steps. Firstly, images from target task are given as input to ConvNet, then that image will be feed-forwarded into pre-trained AlexNet, and the activation features from three fully connected convolutional layers are extracted. Secondly, activation features of three ConvNet layers are concatenated to obtain multiple ConvNet layers representation because it will gain more information about an image. When three fully connected layer features concatenated, the occurring image representation would have 9192 (4096+4096+1000) dimension features. However, features extracted from multiple ConvNet layers are redundant and noisy since they are extracted from the same ConvNet. Thus, a third step, we will use Principal Component Analysis (PCA) to select salient features before the training phase. When salient features are obtained, the classifier can classify image more accurately, and the performance of transfer learning can be improved. To evaluate proposed method, experiments are conducted in three standard datasets (Caltech-256, VOC07, and SUN397) to compare multiple ConvNet layer representations against single ConvNet layer representation by using PCA for feature selection and dimension reduction. Our experiments demonstrated the importance of feature selection for multiple ConvNet layer representation. Moreover, our proposed approach achieved 75.6% accuracy compared to 73.9% accuracy achieved by FC7 layer on the Caltech-256 dataset, 73.1% accuracy compared to 69.2% accuracy achieved by FC8 layer on the VOC07 dataset, 52.2% accuracy compared to 48.7% accuracy achieved by FC7 layer on the SUN397 dataset. We also showed that our proposed approach achieved superior performance, 2.8%, 2.1% and 3.1% accuracy improvement on Caltech-256, VOC07, and SUN397 dataset respectively compare to existing work.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.41 no.2
• /
• pp.79-88
• /
• 2004

Gaze detection is to locate the position on a monitor screen where a user is looking by computer vision. Gaze detection systems have numerous fields of application. They are applicable to the man-machine interface for helping the handicapped to use computers and the view control in three dimensional simulation programs. In our work, we implement it with a computer vision system setting a IR-LED based single camera. To detect the gaze position, we locate facial features, which is effectively performed with IR-LED based camera and SVM(Support Vector Machine). When a user gazes at a position of monitor, we can compute the 3D positions of those features based on 3D rotation and translation estimation and affine transform. Finally, the gaze position by the facial movements is computed from the normal vector of the plane determined by those computed 3D positions of features. In addition, we use a trained neural network to detect the gaze position by eye's movement. As experimental results, we can obtain the facial and eye gaze position on a monitor and the gaze position accuracy between the computed positions and the real ones is about 4.8 cm of RMS error.