• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.17 no.2
• /
• pp.15-23
• /
• 2017

The person Re-identification is the most challenging part of computer vision due to the significant changes in human pose and background clutter with occlusions. The picture from non-overlapping cameras enhance the difficulty to distinguish some person from the other. To reach a better performance match, most methods use feature selection and distance metrics separately to get discriminative representations and proper distance to describe the similarity between person and kind of ignoring some significant features. This situation has encouraged us to consider a novel method to deal with this problem. In this paper, we proposed an enhanced recurrent neural network with three-tier hierarchical network for person re-identification. Specifically, the proposed recurrent neural network (RNN) model contain an iterative expectation maximum (EM) algorithm and three-tier Hierarchical network to jointly learn both the discriminative features and metrics distance. The iterative EM algorithm can fully use of the feature extraction ability of convolutional neural network (CNN) which is in series before the RNN. By unsupervised learning, the EM framework can change the labels of the patches and train larger datasets. Through the three-tier hierarchical network, the convolutional neural network, recurrent network and pooling layer can jointly be a feature extractor to better train the network. The experimental result shows that comparing with other researchers' approaches in this field, this method also can get a competitive accuracy. The influence of different component of this method will be analyzed and evaluated in the future research.

• KIPS Transactions on Software and Data Engineering
• /
• v.3 no.9
• /
• pp.361-368
• /
• 2014

The LPR(License plate recognition) system has been developed to efficient control for complex traffic environment and currently be used in many places. However, because of light, noise, background changes, environmental changes, damaged plate, it only works limited environment, so it is difficult to use in real-time. This paper presents a heuristic segmentation algorithm for robust to noise and illumination changes and introduce a real-time license plate recognition system using it. In first step, We detect the plate utilized Haar-like feature and Adaboost. This method is possible to rapid detection used integral image and cascade structure. Second step, we determine the type of license plate with adaptive histogram equalization, bilateral filtering for denoise and segment accurate character based on adaptive threshold, pixel projection and associated with the prior knowledge. The last step is character recognition that used histogram of oriented gradients (HOG) and multi-layer perceptron(MLP) for number recognition and support vector machine(SVM) for number and Korean character classifier respectively. The experimental results show license plate detection rate of 94.29%, license plate false alarm rate of 2.94%. In character segmentation method, character hit rate is 97.23% and character false alarm rate is 1.37%. And in character recognition, the average character recognition rate is 98.38%. Total average running time in our proposed method is 140ms. It is possible to be real-time system with efficiency and robustness.

• Spatial Information Research
• /
• v.7 no.2
• /
• pp.159-173
• /
• 1999

GIS has been regarded as one of important tools or methodologies for various geoscience applications. Recently, spatial data integration schemes for site-specific or field-specific thematic mapping are newly developed and utilized. However, these kinds of approaches are somewhat insufficient quantitative assessment of integrated layers towards known targets in-detailed . Moreover, GIS analysis scheme is rarely extended to scientific approaches. In this study, simple approach of Multi-Buffer Zone Analysis , related to GIS analystical aspect, is addressed and an actual application for predicting or favorable mapping of mineral occurrences, one of GIS-based geoscientific approaches, is performed, As for geo-processing in GIS itself, this scheme can be regarded as extension or adaptiation of cell-based buffering or proximity analysis to geoscientific data interpretation. This study is based on rationale that surface geological pattern around primitives such as a point, a line, or a polygon in GIS, representing significant geological features, can be efficiently utilized to delineate complex geological behaviors or events, especially handling multiple dta sets originated from multiple sources such as airborne geophysical/radiometric exploration, field survey, and even a classified image of remote sensing. Conclusively, this methodology associated wit GIS is though to be helpful to analyze the spatial pattern of multiple data, pointing given sources, and is expected to effectively utilize for exploratory analysis of cell-based resultant layer integrated with complex or different data sources.

• Geophysics and Geophysical Exploration
• /
• v.11 no.2
• /
• pp.109-115
• /
• 2008

It is very important to estimate the physical properties of survey area and delineate the geological basement in marine site survey for the design of offshore structures. For the purpose of providing high quality data by means of engineering site survey, it is necessary to apply several survey techniques and carry out the integrated interpretation to each other. In this study, we applied single channel seismic reflection method and OBC (Ocean Bottom Cable) type seismic refraction method at shallow marine. We used a dual boomer-single channel streamer as a source-receiver in seismic reflection survey and airgun source-the developed OBC type streamer in seismic refraction survey. We made 24 channels OBC type streamer which has 4m channel interval and each channel is composed of single hydrophone and preamplifier. We tested the field applicability of the proposed method and applied the typical seismic data processing methods to the obtained reflection data in order to enhance the data quality and image resolution. In order to estimate the geological velocity distribution from refraction data, seismic refraction tomography technique was applied. Therefore, we could successfully perform time-depth conversion using the velocity information as an integrated interpretation. The proposed method could provide reliable geologic information such as sediment layer thickness and 3D basement depth map.

• Journal of Korean Society of Coastal and Ocean Engineers
• /
• v.9 no.2
• /
• pp.86-96
• /
• 1997

In order to investigate the formation of a shallow water front and its relation to water quality distributions, oceanographic measurements were made, and the numerical computations of the Simpson-Hunter stratification parameter log(H/U$^3$) were performed. It is shown from satellite image and hydrographic data that the shallow water front is formed near the northern Kaduk channel, and the stratification parameter log(H/U$^3$) near the front is in a range of 2.0-2.5. Measured COD (Chemical Oxygen Demand) concentrations in offshore region of the front and in the western part of the bay are below 2.0 mg/1. whereas the concentrations in Masan Bay located in the northern inside of the frontal zone are high as 3.0-5.5 mg/1. COD concentrations decrease gradually from Masan Bay toward the offshore due to the dilution by strong water mixing. Anoxic and hypoxic water masses at the bottom layer in summcr occur in the western part of Chinhae Bay and in Masan Bay, and DO (Dissolved Oxygen) concentrations become low with increasing the stratification parameter. DO concentrations outside the front are more than about 4.0 mg/1, whereas the concentrations inside the front are low. The shallow water front plays a significant role for material transport from coastal area to oceanic area, and the frontal region seems to be important physical and chemical boundaries.

• Journal of the Korean Geophysical Society
• /
• v.2 no.1
• /
• pp.75-88
• /
• 1999

A recent major subject of geophysical exploration is research into 3-D subsurface imaging with a composite information from the various geophysical data. In an attempt to interpret Schlumberger sounding data for the study area in 2-D and 3-D view, resistivity imaging was firstly performed and then pseudo-3-D resistivity volume was reconstructed by interpolating several 1-D resistivity plots. Electrical resistivity discontinuities such as fracture zone were successfully clarified in pseudo-3-D resistivity volume. The low resistivity zone mainly associated with fracture zone appears to develop down to granitic basement in the central part of the study area. Seismic velocity near the lineament is estimated to be approximately as small as 3,000 m/s, and weathering-layer for the southeastern part is interpreted to be deeper than for the northwestern part. Geophysical attributes such as electrical resistivity, seismic velocity, radioactivity for the Chojeong Area were analysed by utilizing a GIS software Arc/Info. The major fault boundaries and fracture zones were resolved through image enhancement of composite section (electrical resistivity and seismic refraction data) and were interpreted to develop in the southeastern part of the area, as characterized by low electrical resistivity and low seismic velocity. However, radioactivity attribute was found to be less sensitive to geological discontinuities, compared to resistivity and seismic velocity attributes.

• Journal of the Korean Vacuum Society
• /
• v.19 no.2
• /
• pp.128-133
• /
• 2010

ZnO:Al films were deposited by DC-pulsed magnetron sputtering using a two-step process involving the control of the oxygen pressure. The seed layers were prepared with various Ar to oxygen flow ratios and the bulk layers were deposited under pure Ar. As the oxygen pressure during the deposition of the seed layer increased, the crystallinity and degree of (002) texturing increased. The resistivity gradually decreased with increasing crystallinity from $4.7\times10^4\Omega{\cdot}cm$ (no seed) to $3.7\times10^4\Omega{\cdot}cm$ (Ar/$O_2$ = 9/1). The etched surface showed a crater-like structure and an abrupt morphology change appeared as the crystallinity was increased. The sample deposited at an Ar/$O_2$ flow ratio of 9/1 showed a very high haze value of 88% at 500 nm, which was explained by the large feature size of the craters, as shown in the AFM image.

• Journal of the Korean Vacuum Society
• /
• v.17 no.4
• /
• pp.311-316
• /
• 2008

Since the presence of the chemical impurities and defect at surfaces and interfaces greatly influence the properties of various semiconductor devices, an unambiguous chemical characterization of the metal and semiconductor surfaces become more important in the view of the miniaturization of the devices toward nano scale. Among the various conventional surface characterization tools, Electron-induced Auger Electron Spectroscopy (EAES), X-ray Photoelectron Spectroscopy (XPS) and Secondary Electron Ion Mass Spectroscopy (SIMS) are being used for the identification of the surface chemical impurities. Recently, a novel surface characterizaion technique, Positron-annihilation induced Auger Electron Spectroscopy (PAES) is introduced to provide a unique method for the analysis of the elemental composition of the top-most atomic layer. In PAES, monoenergetic positron of a few eV are implanted to the surface under study and these positrons become thermalized near the surface. A fraction of the thermalized positron trapped at the surface state annihilate with the neighboring core-level electrons, creating core-hole excitations, which initiate the Auger process with the emission of Auger electrons almost simultaneously with the emission of annihilating gamma-rays. The energy of electrons is generally determined by employing ExB energy selector, which shows a poor resolution of $6{\sim}10eV$. In this paper, time-of-flight system is employed to measure the electrons energy with an enhanced energy resolution. The experimental result is compared with simulation results in the case of both linear (with retarding tube) and reflected TOF systems.

• Restorative Dentistry and Endodontics
• /
• v.35 no.2
• /
• pp.134-142
• /
• 2010

The purpose of this study was to evaluate the effect of Er,Cr:YSGG laser irradiation with hypersensitivity mode on microtensile bond strength of composite resin. Twenty extracted permanent molars were randomly assigned to six groups, according to the irradiation of Er,Cr:YSGG laser, adhesive system (Optibond FL or Clearfil SE bond) and application time of etchant (15 sec or 20 sec). Then composite resin was build up on each conditioned surface. The restored teeth were stored in distilled water at room temperature for 24 h and twelve specimens for each group were prepared. All specimens were subjected to microtensile bond strength and the fracture modes were evaluated. Also, the prepared dentin surface and laser irradiated dentin surface were examined under SEM. The results were as follows: 1. The microtensile bond strength of laser irradiated group was lower than that of no laser irradiated group. 2. Regardless of laser irradiation, the microtensile bond strength of Optibond FL was higher than that of Clearfil SE bond. And the microtensile bond strength of 20 sec etching group was higher than that of 15 sec etching group when using Optibond FL. 3. The SEM image of laser irradiated dentin surface showed prominent peritubular dentin, opened dentinal tubules and no smear layer.

• Proceedings of the Korean Vacuum Society Conference
• /
• 2010.02a
• /
• pp.362-362
• /
• 2010

Today, vast attention has been paid to periodic arrays of nanostructures due to their potential for applications such as memory with huge storage density. Such application requires large-scale fabrication of well ordered nano-sized structures. One of the most widely used methods for the ordered nanostructures is lithography. This top-down process, however, has the limit to reduce size. Here the promising alternative is the self-organization of ordered nano-sized structures such as large scale 2d carbon-induced reconstructions on W(110). In the present study, we report on the first well-resolved atomic resolution STM studies of the well-known R($15{\times}3$) and R($15{\times}12$) carbon induced reconstruction of the W(110). From the atomic image of R($15{\times}3$) for different values of tunneling gap resistance, we can tell there are no missing atoms in unit cells of R($15{\times}3$) and some atomic displacements are substantial from the clean W(110), even though not all the imaged position of atoms correspond to tungsten, but may include those of carbon. We are considering two cases; First case is related to lattice deformation, or top layer of W(110) is deformed in the process of relief of strain caused by random inserting of carbon atoms possibly in the interstitial position. In the second case, R($15{\times}3$) unit cell results from a coincidence lattice between clean W(110) substrate and tungsten carbide overlayer which has rectangular atomic arrangement and giving R($15{\times}3$) coincidence lattice. beta-W2C showing rectangular unit cell should be a candidate. Further, we report on new reconstructions. Unlike the well-known R($15{\times}12$) consisting of two parts, two inner structures between two "Backbone" structures. The new reconstruction, which we found for the first time, contains more parts between the "Backbone"s. Sometimes we can observe the reconstruction consists of only inner parts without "Backbone" parts. Thus, the observed reconstruction can be built by constructing of two types of "Lego"-like block. Moreover, the rectangle shape of "Backbone" transform to parallelogram-like shape over time, the so-called wavy-R($15{\times}12$). Adsorption of hydrogen can be the reason for this transformation.