• The Journal of Korea Robotics Society
• /
• v.14 no.1
• /
• pp.8-13
• /
• 2019

Visual place recognition is widely researched area in robotics, as it is one of the elemental requirements for autonomous navigation, simultaneous localization and mapping for mobile robots. However, place recognition in changing environment is a challenging problem since a same place look different according to the time, weather, and seasons. This paper presents a feature extraction method using a deep convolutional auto-encoder to recognize places under severe appearance changes. Given database and query image sequences from different environments, the convolutional auto-encoder is trained to predict the images of the desired environment. The training process is performed by minimizing the loss function between the predicted image and the desired image. After finishing the training process, the encoding part of the structure transforms an input image to a low dimensional latent representation, and it can be used as a condition-invariant feature for recognizing places in changing environment. Experiments were conducted to prove the effective of the proposed method, and the results showed that our method outperformed than existing methods.

• International Journal of Advanced Culture Technology
• /
• v.11 no.4
• /
• pp.279-285
• /
• 2023

This study provides academic implications by considering trends of domestic research regarding therapy for Mental disorder schizophrenia and psychosocial. For the analysis of this study, text mining with the use of R program and social network analysis method have been used and 65 papers have been collected The result of this study is as follows. First, collected data were visualized through analysis of keywords by using word cloud method. Second, keywords such as intervention, schizophrenia, research, patients, program, effect, society, mind, ability, function were recorded with highest frequency resulted from keyword frequency analysis. Third, LDA (latent Dirichlet allocation) topic modeling result showed that classified into 3 keywords: patient, subjects, intervention of psychosocial, efficacy of interventions. Fourth, the social network analysis results derived connectivity, closeness centrality, betweennes centrality. In conclusion, this study presents significant results as it provided basic rehabilitation data for schizophrenia and psychosocial therapy through new research methods by analyzing with big data method by proposing the results through visualization from seeking research trends of schizophrenia and psychosocial therapy through text mining and social network analysis.

• Korean Journal of Food Science and Technology
• /
• v.20 no.6
• /
• pp.820-826
• /
• 1988

The freezing point$(t_f)$, latent heat of freezing$({\triangle}\;H_f)$ and kinetic constant of fleering$(k_f)$ were determined from DSC thermogram at cooling rate $-2.5^{\circ}C/min{\sim}-10.0^{\circ}C/min$. The freezing point of various starches was decreased with an increase in cooling rate, and that of whole starches were lower than defatted starches. Changes of the latent heat of freezing was not observed at above cooling rate $-2.5^{\circ}C/min$. The latent heat of freezing$({\triangle}\;H_f)$ could be deduced as a function of water content(W) as follows: ${\triangle}\;H_f=0.700W-13.048$, (Kcal/kg) $(35%{\leqq}W{\leqq}70%)\;{\triangle}\;H_f=1.569W-73.861,\;(Kcal/kg)\;(W{\geqq}70%$) In the water content range $35{\sim}90$(wt %), the activation energy of various starches in freezing process was determined $126{\sim}270$ Kcal/mol.

• KIPS Transactions on Software and Data Engineering
• /
• v.12 no.12
• /
• pp.519-524
• /
• 2023

In hashing-based image retrieval, the hash code of a manipulated image is different from the original image, making it difficult to search for the same image. This paper proposes and evaluates a self-supervised deephashing model that generates perceptual hash codes from feature information such as texture, shape, and color of images. The comparison models are autoencoder-based variational inference models, but the encoder is designed with a fully connected layer, convolutional neural network, and transformer modules. The proposed model is a variational inference model that includes a SimAM module of extracting geometric patterns and positional relationships within images. The SimAM module can learn latent vectors highlighting objects or local regions through an energy function using the activation values of neurons and surrounding neurons. The proposed method is a representation learning model that can generate low-dimensional latent vectors from high-dimensional input images, and the latent vectors are binarized into distinguishable hash code. From the experimental results on public datasets such as CIFAR-10, ImageNet, and NUS-WIDE, the proposed model is superior to the comparative model and analyzed to have equivalent performance to the supervised learning-based deephashing model. The proposed model can be used in application systems that require low-dimensional representation of images, such as image search or copyright image determination.

• Atmosphere
• /
• v.27 no.4
• /
• pp.455-466
• /
• 2017

In Korea, property and human damages occur annually due to heavy precipitation during the summer. On August 8, 2015, heavy rainfall occurred in the Seoul metropolitan area due to an outflow boundary, and $77mmhr^{-1}$ rainfall was recorded in Gwangju, Gyeonggi Province. In this study, the simulation of the WRF numerical model is performed to understand the cause and characteristics of heavy rainfall using the Conditional Instability of the Second Kind (CISK), potential vorticity (PV), frontogenesis function, and convective available potential energy (CAPE) analyses, etc. Convective cells initiated over the Shandong Peninsula and located on the downwind side of an upper level trough. Large amounts of water vapor were supplied to the Shandong Peninsula along the southwestern edge of a high pressure system, and from the remnants of typhoon Soudelor. The mesoscale convective system (MCS) developed through CISK process and moved over to the Yellow Sea. The outflow boundary from the MCS progressed east and pushed cold pool eastward. The warm and humid air over the Korean Peninsula further enhanced convective development. As a result, a new MCS developed rapidly over land. Because of the latent heat release due to convection and precipitation, strong potential vorticity was generated in the lower atmosphere. The rapid development of MCS and the heavy rainfall occurred in an area where the CAPE value was greater than $1300Jkg^{-1}$ and the fronto-genesis function value of 1.5 or greater coincided. The analysis result shows that the MCS driven by an outflow boundary can be identified using CISK process.

• Transactions of Materials Processing
• /
• v.19 no.1
• /
• pp.25-31
• /
• 2010

To achieve desired microstructure and mechanical property of a manufacturing product, heat treatment process is applied as a secondary process after forging. Especially, quenching process is used for improving strength, hardness, and wear resistance since phase transformation occurs owing to rapid heat transfer from the surface of the specimen. In the present paper, a study on surface temperature measurement for water quenching of eutectoid steel was investigated. In order to determine the temperature history in experiments, three different measuring schemes were used by varying installation techniques of K-type thermocouples. Depending on the measured temperature distribution at the surface of the specimen, convective heat transfer coefficients were numerically determined as a function of temperature by the inverse finite element analysis considering the latent heat generation due to phase transformation. Based on the inversely determined convective heat transfer coefficient, temperature, phase, and hardness distributions in the specimen after water quenching were numerically predicted. By comparing the experimental and computational hardness distribution at three different locations in the specimen, the best temperature measuring scheme was determined. This work clearly demonstrates the effect of temperature measurement on the final mechanical property in terms of hardness distribution.

• Laser Solutions
• /
• v.4 no.1
• /
• pp.21-28
• /
• 2001

Simulation on the electron beam welding of Al 2219 alloy was carried out by using commercial FEM code MARC, which encounters moving heat sources. Due to axisymmetry of geometry, a half of the cylinder was simulated. A coupled thermo-mechanical analysis was carried out and subroutine for heat flux was substituted in the program. The material properties such as specific heat, heat transfer coefficient and thermal expansion coefficient were given as a function of temperature and the latent heat associated with a given temperature range is considered. As a result, the proper beam power is 60㎸${\times}$60㎃ and welding speed is 1∼1.5 m/min. The residual stress in the heat-affected zone as well as the fusion zone does not increase. It is necessary to use jigs for preventing distortion of cylinder and improving weld quality.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.35 no.5
• /
• pp.423-430
• /
• 2017

Recently, data analysis research has been carried out using the deep learning technique in various fields such as image interpretation and/or classification. Various types of algorithms are being developed for many applications. In this paper, we propose a precipitation prediction algorithm based on deep learning with high accuracy in order to take care of the possible severe damage caused by climate change. Since the geographical and seasonal characteristics of Korea are clearly distinct, the meteorological factors have repetitive patterns in a time series. Since the LSTM (Long Short-Term Memory) is a powerful algorithm for consecutive data, it was used to predict precipitation in this study. For the numerical test, we calculated the PWV (Precipitable Water Vapor) based on the tropospheric delay of the GNSS (Global Navigation Satellite System) signals, and then applied the deep learning technique to the precipitation prediction. The GNSS data was processed by scientific software with the troposphere model of Saastamoinen and the Niell mapping function. The RMSE (Root Mean Squared Error) of the precipitation prediction based on LSTM performs better than that of ANN (Artificial Neural Network). By adding GNSS-based PWV as a feature, the over-fitting that is a latent problem of deep learning was prevented considerably as discussed in this study.

• Proceedings of the KSRS Conference
• /
• 2005.10a
• /
• pp.392-393
• /
• 2005

Due to the temporal and spatial simultaneity and the high-frequency repetition, the data set retrieved from the satellite observation is considered to be the most desirable ones for the study of air-sea interaction. With rapidly developing sensor technology, satellite-retrieved data has experienced improvement in the accuracy and the number of parameters. Nevertheless, since it is still impossible to directly measure the heat fluxes between air and sea, the bulk method is an exclusive way for the evaluation of the heat fluxes at the sea surface. It was noted that the large deviation of air temperature in the winter season by the linear regression despite good correlation coefficients. We propose a new algorithm based on the Fourier series with which the SST and the air temperature. We found that the mean of air temperature is a function of the mean of SST with the monthly gradient of SST inferred from the latitudinal variation of SST and the spectral energy of air temperature is related linearly to that of SST. An algorithm to obtain the air temperature over the sea was completed with a proper analysis on the relation between of air temperature and of SST. This algorithm was examined by buoy data and therefore the air temperature over the sea can be retrieved based on just satellite data.

• Proceedings of the KSRS Conference
• /
• 2001.03a
• /
• pp.104-110
• /
• 2001

Ocean heat fluxes over a wide region are generally estimated by an aerodynamic bulk fromula. Though a remote sensing technique can be expected to estimated global heat flux, it is difficult to obtain air temperature and specific humidity at sea surface by a remote sensor. In this study present a new method with which to determine near-sea surface air temperature from in situ data. Also, These methods compared with other methods. A new method used a linear regression equation between sea surface temperature and air temperature of the buoys data. In this study new method is validated using observed monthly mean data at the Japan Meteorological Agency(JMA), National Data Buoy Center(NDBC) and Tropical Ocean-Global Atmosphere(TOGA)-Tropical Atmosphere Ocean(TAO) buoys. The result that bias and rmse are 0.28, 1.5$0^{\circ}C$ respectively. The correlation coefficient is 0.98. Also, to retrieve near-sea surface specific humidity(Q) from good nonlinear regression relationship between vapor pressure(Ea) of buoy data and air temperature, after obtained the third-order polynomial function, compared with that of estimated from SSM/I empirical equation by Schussel et al(1995). The result that bias and rmse are -1.42 and 1.75(g/kg).