• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.11 no.8
• /
• pp.3028-3035
• /
• 2010

This paper propose a genetic method based on probability map to detect region of the lips on a natural image with the faces. The method has many solutions in order to detect regions such as the lips instead of one optimal solution of existing methods. To do this, it represents a pair of spatial coordinates as a chromosome, and introduces genetic operations like conservation interval, the number of generations and non-overlapping selection. By using the probability map of the HS in HSV color space, it increases adaptability to similar color that is a property of genetic algorithm. In our experiments, the optimal value of the important parameter $\beta$ was analyzed, which was used as the condition of an ending function and affected performance of the proposed algorithm. Also the algorithm was analyzed on what performance it has when its mating methods are different. The results of the experiment showed that our algorithm could be flexibly adapted for detecting other ROIs.

• The Journal of the Acoustical Society of Korea
• /
• v.37 no.4
• /
• pp.256-261
• /
• 2018

There are three types of noise generated inside the vehicle: BSR (Buzz, Squeak, Rattle). In this paper, we propose a classifier that automatically classifies automotive BSR noise by using features extracted from deep convolutional neural networks. In the preprocessing process, the features of above three noises are represented as noise-map using STFT (Short-time Fourier Transform) algorithm. In order to cope with the problem that the position of the actual noise is unknown in the part of the generated noise map, the noise map is divided using the sliding window method. In this paper, internal parameter of the deep convolutional neural networks is visualized using the t-SNE (t-Stochastic Neighbor Embedding) algorithm, and the misclassified data is analyzed in a qualitative way. In order to analyze the classified data, the similarity of the noise type was quantified by SSIM (Structural Similarity Index) value, and it was found that the retractor tremble sound is most similar to the normal travel sound. The classifier of the proposed method compared with other classifiers of machine learning method recorded the highest classification accuracy (99.15 %).

• Journal of the Korean Society for Precision Engineering
• /
• v.23 no.8 s.185
• /
• pp.127-136
• /
• 2006

Metallic sandwich plates with inner dimpled shell subject to 3-point bending have been analyzed and then optimized for minimum weight. Inner dimpled shells can be easily fabricated by press or roll with high precision and bonded with same material skin sheets by resistance welding or adhesive bonding. Metallic sandwich plates with inner dimpled shell structure can be optimally designed for minimum weight subject to prescribed combination of bending and transverse shear loads. Fundamental findings for lightweight design are presented through constrained optimization. Failure responses of sandwich plates are predicted and formulated with an assumption of narrow sandwich beam theory. Failure is attributed to four kinds of mechanisms: face yielding, face buckling, dimple buckling and dimple collapse. Optimized shape of inner dimpled shell structure is a hemispherical shell to minimize weight without failure. It is demonstrated that bending stiffness of sandwich plate is 2 or 3 times larger than solid plates with the same strength. Failure mode boundaries and iso-strength lines dependent upon the geometry and yield strain of the material are plotted with respect to geometric parameters on the failure map. Because optimal parameters of maximum strength for given material weight can be selected from the map, analytic solutions for maximum strength are expressed as a function of only material property and proposed strength. These optimal parameters match well with numerical optimal parameters.

• Journal of the Korean Magnetics Society
• /
• v.26 no.2
• /
• pp.55-61
• /
• 2016

This study proposes the design algorithm of an electromagnetic linear actuator with a divided coil excitation system, such as the colenoid (COL) system, using the equivalent magnetic circuit (EMC) method. Nowadays, the clamping device is used to hold workpiece in the electrically driven chucking system and is needed to produce a huge clamping force of 40 kN like hydraulic system. The design algorithm for electromagnetic linear actuator can be obtained using the EMC method. At first, the parameter map is used to decide the slot width ratio in the initial design. Next, to make the magnetic flux density uniform at each pole, the pole width is adjusted by the pole width adjusting algorithm with EMC. When the dimensions of the electromagnetic linear actuator are decided, the clamping force is calculated to check the desired clamping force. The design results show that it can be used to hold a workpiece firmly instead of using a hydraulic cylinder in a chucking system.

• Water for future
• /
• v.28 no.5
• /
• pp.117-127
• /
• 1995

The application and analysis for the scale considering GIUH model proposed by the authors in this issue have been performed for the leemokjung sub-basin in the Pyungchang basin one of IHP representative basin in Korea. Scales of topographic maps for model application and fractal analysis are 1:25,000, 1:50,000 and 1:100,000. The ratio between successive scales is therefore constant. Link lengths were measured using a curvimeter with the resolution of 1 mm. Richardson's method was employed to have fractal dimension of streams. Apparent alternations of parameters were found in accordance with variations of map scale. And this tendency could mislead physical meanings of parameters because model parameters had to preserve their own value in spite of map scale change. It was found that uses of fractal transform and Melton's law could help to control the scale problem effectively. This methodlogy also could emphasize the relationship between network and basin to the model. To verify the applicability of GIUH proposed in this research, the model was compared with the exponential GIUH model. It is proven that proposed 2-parameter gamma GIUH model can better simulate the corresponding runoff from any given flood events than exponential GIUH model. The result showed that 2-parameter gamma GIUH model and fractal theory could be used for deriving scale considered IUH of the basin.

• Journal of the Korean Society of Hazard Mitigation
• /
• v.7 no.2 s.25
• /
• pp.53-63
• /
• 2007

This study is to evaluate the probable snowfall depth by the point frequency analysis and to draw the map of probable snowfall depth in Korea. The 14 probability distributions which has been widely used in hydrologic frequency analysis are applied to the annual maximum depth of snowfall data. The parameters of each probability distribution are estimated by method of moments, maximum likelihood method and method of probability weighted moments. The estimated parameters were checked by parameter validity conditions of each assumed probability distribution. Four tests that are $X^2-test$, Kolmogorov-Smirnov test, Cramer von Mises test and probability plot correlation coefficient test are used in this study to determine the goodness of fit of the distributions. Mostly the 2-parameter gamma distribution was determined as appropriate distribution for the annual maximum new snowfall depth. The probable snowfall depth were obtained from appropriate distribution for the selected return periods and the maps of probable snowfall depth were presented. It will be useful to specify the snowfall load for the design of agricultural facilities such as vinyl house and cattle shed.

• Korean Journal of Remote Sensing
• /
• v.34 no.4
• /
• pp.661-670
• /
• 2018

This study aims to develop a methodology of convolutional neural networks (CNNs) to produce thematic maps from remote sensing data. Optimizing the image size for CNNs was studied, since the size of the image affects to accuracy, working as hyper-parameter. The selected study area is Mt. Ung, located in Dangjin-si, Chungcheongnam-do, South Korea, consisting of both coniferous forest and deciduous forest. Spatial structure analysis and the classification of forest type using CNNs was carried in the study area at a diverse range of scales. As a result of the spatial structure analysis, it was found that the local variance (LV) was high, in the range of 7.65 m to 18.87 m, meaning that the size of objects in the image is likely to be with in this range. As a result of the classification, the image measuring 15.81 m, belonging to the range with highest LV values, had the highest classification accuracy of 85.09%. Also, there was a positive correlation between LV and the accuracy in the range under 15.81 m, which was judged to be the optimal image size. Therefore, the trial and error selection of the optimum image size could be minimized by choosing the result of the spatial structure analysis as the starting point. This study estimated the optimal image size for CNNs using spatial structure analysis and found that this can be used to promote the application of deep-learning in remote sensing.

• Journal of the Korean Geographical Society
• /
• v.44 no.4
• /
• pp.557-576
• /
• 2009

Physical based water balance model had better simulation results than conceptial model, however it is difficult to obtain input data for the model. This study suggests some methods to obtain parameter values of BROOK90 from meteorological data, soil map, land-use map. Comparing measured and simulated discharge proved the methods to be valid. For validation model($2001{\sim}2003$), comparing measured and simulated discharge a daily mean bias error, Nash-Sutcliffe's model efficiency coefficient, coefficient of determination equal to -0.517, 0.87 and 0.89 respectively. The results of this study would be helpful to the hydrological study using physical based hydrological model not only in super site but in other catchments.

• Journal of Korean Society for Geospatial Information Science
• /
• v.8 no.2 s.16
• /
• pp.11-22
• /
• 2000

It needs to extract the accurate topological characteristics and hydrological parameters of watershed in order to manage water resource efficiently. But, these data are processed yet by manual wok and simple operation in hydrologic fields. In this paper, we presented algorithm that could extract topological characteristics and hydrological parameters over watershed using GSIS and it gives the saving of data processing tin and the confidency of data. We presented coupling method between GSIS and hydrologic model by using extracted parameters into the input parameter of HEC-HMS hydrologic model. The extraction procedure of topological characteristics and hydrological parameters is as below. First, watershed and stream are extracted by DEM and curve unmber is extracted throughout the overlay of landuse map and soil map. Also, we extracted surface parameters like the length of the longest flow path and the slope of the longest flow path by Grid computation into watershed and stream. And we gave the method that could extract hydrologic parameters like Muskingum K and sub-basin lag tin by executing computation into surface parameters and average Sn curve number being extracted.

• Journal of Korea Water Resources Association
• /
• v.46 no.5
• /
• pp.439-447
• /
• 2013

The applicability of the parameter map of the Modified Bartlett-Lewis Rectangular Pulse (MBLRP) model for the Korean Peninsula was assessed from the perspective of flood prediction. The design rainfalls estimated from the MBLRP model were smaller than those from observed values by 5% to 40%, and the degree of underestimation of design rainfall increases with the increase of the recurrence interval of the design rainfall. The design floods at a virtual watershed estimated using the simulated rainfall time series based on MBLRP model were also smaller than those derived from the observed rainfall time series by 20% to 45%. The degree of underestimation of design flood increases with the increase of the recurrence interval of the design flood.