• Journal of the Korean Institute of Illuminating and Electrical Installation Engineers
• /
• v.19 no.6
• /
• pp.109-116
• /
• 2005

Recently, images/videos have been preferred as the communication media because of their information-implication and easy recognizability. But the amount of their data is so large that it has been the major research area to compress their data. This paper is related to optimization in th image data format, which can make a great effect in performance of data compression and is based on the wavelet transform and JPEG2000. This paper established a criterion to decide the data format to be used in wavelet transform, which is on the bases of the data errors in frequency transform and quantization. This criterion has been used to extract the optimal data format experimentally. The result were(1, 9) of 10-bit fixed-point format for filter coefficients and (9, 7) of 16-bit fixed-point data dormat for wavelet coefficients and their optimality was confirmed.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.12 no.9
• /
• pp.4317-4335
• /
• 2018

Semi-supervised multi-view latent space learning is gaining considerable popularity recently in many machine learning applications due to the high cost and difficulty to obtain the large amount of label information of data. Although some semi-supervised multi-view latent space learning methods have been presented, there is still much space for improvement: 1) How to learn latent discriminant intact feature representations by employing data of multiple views; 2) How to exploit the manifold structure of both labeled and unlabeled point in the learned latent intact space effectively. To address the above issues, we propose an approach called semi-supervised multi-view manifold discriminant intact space learning ($SM^2DIS$) for image classification in this paper. $SM^2DIS$ aims to seek a manifold discriminant intact space for data of different views by making use of both the discriminant information of labeled data and the manifold structure of both labeled and unlabeled data. Experimental results on MNIST, COIL-20, Multi-PIE, and Caltech-101 databases demonstrate the effectiveness and robustness of our proposed approach.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.26 no.9
• /
• pp.1330-1339
• /
• 2022

3D object detection generally aims to detect relatively large data such as automobiles, buses, persons, furniture, etc, so it is vulnerable to small object detection. In addition, in an environment with limited resources such as embedded devices, it is difficult to apply the model because of the huge amount of computation. In this paper, the accuracy of small object detection was improved by focusing on local features using only one layer, and the inference speed was improved through the proposed knowledge distillation method from large pre-trained network to small network and adaptive quantization method according to the parameter size. The proposed model was evaluated using SUN RGB-D Val and self-made apple tree data set. Finally, it achieved the accuracy performance of 62.04% at mAP@0.25 and 47.1% at mAP@0.5, and the inference speed was 120.5 scenes per sec, showing a fast real-time processing speed.

• The Journal of Korean Institute of Electromagnetic Engineering and Science
• /
• v.12 no.5
• /
• pp.787-794
• /
• 2001

In this paper, the development of large-signal model extractor for GaAs FET device through the Monolithic Microwave integrated Circuit(MMIC) is presented. The measurement program controlled by personal computer is developed for the processing of an amount of measured data, and the de-embedding algorithm is added to the program for voltage dropping as attached series resistance on measurement system. The small-signal model parameters are typically consisted of 7 elements that are considered as complexity of large-signal model and its the accuracy of the small-signal model is verified through comparing with measured data as varied bias point. The fitting function model, one of the empirical model, is used for quick simulation. In the process of large-signal model parameter extraction, one-dimensional optimization method is proposed and optimized parameters are extracted. This study can reduce the modeling and measuring time and can secure a suitable model for circuit.

• Proceedings of the SAREK Conference
• /
• 2009.06a
• /
• pp.1306-1311
• /
• 2009

In recent semiconductor manufacturing clean rooms, in order to improve clean room air quality, air washers are used to remove airborne gaseous contaminants such as $NH_3$, SOx and organic gases from the outdoor air introduced into clean room. Meanwhile, there is a large amount of exhaust air from a clean room. From an energy conservation point of view, heat recovery is therefore useful for reducing the outdoor air conditioning load for a clean room. Therefore it is desirable to recover heat from the exhaust air and use it to reheat the outdoor air. In the present study, numerical analysis and experiment was conducted to simulate the amount of energy reduction of exhaust air heat recovery type air washer system. The present numerical results showed good agreement with the results of the experimental data.

• Journal of Korea Multimedia Society
• /
• v.20 no.11
• /
• pp.1803-1810
• /
• 2017

A large amount of processing time is required if the process of detecting the touch position on the touch screen and displaying it on the display panel is performed only by software. In this paper, we propose a method to output information touched on the screen using H/W method in order to improve the response speed delay. In the FPGA module designed for the HDMI signal output to the display module, the touch information is input to the serial data signal including touch coordinate information, point size, and color information. Then the module render the image using HDMI signal input to the module and the touch information. This method has a pipeline structure so it has effect of reducing the delay time that occurs in outputting the touch information compared with the conventional software processing method.

• Smart Structures and Systems
• /
• v.32 no.2
• /
• pp.123-133
• /
• 2023

This paper proposes a thermography-based coating thickness estimation method for steel structures using model-agnostic meta-learning. In the proposed method, a halogen lamp generates heat energy on the coating surface of a steel structure, and the resulting heat responses are measured using an infrared (IR) camera. The measured heat responses are then analyzed using model-agnostic meta-learning to estimate the coating thickness, which is visualized throughout the inspection surface of the steel structure. Current coating thickness estimation methods rely on point measurement and their inspection area is limited to a single point, whereas the proposed method can inspect a larger area with higher accuracy. In contrast to previous ANN-based methods, which require a large amount of data for training and validation, the proposed method can estimate the coating thickness using only 10- pixel points for each material. In addition, the proposed model has broader applicability than previous methods, allowing it to be applied to various materials after meta-training. The performance of the proposed method was validated using laboratory-scale and field tests with different coating materials; the results demonstrated that the error of the proposed method was less than 5% when estimating coating thicknesses ranging from 40 to 500 ㎛.

• Journal of the Korean Society for Aeronautical & Space Sciences
• /
• v.49 no.3
• /
• pp.197-204
• /
• 2021

If a failure or an abnormal maneuver occurs during the flight test of a missile, the missile is deliberately self-destructed so as not to continue the flight. At this time, debris are produced and it is important to estimate the impact area in real-time whether it is out of the safety area. In this paper, we propose a method to estimate the debris dispersion area and falling time in real-time using a Fully-Connected Neural Network (FCNN). We applied the Unscented Transform (UT) to generate a large amount of training data. UT parameters were selected by comparing with Monte-Carlo (MC) simulation to secure reliability. Also, we analyzed the performance of the proposed method by comparing the estimation result of MC.

• Journal of the Korea Society of Computer and Information
• /
• v.18 no.12
• /
• pp.75-82
• /
• 2013

This paper proposes a $O(n^2)$ polynomial time algorithm to obtain optimal solution for Traveling Salesman problem that is a NP-complete because polynomial time algorithm has been not known yet. The biggest problem in a large-scale Traveling Salesman problem is the fact that the amount of data to be processed is $n{\times}n$, and thus as n increases, the data increases by multifold. Therefore, this paper proposes a methodology by which the data amount is first reduced to approximately n/2. Then, it seeks a bi-directional route at a random point. The proposed algorithm has proved to be successful in obtaining the optimal solutions with $O(n^2)$ time complexity when applied to TSP-1 with 26 European cities and TSP-2 with 46 cities of the USA. It could therefore be applied as a generalized algorithm for TSP.

• Magazine of the Korean Society of Agricultural Engineers
• /
• v.32 no.1
• /
• pp.55-71
• /
• 1990

The purpose of this study is to find out the basic data for irrigation plans of red pepper and radish during the growing period, such as total amount of evapotranspiration, coefficent of evapotranspiration at each growth stage, the peak stage of evapotranspiration, the maximum ten day evapotranspiration , optimum irrigation point, total readily available moisture and intervals of irrigation date. The plots of experiment were arranged with split plot design which were composed of two factors, irrigation point for main plot and soil texture for split plot, and three levels ; irrigation point with pH1.7-2.0, pF2.1-2.4 and pF2.5-2.8, at soil texture of sandy soil, sandy loam and silty clay for both red pepper and radish, with two replications. The results obtained are summarized as follows. 1.1/10 exceedance probability values of maximum total pan evaporation during growing period for red peppr and radish were shown as 663.6 mm and 251.8 mm. respectively, and those of maximum ten day pan evaporation for red pepper and radish, 67.1 mm and 46.9 mm, respectively. 2.The time that annual maximum of ten day pan evaporation can he occurred, exists at any stage between the middle of May and the late of August for red pepper, and at any stage between the late of August and the late September for radish. 3.The magnitude of evapotranspiration and its coefficient for red pepper was occurred large in order of pF1.7-2.0 pF2.1-2.4 and pF2.5~2.8 in aspect of irrigation point and the difference in the magnitude of evapotranspiration and of its coefficient between levels of irrigation point was difficult to be found out due to the relative increase in water consumption resulted from large flourishing growth at the irrigation point in lower water content for radish. In aspect of soil texture they were appeared large in order of sandy loam, silty clay and sandy soil for both red pepper and radish. 4.The magnitude of leaf area index was shown large in order of pF2.1-2.4, pF2.5-2.8, and pFl.7-2.0, for red pepper and of pF2.5-2.8, pF2.1-2.4, pFl.7-2.0 for radish in aspect of irrigation point, and large in order of sandy loam, silty clay, sandy soil for both red pepper and radish in aspect of soil texture 5.1/10 exceedance probability value of evapotranspiration and its coefficient during the growing period for red pepper were shown as 683.5 mm and 1.03, respectively, while those of radish, 250.3 mm and 0, 99. respectively. 6.The time that the maximum evapotranspiration of red pepper can be occurred is in the middle of August around the date of ninetieth to hundredth after transplanting, and the time for radish is presumed to be in the late of September, around the date of thirtieth to fourtieth after sowing. At that time, 1/10 exceedance probability value of ten day evapotranspiration and its coefficient for red pepper is assumed to be 81.8 mm and 1.22, respectively, while those of radish, 49, 7 mm and 1, 06, respectively. 7.Optimum irrigation point for red pepper on the basis of the yield of raw matter is assumed to be pFl.7-2.0 for sandy soil, pF2.5-2.8 for sandy loam, and pF2.1-2.4 for silty clay. while that for radish is appeared to be pF2.5-2.8 in any soil texture used. 8.The soil moisture extraction patterns of red pepper and radish have shown that maximum extraction rates exist at 7 cm deep layer at the beginning stage of growth in any soil texture and that extraction rates of 21 cm to 35 cm deep layer are increased as getting closer to the late stage of growth. And especially the extraction rates have shown tendency to be greatest at 21cm deep layer from the most flourishing stage of growth for red pepper and at the last stage of growth for radish. 9.The total readily available moisture on the basic of the optimum irrigation point become 3.77-8.66 mm for sandy soil, 28.39-34.67 mm for sandy loam and 18.40-25.70 mm for silty clay for red pepper of each soil texture used but that of radish that has shown the optimum irrigation point of pF2.5-2.8 in any soil texture used. 12.49-15.27 mm for sandy soil, 23.03-28.13 mm for sandy loam, and 22.56~27.57 mm for silty clay. 10.On the basis of each optimum irrigation point. the intervals of irrigation date at the growth stage of maximum consumptive use of red pepper become l.4 days for sandy soil, 3.8 days for sandy loam and 2.6 days for silty clay, while those of radish, about 7.2 days.