• Journal of the Korea Society of Computer and Information
• /
• v.25 no.10
• /
• pp.15-22
• /
• 2020

In this study, we propose an improvement method that can create U-Net model which detect fine concrete cracks by applying a weighted loss function. Because cracks in concrete are a factor that threatens safety, it is important to periodically check the condition and take prompt initial measures. However, currently, the visual inspection is mainly used in which the inspector directly inspects and evaluates with naked eyes. This has limitations not only in terms of accuracy, but also in terms of cost, time and safety. Accordingly, technologies using deep learning is being researched so that minute cracks generated in concrete structures can be detected quickly and accurately. As a result of attempting crack detection using U-Net in this study, it was confirmed that it could not detect minute cracks. Accordingly, as a result of verifying the performance of the model trained by applying the suggested weighted loss function, a highly reliable value (Accuracy) of 99% or higher and a harmonic average (F1_Score) of 89% to 92% was derived. The performance of the learning improvement plan was verified through the results of accurately and clearly detecting cracks.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.25 no.4
• /
• pp.377-385
• /
• 2015

This paper proposes a novel forward vehicle tracking algorithm based on the WMIL(Weighted Multiple Instance Learning) equipped with a particle filter. In the proposed algorithm Haar-like features are used to train a vehicle object detector to be tracked and the location of the object are obtained from the recognition result. In order to combine both the WMIL to construct the vehicle detector and the particle filter, the proposed algorithm updates the object location by executing the propagation, observation, estimation, and selection processes involved in particle filter instead of finding the credence map in the search area for every frame. The proposed algorithm inevitably increases the computation time because of the particle filter, but the tracking accuracy was highly improved compared to Ababoost, MIL(Multiple Instance Learning) and MIL-based ones so that the position error was 4.5 pixels in average for the videos of national high-way, express high-way, tunnel and urban paved road scene.

• Journal of Korean Society for Geospatial Information Science
• /
• v.20 no.1
• /
• pp.93-99
• /
• 2012

This study aims to identify spatio-temporal land price change pattern in Gongju city including the area incorporated and surrounding area depending on the Multifunctional Administrative City Construction. For this, GIS data was built by calculating the average land price each 209 Dong and Ri by the time of the year 2000, 2005 and 2010 based on. The first, the change in the land price was to identify in the 5-year intervals through a kriging interpolation as a kind of geostatistical techniques. The second, a trend analysis was conducted to know directional change pattern of the east-west axis and the north-south axis. Finally, the weighted mean center was calculated by the land price at a weight to examine moving direction on the center point of land price, point of view. The result is that the land price change pattern appeared visible higher growth on the eastern built in the Multifunctional Administrative City, moving direction on the center point of the land price appeared that the phenomenon was concentrated in the northeastern area.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.13 no.1
• /
• pp.333-342
• /
• 2012

This paper presents a vowel onset point (VOP) detection method based on the human auditory system. This method maps the "perceptual" frequency scale, i.e. Mel scale onto a linear acoustic frequency, and then establishes a series of Triangular Mel-weighted Filter Bank simulate the function of band pass filtering in human ear. This nonlinear critical-band filter bank helps greatly reduce the data dimensionality, and eliminate the effect of harmonic waves to make the formants more prominent in the nonlinear spaced Mel spectrum. The sum of mel spectrum peaks energy is extracted as feature for each frame, and the instinct at which the energy amplitude starts rising sharply is detected as VOP, by convolving with Gabor window. For the single-word database which contains 12 vowels articulated with different kinds of consonants, the experimental results showed a good average detection rate of 72.73%, higher than other vowel detection methods based on short-time energy and zero-crossing rate.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.9 no.6
• /
• pp.335-344
• /
• 2014

As a kind of 2D spatial coordinate transform, image warping is a basic image processing technique utilized in various applications. Though image warping algorithm is composed of relatively simple operations such as memory accesses and computations of weighted average, real-time implementations on embedded vision systems suffer from limited computational power because the simple operations are iterated as many times as the number of pixels. This paper presents a real-time implementation of a look-up table(LUT)-based image warping using an FPGA. In order to ensure sufficient data transfer rate from memories storing mapping LUT and image data, appropriate memory devices are selected by analyzing memory access patterns in an LUT-based image warping using backward mapping. In addition, hardware structure of a parallel and pipelined architecture is proposed for fast computation of bilinear interpolation using fixed-point operations. Accuracy of the implemented hardware is verified using a synthesized test image, and an application to real-time lens distortion correction is exemplified.

• The Journal of Industrial Distribution & Business
• /
• v.8 no.6
• /
• pp.87-95
• /
• 2017

Purpose - The structural changes of Korean agriculture are complex due to heterogeneous production processes and farms' features. This study analyzed trends of dualism in Korean agriculture over the period 2000-15 based on farm-level data to clarify the specific trends of dualism in terms of farm income, farm-size, and farm operators' age. From the results of this study, we would be able to understand the features of structural changes in Korean agriculture more profoundly. Research design, data, and methodology - We incorporated farm-level data in South Korea: Agricultural census and Farm household economy survey. As measures of inequality, we used size-weighted quantiles, and normalized Gini coefficients as well as mean and conventional quantiles. The size-weighted quantiles are more robust to changes in the number of small farms, but they are more sensitive to changes in the distribution of farm-size. Thus, they would be more useful to identify trends of dualism of Korean agriculture. Results - The results show that the farmland distribution of crop farms became more skewed and dispersed. However, the herd distribution of livestock farms became more concentrated. To be specific, their mean and 1st quantile increases more rapidly than their size-weighted 2nd quantile and size-weighted 3rd quantile. Gini coefficients of livestock farms regarding their herd distribution decreased by 0.1 on average. In the case of income distribution, the results indicate that the polarization regarding farm household/agricultural/non-agricultural income became more severe. However, we also found that the distribution of transfer income became concentrated continuously. The results imply that transfer income including subsidies would decrease farm income polarization. Lastly, during the study periods, Korean farms were aging over time, and age distribution of them more concentrated. Conclusions - The structure of Korean agriculture has been changing, even though the absolute size of it decreased over time. Land (herd) distribution became more dispersed (concentrated). Inequality regarding agricultural income became more severe, and it made farm household income more polarized even though transfer income would decrease income gaps among farms. Lastly, farms continue to age regardless of farm types and this might affect the structural changes in Korean agriculture in the future.

• Journal of Magnetics
• /
• v.20 no.4
• /
• pp.381-386
• /
• 2015

The purpose of this study is to improve diagnostic efficiency of clinical study by setting up guidelines for more precise examination with a comparative analysis of signal intensity and image distortion depending on the location of X axial of object when performing magnetic resonance diffusion weighted imaging (MR DWI) examination. We arranged the self-produced phantom with a 45 mm of interval from the core of 44 regent bottles that have a 16 mm of external diameter and 55 mm of height, and were placed in 4 rows and 11 columns in an acrylic box. We also filled up water and margarine to portrait the fat. We used 3T Skyra and 18 Channel Body array coil. We also obtained the coronal image with the direction of RL (right to left) by using scan slice thinkness 3 mm, slice gap: 0mm, field of view (FOV): $450{\times}450mm^2$, repetition time (TR): 5000 ms, echo time (TE): 73/118 ms, Matrix: $126{\times}126$, slice number: 15, scan time: 9 min 45sec, number of excitations (NEX): 3, phase encoding as a diffusion-weighted imaging parameter. In order to scan, we set b-value to $0s/mm^2$, $400s/mm^2$, and $1,400s/mm^2$, and obtained T2 fat saturation image. Then we did a comparative analysis on the differences between image distortion and signal intensity depending on the location of X axial based on iso-center of patient's table. We used "Image J" as a comparative analysis programme, and used SPSS v18.0 as a statistic programme. There was not much difference between image distortion and signal intensity on fat and water from T2 fat saturation image. But, the average value depends on the location of X axial was statistically significant (p < 0.05). From DWI image, when b-value was 0 and 400, there was no significant difference up to $2^{nd}$ columns right to left from the core of patient's table, however, there was a decline in signal intensity and image distortion from the $3^{rd}$ columns and they started to decrease rapidly at the $4^{th}$ columns. When b-value was 1,400, there was not much difference between the $1^{st}$ row right to left from the core of patient's table, however, image distortion started to appear from the $2^{nd}$ columns with no change in signal intensity, the signal was getting decreased from the $3^{rd}$ columns, and both signal intensity and image distortion started to get decreased rapidly. At this moment, the reagent bottles from outside out of 11 reagent bottles were not verified from the image, and only 9 reagent bottles were verified. However, it was not possible to verify anything from the $5^{th}$ columns. But, the average value depends on the location of X axial was statistically significant. On T2 FS image, there was a significant decline in image distortion and signal intensity over 180mm from the core of patient's table. On diffusion-weighted image, there was a significant decline in image distortion and signal intensity over 90 mm, and they became unverifiable over 180 mm. Therefore, we should make an image that has a diagnostic value from examinations that are hard to locate patient's position.

• Journal of Korean Society of Occupational and Environmental Hygiene
• /
• v.9 no.1
• /
• pp.87-98
• /
• 1999

The objectives of this study were to evaluate exposures to airborne methylene chloride and postshift carboxyhemoglobin (COHb) in blood of workers engaged in processes using blowing or cleaning agents, and to investigate correlation between methylene chloride concentrations and the blood COHb levels of workers. The geometric mean (GM) of workers' exposures (8 hour-time weighted averages, TWA) to airborne methylene chloride during cleaning molds using rags wetted with the solvent in the manufacture of flexible polyurethane foam (GM = 61.4 ppm), during operating the dip tank for cleaning molds in the manufacture of lens (GM = 61.0 ppm), and during cleaning the blowing nozzles by spraying the solvent in the manufacture of shoes (GM = 117.2 ppm) were exceeded the American Conference of Governmental Industrial Hygienists (ACGIH) Threshold Limit Value ($TLV^{(R)}$)-Time Weighted Average (TWA) (50 ppm). The COHb levels were significantly different among groups (p<0.05). The average COHh levels in blood of non-smoking workers were 2.0% in. low-level (<50 ppm) exposure group, and 3.9 % in high-level (>50ppm) exposure group. The average COHb levels in smoking workers were 3.1% in low-level exposure group, and 4.8% in high-level exposure group. The blood COHb levels of no-exposed workers to methylene chloride were 1.8% in non-smoking group, and 2.8% in smoking group. It was found that the COHb level dependeds on the methylene chloride concentration and smoking habit, and was highly correlated with methylene chloride concentration in air. The correlation coefficient was 0.81 among non-smoking workers. The estimated COHb level (3.6%) and 95% upper confidence limit (4.0%) corresponding to TLV-TWA of methylene chloride exceeded the current ACGIH Biological Exposure Index (COHb 3.5%) for carbon monoxide. The estimated COHb level (5.4%) at 100 ppm exceeded the standard (5%) recommended by National Institute for Occupational Safety and Health (NIOSH) for preventing adverse cardiovascular effect. The estimated COHb value and 95% upper confidence limit at 25 ppm of the Occupational Safety and Health. Adminstration (OSHA) Permissible Exposure Limit-TWA (PEL-TWA) were 2.6% and 3.0%, respectively. It is suggested that COHb in blood be kept below 3.0% to comply with OSHA PEL-TWA.

• The Journal of KAIRB
• /
• v.5 no.1
• /
• pp.14-20
• /
• 2023

Purpose: To obtain fundamental data on selection tools for an internal audit and develop a new guideline. We scored the indicated points from the internal audit, identified the research progress and problems that occurred, and confirmed the validity of the risk factors involved. Methods: Of the 63 internal audits conducted by Keimyung University Dongsan Hospital from 2014 to 2021, we analyzed 55 clinical trials with an inspection checklist. We excluded 8 that failed to transfer data and refused to comply with the internal audit. The statistical summary of the collected data was verified and interpreted by using frequency analysis and a chi-square test. Result: Of total 55 cases included in the internal audit, sponsor-initiated trial (SIT) was 63.6% (vs. investigator-initiated trial [IIT]), clinical trial for investigational drug was 71.0% (vs. nonclinical or clinical trial for investigational device), domestic multicenter trial was 60.0% (vs. single center or multinational multicenter trial), and trial requisition for MFDS approval was 69.1% (vs. exception for MFDS approval). The 10 areas of the clinical trial inspection checklist (reports, protection of subjects, compliance with protocols, records, management of investigational drug and/or device, delegation of duties, qualification of investigators, management of specimen, contract-agreement and approval of protocols, and preservation of recorded documents) were weighted between 2 to 5 points. The average of the total points was 16.09±13.2 and 20 clinical trials were above the average. As a result of comparing the average of the total points weighted by year, the highest score was in 2020. The 4 factors that play significant roles in determining the internal quality were (1) principal subjects that initiated the clinical trials (p=0.049), (2) type (p=0.003), (3) phase of clinical trials (p=0.024), and (4) number of registered subjects reported at the time of continuing deliberation (p=0.019). Of the 10 areas of the clinical trial inspection checklist, 'record' was the most inappropriate and insufficient. We found more indicated points; the quality of performance declined in IIT, nonclinical trials, and other clinical trials that were not in phase I1-IV4, and the study of more than 30 registered subjects at the time of continuing review. Conclusion: If an institution has an internal audit selection tool that reflects the aforementioned risk factors, it will be possible to effectively manage high-risk studies; thereby, contributing to an efficient internal audit and improving the quality of clinical trials.

• Management Science and Financial Engineering
• /
• v.11 no.3
• /
• pp.79-93
• /
• 2005

Despite the fact that the sourcing process in B2B e-Marketplaces is one of the most important tasks, the evaluation and selection process of suppliers have been ad-hoc based and mainly dependent on the experience of sourcing managers' subjective knowledge. To remedy the problem, we developed a decision support System (called Wise - I) that helps sourcing managers evaluate suppliers in a more systematic way. The system reflects company's strategy and know-how by adopting company enforced weighted scores for different factors and employing a more scientific method of considering factors other than price and on-time delivery rate, utilizing the AHP method. This paper reports the effectiveness of the system as well as the detailed description of the system. To investigate the effectiveness of the system, we collected information through interview and questionnaire survey. The information was also augmented through the firm key index system, which monitors average delivery lead time and on-time delivery rate. The result indicates that the system leads to the efficiency of purchasing section and the transparency of buying process, therefore reduces delivery time and cost.