• Journal of the Institute of Convergence Signal Processing
• /
• v.22 no.3
• /
• pp.141-148
• /
• 2021

In this paper, we propose an algorithm of welding bead detection and evaluation using geodesic active contour algorithm and high pass filter with image processing technique. The algorithm uses histogram equalization and high pass filter as gaussian filter to improve contrast. The image processing techniques smoothens the welding beads reduce the noise on an image. Then, the algorithm detects the welding bead area by applying the geodesic active contour algorithm and morphological ooperation. It also applies the balloon force that either inflates in, or deflates out the evolving contour for a better segmentation. After that, we propose a method for determining the quality of welding bead using effective length and width of the detected bead. In the experiments, our algorithm achieved the highest recall, precision, F-measure and IOU as 0.9894, 0.9668, 0.9780, and 0.8957 respectively. We compared the proposed algorithm with the conventional algorithms to evaluate the performance of the proposed algorithm. The proposed algorithm achieved better performance compared to the conventional ones with a maximum computational time of 0.6 seconds for segmenting and evaluating one welding bead.

• Journal of Broadcast Engineering
• /
• v.27 no.3
• /
• pp.378-390
• /
• 2022

With the recent development of deep composite multiplication neural network learning, deep learning techniques applied to single-image super-resolution have shown good results, and the strong expression ability of deep networks has enabled complex nonlinear mapping between low-resolution and high-resolution images. However, there are limitations in applying it to real-time or low-power devices with increasing parameters and computational amounts due to excessive use of composite multiplication neural networks. This paper uses blocks that extract hierarchical characteristics little by little using information distillation and suggests the Recursive Distillation Super Resolution Network (RDSRN), a lightweight network that improves performance by making more accurate high frequency components through high frequency residual purification blocks. It was confirmed that the proposed network restores images of similar quality compared to RDN, restores images 3.5 times faster with about 32 times fewer parameters and about 10 times less computation, and produces 0.16 dB better performance with about 2.2 times less parameters and 1.8 times faster processing time than the existing lightweight network CARN.

• Journal of the Korea Computer Graphics Society
• /
• v.28 no.3
• /
• pp.55-65
• /
• 2022

Recently, deep learning-based automated systems for identifying and detecting landmarks have been proposed. In order to train such a deep learning-based model without overfitting, a large amount of image and labeling data is required. Conventionally, an experienced reader manually identifies and labels landmarks in a patient's image. However, such measurement is not only expensive, but also has poor reproducibility, so the need for an automated labeling method has been raised. In addition, in the X-ray image, since various human tissues on the path through which the photons pass are displayed, it is difficult to identify the landmark compared to a general natural image or a 3D image modality image. In this study, we propose a geometric data augmentation technique that enables the generation of a large amount of labeling data in X-ray images. In addition, the optimal attention mechanism for landmark detection was presented through the implementation and application of various attention techniques to improve the detection performance of 16 major landmarks in the skull. Finally, among the major cranial landmarks, markers that ensure stable detection are derived, and these markers are expected to have high clinical application potential.

• Land and Housing Review
• /
• v.14 no.3
• /
• pp.93-106
• /
• 2023

This study attempted to simulate changes in the thermal environment according to the type of apartment complex in Korea using CFD techniques and evaluate the thermal environment by type of apartment. First, apartment complex types in the 2000s and 2010s were referred from previous studies and four types of apartment complex were extracted from. Second, the layout of the apartment complex and temperature changes were analyzed by the direction of wind inflow. Third, a standardized model was created from each type using tower type, plate type, and mixed driving. Last, CFD simulations were performed by setting up the inflow of wind from a total of eight directions. The temperature was relatively low in the type consisting of only the tower type and the type of placing the tower type in the center of the complex, regardless of the direction of the wind. It was due to the good inflow of wind from these types to the inside of the complex. It can be interpreted because wind flows easily into the complex in these types. The findings showed that wind flow and resulting temperature distribution patterns differed depending on the building type and complex layout type, confirming the need for careful consideration of the complex layout in the early design stage. The results are expected to be used as basic data for creating a sustainable residential environment in the early design stage of apartment complexes in the future.

• Nutrition Research and Practice
• /
• v.17 no.4
• /
• pp.682-697
• /
• 2023

BACKGROUND/OBJECTIVES: Tibetan tea is a kind of dark tea, due to the inherent complexity of natural products, the chemical composition and beneficial effects of Tibetan tea are not fully understood. The objective of this study was to unravel the composition of Tibetan tea using knowledge-guided multilayer network (KGMN) techniques and explore its potential antioxidant and hypolipidemic mechanisms in mice. MATERIALS/METHODS: The C57BL/6J mice were continuously gavaged with Tibetan tea extract (T group), green tea extract (G group) and ddH2O (H group) for 15 days. The activity of total antioxidant capacity (T-AOC) and superoxide dismutase (SOD) in mice was detected. Transcriptome sequencing technology was used to investigate the molecular mechanisms underlying the antioxidant and lipid-lowering effects of Tibetan tea in mice. Furthermore, the expression levels of liver antioxidant and lipid metabolism related genes in various groups were detected by the real-time quantitative polymerase chain reaction (qPCR) method. RESULTS: The results showed that a total of 42 flavonoids are provisionally annotated in Tibetan tea using KGMN strategies. Tibetan tea significantly reduced body weight gain and increased T-AOC and SOD activities in mice compared with the H group. Based on the results of transcriptome and qPCR, it was confirmed that Tibetan tea could play a key role in antioxidant and lipid lowering by regulating oxidative stress and lipid metabolism related pathways such as insulin resistance, P53 signaling pathway, insulin signaling pathway, fatty acid elongation and fatty acid metabolism. CONCLUSIONS: This study was the first to use computational tools to deeply explore the composition of Tibetan tea and revealed its potential antioxidant and hypolipidemic mechanisms, and it provides new insights into the composition and bioactivity of Tibetan tea.

• The Sea:JOURNAL OF THE KOREAN SOCIETY OF OCEANOGRAPHY
• /
• v.28 no.4
• /
• pp.133-142
• /
• 2023

In this paper, we built a data-driven artificial intelligence model using RNN-LSTM (Recurrent Neural Networks-Long Short-Term Memory) to predict the Lorenz system, and examined the possibility of whether this model can replace chaotic dynamic models. We confirmed that the data-driven model reflects the chaotic nature of the Lorenz system, where a small error in the initial conditions produces fundamentally different results, and the system moves around two stable poles, repeating the transition process, the characteristic of "deterministic non-periodic flow", and simulates the bifurcation phenomenon. We also demonstrated the advantage of adjusting integration time intervals to reduce computational resources in data-driven models. Thus, we anticipate expanding the applicability of data-driven artificial intelligence models through future research on refining data-driven models and data assimilation techniques for data-driven models.

• The Journal of the Acoustical Society of Korea
• /
• v.42 no.6
• /
• pp.519-528
• /
• 2023

The aim of this study was to enhance the flow rate and noise performance of a centrifugal pump in dishwashers by designing an optimized impeller shape through numerical and experimental investigations. To evaluate the performance of the target centrifugal pump, experiment was conducted using a pump performance tester and noise experiment was carried out in a semi-anechoic chamber with microphones and a reflecting wall behind the dishwasher. Through the use of advanced computational fluid dynamics techniques, numerical simulations were performed to analyze the flow and aeroacoustics performance of our target centrifugal pump impeller. To achieve this, numerical simulations were carried out using the Reynolds-Average Navier-Stokes equations and Ffowcs-Willliams and Hawkings equations as governing equations. In order to ensure the validity of numerical methods, a thorough comparison of numerical results with experimental results. After having confirmed the reliability of the current numerical method of this study, the optimization of the target centrifugal pump impeller was conducted. An improvement in flow rate was confirmed numerically, and a manufactured proto-type of the optimized model was used for experimental investigation. Furthermore, it was observed that by applying the fan law, we could effectively reduce noise levels without reducing the flow rate.

• The Journal of the Acoustical Society of Korea
• /
• v.43 no.1
• /
• pp.103-111
• /
• 2024

Axial-flow fans are used to transport fluids in relatively low-pressure flow regimes, and a variety of design variables are employed. The tip geometry of an axial fan plays a dominant role in its flow and noise performance, and two of the most prominent flow phenomena are the tip vortex and the tip leakage vortex that occur at the tip of the blade. Various studies have been conducted to control these three-dimensional flow structures, and winglet geometries have been developed in the aircraft field to suppress wingtip vortices and increase efficiency. In this study, a numerical and experimental study was conducted to analyze the effect of winglet geometry applied to an axial fan blade for an air conditioner outdoor unit. The unsteady Reynolds-Averaged Navier-Stokes (RANS) equation and the FfocwsWilliams and Hawkings (FW-H) equation were numerically solved based on computational fluid dynamics techniques to analyze the three-dimensional flow structure and flow noise numerically, and the validity of the numerical method was verified by comparison with experimental results. The differences in the formation of tip vortex and tip leakage vortex depending on the winglet geometry were compared through a three-dimensional flow field, and the resulting aerodynamic performance was quantitatively compared. In addition, the effect of winglet geometry on flow noise was evaluated by numerically simulating noise based on the predicted flow field. A prototype of the target fan model was built, and flow and noise experiments were conducted to evaluate the actual performance quantitatively.

• The Transactions of the Korea Information Processing Society
• /
• v.13 no.7
• /
• pp.326-334
• /
• 2024

Extra-oral imaging techniques such as Panoramic X-rays (PXs) and Cone Beam Computed Tomography (CBCT) are the most preferred imaging modalities in dental clinics owing to its patient convenience during imaging as well as their ability to visualize entire teeth information. PXs are preferred for routine clinical treatments and CBCTs for complex surgeries and implant treatments. However, PXs are limited by the lack of third dimensional spatial information whereas CBCTs inflict high radiation exposure to patient. When a PX is already available, it is beneficial to reconstruct the 3D oral structure from the PX to avoid further expenses and radiation dose. In this paper, we propose 3DentAI - an U-Net based deep learning framework for 3D reconstruction of oral structure from a PX image. Our framework consists of three module - a reconstruction module based on attention U-Net for estimating depth from a PX image, a realignment module for aligning the predicted flattened volume to the shape of jaw using a predefined focal trough and ray data, and lastly a refinement module based on 3D U-Net for interpolating the missing information to obtain a smooth representation of oral cavity. Synthetic PXs obtained from CBCT by ray tracing and rendering were used to train the networks without the need of paired PX and CBCT datasets. Our method, trained and tested on a diverse datasets of 600 patients, achieved superior performance to GAN-based models even with low computational complexity.

• Korean Chemical Engineering Research
• /
• v.62 no.3
• /
• pp.274-280
• /
• 2024

Machine learning techniques utilizing neural networks have been employed in various fields such as disease gene discovery and diagnosis, drug development, and prediction of drug-induced liver injury. Disease features can be investigated by molecular information of DNA. In this study, we developed a neural network to predict the length of DNA and the number of DNA species in mixture solution which are representative molecular information of DNA. In order to address the time-consuming limitations of gel electrophoresis as conventional analysis, we analyzed the dynamic data of a microfluidic concentrating device. The dynamic data were reconstructed into a spatiotemporal map, which reduced the computational cost required for training and prediction. We employed a convolutional neural network to enhance the accuracy to analyze the spatiotemporal map. As a result, we successfully performed single DNA length prediction as single-variable regression, simultaneous prediction of multiple DNA lengths as multivariable regression, and prediction of the number of DNA species in mixture as binary classification. Additionally, based on the composition of training data, we proposed a solution to resolve the problem of prediction bias. By utilizing this study, it would be effectively performed that medical diagnosis using optical measurement such as liquid biopsy of cell-free DNA, cancer diagnosis, etc.