Sleep apnea refers to a condition in which a person does not breathe during sleep, and is a dangerous symptom that blocks oxygen supply in the body, causing various complications, and the elderly and infants can die if severe. In this paper, we present an algorithm that classifies sleep breathing by analyzing the intensity of breathing with images alone in preparation for the risk of sleep apnea. Only the chest of the person being measured is set to the Region of Interest (ROI) to determine the breathing strength by the differential image within the corresponding ROI area. The adult was selected as the target of the measurement and the breathing strength was measured accurately, and the difference in breathing intensity was also distinguished using depth information. Two videos of sleeping babies also show that even microscopic breathing motions smaller than adults can be detected, which is also expected to help prevent infant death syndrome (SIDS).

Recognition of characters printed or marked on the PCB components from images captured using cameras is an important task in PCB components inspection systems. Previous optical character recognition (OCR) of PCB components typically consists of two stages: character segmentation and classification of each segmented character. However, character segmentation often fails due to corrupted characters, low image contrast, etc. Thus, OCR without character segmentation is desirable and increasingly used via deep neural networks. Typical implementation based on deep neural nets without character segmentation includes convolutional neural network followed by recurrent neural network (RNN). However, one disadvantage of this approach is slow execution due to RNN layers. LPRNet is a segmentation-free character recognition network with excellent accuracy proved in license plate recognition. LPRNet uses a wide convolution instead of RNN, thus enabling fast inference. In this paper, LPRNet was adapted for recognizing characters printed on PCB components with fast execution and high accuracy. Initial training with synthetic images followed by fine-tuning on real text images yielded accurate recognition. This net can be further optimized on Intel CPU using OpenVINO tool kit. The optimized version of the network can be run in real-time faster than even GPU.

I performed a semi-analytical numerical analysis of the effects of core size and electric field intensity on the light emission wavelength of CdSe/ZnS quantum dots (QDs). The analysis used a quantum mechanical approach; I solved the Schrödinger equation describing the electron-hole pairs of QDs. The numerical solutions are described using a basis set composed of the eigenstates of the Schrödinger equation; they are thus equivalent to analytical solutions. This semi-analytical numerical method made it simple and reliable to evaluate the dependency of QD characteristics on the QD core size and electric field intensity. As the QD core diameter changed from 9.9 to 2.5 nm, the light emission wavelength of CdSe core-only QDs varied from 262.9 to 643.8 nm, and that of CdSe/ZnS core/shell QDs from 279.9 to 697.2 nm. On application of an electric field of 8 × 10⁵ V/cm, the emission wavelengths of green-emitting CdSe and CdSe/ZnS QDs increased by 7.7 and 3.8 nm, respectively. This semi-analytical numerical analysis will aid the choice of QD size and material, and promote the development of improved QD light-emitting devices.

This paper analyzes the distortion correction of the in the micro DMD(digital micro mirror device) projector system using 0.25 or less optical throwing distance ratio. The distortion of projected image occurs depending on the performance of the optical lens, the installation location of the projection system, and the tilt of the screen. This study analyzed the physical tilt values influencing of the distortion of projected image, removed the tilt distortion of throwing distance ratio optical lens, and adjusted the distortion image by the simulation of calibration displacements. The results of this study demonstrated within 5% TV distortion reference. Moreover, the correction method reduced the pin-distortion correction of projection system.

Thin-film encapsulation (TFE) technology is most effective in preventing water vapor and oxygen permeation in the organic light emitting diodes (OLED). Of those, a laminated structure of Al₂O₃ and MgO were applied to provide efficient barrier performance for increasing the stability of devices in air. Atomic layer deposition (ALD) method is known as the most promising technology for making the laminated Al₂O₃/MgO and is used to realize a thin film encapsulation technology in organic light-emitting diodes. Atomic layer deposited inorganic films have superior barrier performance and have advantages of excellent uniformity over large scales at relatively low deposition temperatures. In this study, the control system of the MgCP₂ precursor for the atomic layer deposition of MgO was established in order to deposit the MgO layer stably by the injection time of second level and the stable heating temperature. The deposition rate was obtained stably to be from 4 to 10 Å/cycle using the injection pulse times ranging from 3 to 12 sec and a substrate temperature ranging from 80 to 150 ℃.

VI probe, which is one of various in-situ plasma monitoring sensor, is frequently used for in-situ process monitoring in mass production environment. In this paper, we correlated the plasma etch results with VI probe data with the small amount of gas flow rate changes to propose usefulness of the VI probe in real-time process monitoring. Several different sized contact holes were employed for the etch experiment, and the etched profiles were measured by scanning electron microscope (SEM). Although the shape of etched hole did not show satisfactory relationship with VI probe data, the chamber status changed along the incremental/decremental modification of the amount of gas flow was successfully observed in terms of impedance monitoring.

We developed an inductively coupled plasma (ICP) etcher for GaN etching using a parallel plasma electrode source with a multifunctional chuck matched to it in order for the low power consumption and low process cost in comparison with the conventional ICP system with a helical-type plasma electrode source. The optimization process condition using it for the micro light-emitting diode (µ-LED) chip fabrication was established, which is an ICP RF power of 300 W, a chuck power of 200 W, a BCl₃/Cl₂ gas ratio of 3:2. Under this condition, the mesa structure with the etch depth over 1 ㎛ and the etch angle over 75° and also with no etching residue was obtained for the µ-LED chip. The developed ICP showed the improved values on the process pressure, the etch selectivity, the etch depth uniformity, the etch angle profile and the substrate temperature uniformity in comparison with the commercial ICP. The µ-LED chip fabricated using the developed ICP showed the similar or improved characteristics in the L-I-V measurements compared with the one fabricated using the conventional ICP method

Since the recent Corona 19, the importance of cloud computing is increasing, and at the same time, competition among clouds is intensifying. Cloud companies are competing for survival by promoting various management innovation methods for continuous growth and development amid a rapidly changing business environment. They are also increasingly interested in performance management in their operations and growth. In this paper, we propose Cloud BSC, an IT BSC-based performance measurement model for cloud enterprise performance management. The validity of the proposed model is verified through statistical analysis and causal analysis. Eventually, the proposed model is expected to be utilized as a management evaluation tool that can provide useful performance analysis information to cloud companies.

The adjustment of lens magnification can make the degree of precision in pattern measurement be improved, but several problems such as high cost, smaller field of view and stage error accumulation are followed. In this paper, a method for precisely measuring patterns is proposed based on gradient transition vector, in order to solve these problems. The performance of our method is evaluated using pattern images with several directions. Also, it is compared with previous methods based on edge and gray-level moment. It is judged that the proposed method outperforms consistent pattern width results, and so could be applied to automation processes for measurement and inspection of precise and complexed patterns in IT, BT industry products.

Using an inkjet printing process, we have investigated a droplet formation of poly(3,4-ethylenedioxythiophene): poly(styrenesulfonate) (PEDOT:PSS) near the orifice of a piezoelectric inkjet head. With an attempt to form the smallest droplet without any satellites, we have applied various waveforms such as the unipolar, bipolar, and M-shaped waveforms. It is found that the droplet velocity and volume vary depending sensitively on the waveform width and voltage. Of those, the M-shaped waveform is shown to provide the smallest droplet volume, followed by the bipolar and then unipolar waveforms. The droplet printed on a PET film roll by the M-shaped waveform has the diameter as small as 46.1 ㎛. It is likely that the second short unipolar in the M-shape waveform increases the droplet velocity gradient, rendering the droplet smaller.

The domestic semiconductor industry can produce various products that will satisfy customer needs by diversifying assembly parts and increasing compatibility between them. It is necessary to improve the production line as a method to reduce the work-in-process inventory (WIP) in the assembly line, the idle time of the worker, and the idle time of the process. The improvement of the production line is to balance the capabilities of each process as a whole, and to determine the timing of product input or the order of the work process so that the time required between each process is balanced. The purpose of this study is to find the optimal WIP and buffer size through SA (Simulated Annealing) that minimizes lead time while matching the number of two parts in a parallel assembly line with bottleneck process. The WIP level and buffer size obtained by the SA algorithm were applied to the CONWIP and DBR systems, which are the existing production systems, and the simulation was performed by applying them to the new hybrid production system. Here, the Hybrid method is a combination of CONWIP and DBR methods, and it is a production system created by setting new rules. As a result of the Simulation, the result values were derived based on three criteria: lead time, production volume, and work-in-process inventory. Finally, the effect of the hybrid production method was verified through comparative analysis of the result values.

According to the trend of increasingly sophisticated cyber threats, the need for technology research that can be applied to cyber security personnel management and training systems is constantly being raised not only overseas but also in Korea. Previously, the US and UK have already recognized the need and have been steadily conducting related research from the past. In the United States, by encouraging applications based on related research (NICE Cybersecurity Workforce Framework) and disclosing successful use cases to the outside, it is laying the groundwork for profiling cyber security experts. However in Korea, research on cyber security expert training and profiling is insufficient compared to other countries. Therefore, in this study, in order to create a system suitable for the domestic situation, research and analysis of cases in the United States and the United Kingdom were conducted over the past few years, and based on this, a prototype was produced for the study of profiling technology for domestic cyber security experts.

Digital decimation filters are used in various digital signal processing systems using ADCs, including digital communication systems and sensor network systems. When the sampling rate of digital data is reduced, aliasing occurs. So, an anti-aliasing filter is necessary to suppress aliasing before down-sampling the data. Since the anti-aliasing filter has to have a sharp transition band between the passband and the stopband, the order of the filter is very high. However, as the order of the filter increases, the complexity and area of the filter increase, and more power is consumed. Therefore, in this paper, we propose two types of decimation filters, focusing on reducing the area of the hardware. In both cases, the complexity of the circuit is reduced by applying the required down-sampling rate in two times instead of at once. In addition, CIC decimation filters without a multiplier are used as the decimation filter of the first stage. The second stage is implemented using a CIC filter and a down sampler with an anti-aliasing filter, respectively. It is designed with Verilog-HDL and its function and implementation are validated using ModelSim and Quartus, respectively.

This numerical study focuses on the analysis of fume particle dispersion characteristics over the surface of target workpiece in laser micro-hole machining process. The effects of oblique stagnation flow over fume generating machining point are examined by carrying out a series of three-dimensional random particle simulations along with probabilistic particle generation model and particle drag correlation of low Reynolds number. Present computational model of fume particle dispersion is found to be capable of assessing and quantifying the fume particle contamination in precision hole machining which may influenced by different types of air flow patterns and their flow intensity. The particle size dependence on dispersion distance of fume particles from laser machining point is significant and the effects of increasing flow oblique angle are shown quite differently when slot blowing or slot suction flows are applied in micro-hole machining.

In this paper, electric fields in electromagnetic reverberation chambers, which are used as a substitute facility for EM-free anechoic chambers, are analyzed. This paper focused on the 4-5 GHz band, which is expected to adversely affect equipment. To analyze the field uniformity inside the electromagnetic reverberation chamber, electric field strengths are sampled and finite-difference time-domain method was used for numerical analysis. Moreover, Quadratic residue diffuser was used to improve the characteristics of the electromagnetic reverberation chamber and the uniformity of the internal field strength. Standard deviation, tolerance characteristics, and partiality characteristics were compared while varying the aiming point of transmission antenna.

Cloud is emerging as a key factor in digital transformation. In particular, the industry is paying attention to cloud as a key factor in economic revitalization after COVID-19. In this economic environment, cloud companies can survive only through continuous innovation. In this study, we propose the innovation types of cloud companies using the meta model, which is a high-level conceptual model that generalizes various corporate activities. The proposed cloud company innovation type applies a conceptual model divided into two axes such as the corporate value chain and the business management layer. The contribution of this study is to provide a basis for research in a new academic field called corporate innovation in the cloud domain, and to provide decision-making tools for diagnosing innovation levels or exploring innovation directions for cloud companies.

The desire for a technology that can mechanically acquire 2D images starting with the manual method of drawing has been making possible a wide range of modern image-based technologies and applications over a period. Moreover, this trend of the utilization of image-related technology as well as image-based information is likely to continue. Naturally, as like other technology areas, the function that humans produce and utilize by using images needs to be automated by using computing-based technologies. Surprisingly, technology using images in the future will be able to discover knowledge that humans have never known before through the information-related process that enables new perception, far beyond the scope of use that human has used before. Regarding this trend, the manipulation and configuration of massively distributed image database system is strongly demanded. In this paper, we discuss image identifier production methods based on the utilization of the image hashing technique which especially puts emphasis over an entropy operator.

In this study, we developed the service that can efficiently making friends among college students in metaverse world. Metaverse technology has recently emerged as an important topic across the industry.' The development of virtual and augmented reality technologies, which have emerged as a new paradigm to drive the next generation of the Internet, is bringing us closer to the metaverse world. Metaverse is spreading around the gaming, entertainment, music, and content industries[1]. In particular, as non-face-to-face transitions have accelerated since the COVID-19 outbreak, lifestyles and industrial sites are rapidly changing beyond untacting to metaverseization, a three-dimensional virtual space. After discovering the needs of users through surveys and interviews, the research method added functions to the service that matched those needs. Users were pleased that they could make friends who matched their preferences and tastes, play like a game in a virtual world called metaverse, and customize their avatars to their liking. It was also very fresh to customize the goods so that they could be gifted and kept by themselves.

In this paper, a capacitive water level sensor for boilers was developed. In order to accurately monitor the water level in a high-temperature boiler that generates a lot of precipitates, the occurrence of precipitates on the surface of the water level sensor should be small, and a sensor capable of measuring even if the sensor surface is somewhat contaminated is required. The capacitive water level sensor has a structure in which one of the two electrodes is insulated with Teflon coating, and the stainless steel package of the water level sensor is brought into contact with the water tank so that the entire water tank becomes another electrode of the water level sensor. A C-V converter that converts the capacitance change of the capacitive water level sensor into a voltage change was developed and integrated with the water level sensor to minimize noise. The performance of the developed capacitive water level sensor was evaluated through measurement.

For smart mobile devices, the wireless communication module is one of the hardware modules that consume the most energy. If we can build a multi-channel multi-interface environment using heterogeneous communication modules and operate them dynamically, data transmission performance can be highly improved by increasing the parallelism. Also, because these heterogeneous modules have different data rates, transmission ranges, and power consumption, we can save energy by exploiting a power efficient and low speed wireless interface module to transmit/receive sporadic small data. In this paper, we propose a power efficient data transmission method using heterogeneous communication networks. We also compared the performance of our proposed scheme to a conventional scheme, and proved that our proposed scheme can save energy while guaranteeing reasonable data delivery time.

As artificial intelligence technologies, including deep learning, develop, these technologies are being introduced to code similarity analysis. In the traditional analysis method of calculating the graph edit distance (GED) after converting the source code into a control flow graph (CFG), there are studies that calculate the GED through a trained graph neural network (GNN) with the converted CFG, Methods for analyzing code similarity through CNN by imaging CFG are also being studied. In this paper, to determine which approach will be effective and efficient in researching code similarity analysis methods using artificial intelligence in the future, code similarity is measured through funcGNN, which measures code similarity using GNN, and Siamese Network, which is an image similarity analysis model. The accuracy was compared and analyzed. As a result of the analysis, the error rate (0.0458) of the Siamese network was bigger than that of the funcGNN (0.0362).

With the development of the industry, the use of drones in specific mission flight is being actively studied. These drones fly a specified path and perform repetitive tasks. if the drone system will detect objects in real time, the performance of these mission flight will increase. In this paper, we implement object detection system and mount GPU acceleration to maximize the efficiency of limited device resources with drone video using Tensorflow Lite which enables in-device inference from a mobile device and Mobile SDK of DJI, a drone manufacture. For performance comparison, the average processing time per frame was measured when object detection was performed using only the CPU and when object detection was performed using the CPU and GPU at the same time.

Recently, human attribute recognition (HAR) attracts a lot of attention due to its wide application in video surveillance systems. Recent deep-learning-based solutions for HAR require time-consuming training processes. In this paper, we propose a post-processing technique that utilizes the historical video frames to improve prediction results without invoking re-training or modifying existing deep-learning-based classifiers. Experiment results on a large-scale benchmark dataset show the effectiveness of our proposed method.

Scaffold is used to produce bio sensor. Scaffold is required high dimensional accuracy. 3D printer is used to manufacture scaffold. 3D printer can't detect defect during printing. Defect detection is very important in scaffold printing. Real-time defect detection is very necessary on industry. In this paper, we proposed the method for real-time scaffold defect detection. Real-time defect detection model is produced using CNN(Convolution Neural Network) algorithm. Performance of the proposed model has been verified through evaluation. Real-time defect detection system are manufactured on hardware. Experiments were conducted to detect scaffold defects in real-time. As result of verification, the defect detection system detected scaffold defect well in real-time.

Numbers of studies related to optimization of design of organic light emitting diodes(OLED) through machine learning are increasing. We propose the generative method of the image to assess the performance of the device combining with machine learning technique. Principle parameter regarding dark spot growth mechanism of the OLED can be the key factor to determine the long-time performance. Captured images from actual device and randomly generated images at specific time and initial pinhole state are fed into the deep neural network system. The simulation reinforced by the machine learning technique can predict the device parameters accurately and faster. Similarly, the inverse design using multiple layer perceptron(MLP) system can infer the initial degradation factors at manufacturing with given device parameter to feedback the design of manufacturing process.

In this paper, we studied the charging characteristics of high-capacity supercapacitor with high current for RF wireless charging system for smart phone charging. The dc output of the RF-DC receiver is connected to supercapacitor after which is connected to DC-DC converter for charging a smart phone. This configuration stably supplies voltage and current for charging it. Studies show that the higher charging current use, the rapidly shorter the charging time of supercapacitor is. The currents of 2A, 10A and 27A were used for charging supercapacitors. The charging time was measured for 3000F, 6000F, 12000F supercapacitors which is parallelly connected with 3000F supercapacitors.

This paper deals with the development of a deep-learning-based runway cleaning robot using an optical camera. A suitable model to realize real-time object detection was investigated, and the differences between the selected YOLOv3 and other deep learning models were analyzed. In order to check whether the proposed system is applicable to the actual runway, an experiment was conducted by making a prototype of the robot and a runway model. As a result, it was confirmed that the robot was well developed because the detection rate of FOD (Foreign Object Debris) and cracks was high, and the collection of foreign substances was carried out smoothly.

We introduce an arbitrary waveform generation method and its H/W implementation case based on Rademacher and Walsh function. According to the orthogonal and periodic features of Rademacher and Walsh function, simple calculations can generate arbitrary waves with affordable logics. We implemented an FPGA-based AWS using above two functions, and verified. HDL simulation shows the proposed idea can draw desired analog test waveforms very fast, and its H/W size is promising to Built-Out Self-Test(BOST) logics for AI ICs.

Cartesian robot is used in semiconductor manufacturing. Friction between steel band and mover wears the steel band. The emission of wear particle from steel band contaminates semiconductor equipment. At the manufacturing site, the steel band is replaced periodically to minimize the generation of wear particle. But this is not a good way to minimize the generation of wear particle, because it is hard to specify the moment of replacement. We suggested the methodology to minimize the generation of wear particle using TRIZ technique. Also we made prototype robot which the solution is applied, and the performance of the solution was verified through experiments. As a result of verification, it was confirmed that the solution significantly reduced the generation of wear particle compared to the standard way.

In this paper, a 20W class multi-output QR flyback converter for auxiliary power supply was designed to stabilize 4 output voltages, and the efficiency and load characteristics were compared and analyzed. It was checked if each output affects other output characteristics through experiment. As a result, the experimental circuit reached a high efficiency of 82.5% or more at a load power of over 20W, and the maximum power loss was 2.6W. Consequently, it was confirmed that all of 4 output voltages of the multi-output QR flyback converter constructed in this paper were stabilized within 0.5% in full-load range, and each output was independently controlled in an electrically isolated state.

To prevent accidents caused by changes in the surrounding environment or other factors, various protection facilities are installed at the photovoltaic connection module. The main causes of fire are sparks due to foreign substances inside the photovoltaic connection module through high temperature rise and dew condensation in the photovoltaic connection module, and fire due to heat from the power diode. The proposed method can predict the fire by measuring flame, carbon dioxide, carbon monoxide, temperature, humidity, input voltage, and current on the photovoltaic connection module, and when the fire conditions are reached, fire alarm and power off can be sent to managers and users in real time to prevent fire in advance.

The core-shell InGaN/GaN Multi Quantum Well-Nanowires (MQW-NWs) that were selectively grown on oxide templates with perfectly circular hole patterns were highly crystalline and were shaped as high-aspect-ratio pyramids with semi-polar facets, indicating hexagonal symmetry. The formation of the InGaN active layer was characterized at its various locations for two types of the substrates, one containing defect-free MQW-NWs with GQDs and the other containing MQW-NWs with defects by using HRTEM. The TEM of the defect-free NW showed a typical diode behavior, much larger than that of the NW with defects, resulting in stronger EL from the former device, which holds promise for the realization of high-performance nonpolar core-shell InGaN/GaN MQW-NW substrates. These results suggest that well-defined nonpolar InGaN/GaN MQW-NWs can be utilized for the realization of high-performance LEDs.

The HUD (Head-up Display) device for vehicles has gradually been advanced in connection with ADAS (Advanced Driver Assistant System) for the safety and the convenience of driving. In this paper, the major features (e.g. speed, RPM, etc.) of vehicles is received through the ECU and the route information is received through the navigating API, configurating the integrated GUI. And, the optical system is configured based on DLP (Digital Light Processing) to evaluate the visibility depending on the resolution change of the GUI. The IoT HUD system proposed in this paper has the scalability to flexibly add not only the ECU but also various cloud-based driving-related information.

This paper describes a dynamic spectroscopic ellipsometry based on dual-wavelength calibration. DSE provides ellipsometric parameters at rates above 20 Hz, but the interferometer's sensitivity to temperature makes it difficult for that proposed system to maintain stable 𝜟k over long periods of time. To solve this problem, we set up an additional path in the DSE to perform simulations of the polarization phase calibration method using dual wavelengths. Through simulation, we were able to eliminate most of the polarization phase error and maintain a stable 𝜟k in the long-term stability experiment for 10 hours. This is the result that the 𝜟k stability of the proposed system is improved tens of times compared to the existing system.

In general, digital signage is a next-generation smart media that provides various information and advertisement services to many people indoors or outdoors using the Internet. Recently, digital signage is rapidly spreading in such a small indoor environment, that is, in an area closely related to daily life, for example, inside an elevator. However, in this kind of indoor environment where the stay time of persons is extremely limited, it would be not easy for them to keep advertisements in the user memory for a long time. In the digital signage display installed in an indoor environment, it is possible to think about the possibility for a function such as expanding the screen to a user's smartphone, which is now widely spread, to contain, store, and use the transmitted content. In this paper, we propose a method to extend the display of digital signage contents to personal smart phones with interaction function in such a limited environment. In order to make the system operation, the proposed system was verified by confirming the result of dual screen implementation in a smart phone through the prototype implementation of a digital signage system in an embedded Linux environment.