• 전자통신동향분석
• /
• 제37권3호
• /
• pp.64-73
• /
• 2022

As the COVID-19 pandemic has decreased face-to-face communication in everyday life, our interest in cultural communication via virtual world has grown significantly. In particular, the demand for applications that use three-dimensional (3D) data generation technology such as virtual reality, augmented reality, virtual performances, and realistic content is rapidly increasing in the entertainment and gaming industries. Additionally, improved computing capacity has increased the demand for high-resolution data. This study investigates the trends in 3D scanning and photogrammetry technologies that can support high-quality 3D data generation and introduces the high-resolution 3D data generation technology developed and reported in ETRI.

• 방송공학회논문지
• /
• 제25권2호
• /
• pp.275-278
• /
• 2020

본 논문에서는 가상 데이터 생성을 통한 딥러닝 기반 문자인식 시스템을 제안한다. 지도학습에서 가장 큰 비중을 차지하는 학습 데이터를 확보하기 위하여 가상 데이터를 생성하였다. 또한 가상 데이터를 생성 후 증강 파라미터를 이용하여, 실제 다양한 데이터에 대응하기 위해서 데이터 일반화를 하였다. 최종적으로 학습 데이터 구성은 증강 파라미터와 폰트 인자에 다양한 값을 대입하여 데이터를 생성하였다. 문자인식 성능을 측정하기 위한 테스트 데이터는 실제 촬영된 이미지 데이터에서 문자영역을 크롭하여 구성하였다. 테스트 데이터는 실제환경에서 발생할 수 있는 이미지 왜곡을 고려하여 데이터 증강하였다. 딥러닝 알고리즘은 실시간 검출에 용이한 YOLO v3를 사용하였으며, 추론결과는 후처리를 통하여 최종 검출결과를 출력한다.

• Journal of Information Processing Systems
• /
• 제20권1호
• /
• pp.116-130
• /
• 2024

In the Internet-of-Things (IoT) or blockchain-based network systems, secure keys may be stored in individual devices; thus, individual devices should protect data by performing secure operations on the data transmitted and received over networks. Typically, secure functions, such as a physical unclonable function (PUF) and fully homomorphic encryption (FHE), are useful for generating safe keys and distributing data in a network. However, to provide these functions in embedded devices for IoT or blockchain systems, proper inspection is required for designing and implementing embedded system-on-chip (SoC) modules through overhead and performance analysis. In this paper, a virtual platform (SoC VP) was developed that includes a secure key generation module with a PUF and FHE. The SoC VP platform was implemented using SystemC, which enables the execution and verification of various aspects of the secure key generation module at the electronic system level and analyzes the system-level execution time, memory footprint, and performance, such as randomness and uniqueness. We experimentally verified the secure key generation module, and estimated the execution of the PUF key and FHE encryption based on the unit time of each module.

• 한국세라믹학회지
• /
• 제56권6호
• /
• pp.596-600
• /
• 2019

Artificial intelligence with a sufficient amount of realistic big data in certain applications has been demonstrated to play an important role in designing new materials or in manufacturing high-quality products. To reduce cracks in ceramic products using machine learning, it is desirable to utilize big data in recently developed data-driven optimization schemes. However, there is insufficient big data for ceramic processes. Therefore, we developed a numerical algorithm to make "virtual" manufacturing data sets using indirect methods such as computer simulations and image processing. In this study, a numerical algorithm based on the random walk was demonstrated to generate images of cracks by adjusting the conditions of the random walk process such as the number of steps, changes in direction, and the number of cracks.

• 국제학술발표논문집
• /
• The 9th International Conference on Construction Engineering and Project Management
• /
• pp.893-903
• /
• 2022

Recently, computer vision (CV)-based safety monitoring (i.e., object detection) system has been widely researched in the construction industry. Sufficient and high-quality data collection is required to detect objects accurately. Such data collection is significant for detecting small objects or images from different camera angles. Although several previous studies proposed novel data augmentation and synthetic data generation approaches, it is still not thoroughly addressed (i.e., limited accuracy) in the dynamic construction work environment. In this study, we proposed a game engine-driven synthetic data generation model to enhance the accuracy of the CV-based object detection model, mainly targeting small objects. In the virtual 3D environment, we generated synthetic data to complement training images by altering the virtual camera angles. The main contribution of this paper is to confirm whether synthetic data generated in the game engine can improve the accuracy of the CV-based object detection model.

• 한국자동차공학회논문집
• /
• 제11권5호
• /
• pp.201-209
• /
• 2003

Virtual human models are widely used to save time and expense in vehicle safety studies. A human model is an essential tool to visualize and simulate a vehicle driver in virtual environments. This research is focused on creation and application of a human model fer virtual reality. The Korean anthropometric data published are selected to determine basic human model dimensions. These data are applied to GEBOD, a human body data generation program, which computes the body segment geometry, mass properties, joints locations and mechanical properties. The human model was constituted using MADYMO based on data from GEBOD. Frontal crash and bump passing test were simulated and the driver's motion data calculated were transmitted into the virtual environment. The human model was organized into scene graphs and its motion was visualized by virtual reality techniques including OpenGL Performer. The human model can be controlled by an arm master to test driver's behavior in the virtual environment.

• 대한치과기공학회지
• /
• 제46권2호
• /
• pp.36-41
• /
• 2024

Purpose: This study aimed to generate virtual mandibular left first molar teeth using deep convolutional generative adversarial networks (DCGANs) and analyze their matching accuracy with actual tooth morphology to propose a new paradigm for using medical data. Methods: Occlusal surface images of the mandibular left first molar scanned using a dental model scanner were analyzed using DCGANs. Overall, 100 training sets comprising 50 original and 50 background-removed images were created, thus generating 1,000 virtual teeth. These virtual teeth were classified based on the number of cusps and occlusal surface ratio, and subsequently, were analyzed for consistency by expert dental technicians over three rounds of examination. Statistical analysis was conducted using IBM SPSS Statistics ver. 23.0 (IBM), including intraclass correlation coefficient for intrarater reliability, one-way ANOVA, and Tukey's post-hoc analysis. Results: Virtual mandibular left first molars exhibited high consistency in the occlusal surface ratio but varied in other criteria. Moreover, consistency was the highest in the occlusal buccal lingual criteria at 91.9%, whereas discrepancies were observed most in the occusal buccal cusp criteria at 85.5%. Significant differences were observed among all groups (p<0.05). Conclusion: Based on the classification of the virtually generated left mandibular first molar according to several criteria, DCGANs can generate virtual data highly similar to real data. Thus, subsequent research in the dental field, including the development of improved neural network structures, is necessary.

• 소성∙가공
• /
• 제12권6호
• /
• pp.550-557
• /
• 2003

In this study, a web-based system was developed by utilizing finite element method and virtual system designed using Virtual Reality Modeling Language (VRML). The simulation program for axi-symetric sheet forming is developed using finite flement method. The developed system consists of two modules, client module and server module. The client module was developed by using Active-X control. The input data for FEM calculation is transferred to the server module by using communication protocol. Then sever module performs several successive processes: input data generation, forming simulation, conversion of results to VRML format. After that, the results from the simulation can be visualized on the web browser in client computer. Besides, client module offers the capability to control and navigate on virtual forming machine and calculated result. By using this system simulation result can be investigated more realistically in virtual environment including forming machine.

• Wind and Structures
• /
• 제28권5호
• /
• pp.271-284
• /
• 2019

This paper describes the application of a novel virtual prototyping methodology to wind turbine blade design. Numeric modelling data and experimental data about turbine blade geometry and structural/dynamical behaviour are combined to obtain an affordable digital twin model useful in reducing the undesirable uncertainties during the entire turbine lifecycle. Moreover, this model can be used to track and predict blade structural changes, due for example to structural damage, and to assess its remaining life. A new interactive and recursive process is proposed. It includes CAD geometry generation and finite element analyses, combined with experimental data gathered from the structural testing of a new generation wind turbine blade. The goal of the research is to show how the unique features of a complex wind turbine blade are considered in the virtual model updating process, fully exploiting the computational capabilities available to the designer in modern engineering. A composite Sandia National Laboratories Blade System Design Study (BSDS) turbine blade is used to exemplify the proposed process. Static, modal and fatigue experimental testing are conducted at Clarkson University Blade Test Facility. A digital model was created and updated to conform to all the information available from experimental testing. When an updated virtual digital model is available the performance of the blade during operation can be assessed with higher confidence.

• Journal of Advanced Marine Engineering and Technology
• /
• 제28권8호
• /
• pp.1185-1194
• /
• 2004

Virtual images for PIV are produced for the construction of a benchmark test tool of PTV systems, Camera parameters obtained by an actual experiment are used to construct the virtual images, LES(Large Eddy Simulation) data sets of a channel flow are used for generation of the virtual images, Using the virtual images and the camera's parameters. three-dimensional velocity vectors are obtained for a channel flow. The capabilities of a 3D-PTV algorithm are investigated by comparing the results obtained by the virtual images and those by an actual measurement for the channel flow.