• Current Optics and Photonics
• /
• 제2권1호
• /
• pp.69-78
• /
• 2018

This paper presents a precise edge detection algorithm for the critical dimension (CD) measurement of a Thin-Film Transistor Liquid-Crystal Display (TFT-LCD) pattern. The sigmoid surface function is proposed to model the blurred step edge. This model can simultaneously find the position and geometry of the edge precisely. The nonlinear least squares fitting method (Levenberg-Marquardt method) is used to model the image intensity distribution into the proposed sigmoid blurred edge model. The suggested algorithm is verified by comparing the CD measurement repeatability from high-magnified blurry and noisy TFT-LCD images with those from the previous Laplacian of Gaussian (LoG) based sub-pixel edge detection algorithm and error function fitting method. The proposed fitting-based edge detection algorithm produces more precise results than the previous method. The suggested algorithm can be applied to in-line precision CD measurement for high-resolution display devices.

• 한국연소학회지
• /
• 제5권2호
• /
• pp.19-27
• /
• 2000

A turbulent combustion model, based on edge flame dynamics, is discussed in order to predict global extinction of turbulent flames. The model is applicable to the broken flamelet regime of turbulent combustion, in which global extinction of turbulent flame is achieved by gradual expansion of flame holes. The edge flame dynamics is the key mechanism to describe the flame hole expansion or contraction. For flames with Lewis numbers near unity, there is a $Damk{\ddot{o}}hler$ number, namely the crossover $Damk{\ddot{o}}hler$ number, at which edge flame changes its direction of propagation. The parametric region between the quasi-steady extinction condition and the edge-flame crossover condition is a metastable region, in that flames without edge can stay in their burning states while flames with edge have to retract to expand quenching holes. Using the above properties of edge flame, Hartley and Dold proposed a Lagrangian hole dynamics, which allows us to simulate transient variation of quenching holes. In their model, each stoichiometric surface is subjected to a random sequence of scalar dissipation rate compatible to the equilibrium turbulence. Then, each stoichiometric surface will evolve, according to the combustion map, dependent on the scalar dissipation rate and existence of flame edge, If all the burning surfaces are annihilated, the event can be declared as a global extinction. The consequence obtained from the above model also can be used as a subgrid model to determine local extinction occurring in a calculation grid.

• Journal of Information Processing Systems
• /
• 제20권3호
• /
• pp.375-390
• /
• 2024

Edge computing architecture has effectively alleviated the computing pressure on cloud platforms, reduced network bandwidth consumption, and improved the quality of service for user experience; however, it has also introduced new security issues. Existing anomaly detection methods in big data scenarios with cloud-edge computing collaboration face several challenges, such as sample imbalance, difficulty in dealing with complex network traffic attacks, and difficulty in effectively training large-scale data or overly complex deep-learning network models. A lightweight deep-learning model was proposed to address these challenges. First, normalization on the user side was used to preprocess the traffic data. On the edge side, a trained Wasserstein generative adversarial network (WGAN) was used to supplement the data samples, which effectively alleviates the imbalance issue of a few types of samples while occupying a small amount of edge-computing resources. Finally, a trained lightweight deep learning network model is deployed on the edge side, and the preprocessed and expanded local data are used to fine-tune the trained model. This ensures that the data of each edge node are more consistent with the local characteristics, effectively improving the system's detection ability. In the designed lightweight deep learning network model, two sets of convolutional pooling layers of convolutional neural networks (CNN) were used to extract spatial features. The bidirectional long short-term memory network (BiLSTM) was used to collect time sequence features, and the weight of traffic features was adjusted through the attention mechanism, improving the model's ability to identify abnormal traffic features. The proposed model was experimentally demonstrated using the NSL-KDD, UNSW-NB15, and CIC-ISD2018 datasets. The accuracies of the proposed model on the three datasets were as high as 0.974, 0.925, and 0.953, respectively, showing superior accuracy to other comparative models. The proposed lightweight deep learning network model has good application prospects for anomaly traffic detection in cloud-edge collaborative computing architectures.

• 대한원격탐사학회:학술대회논문집
• /
• 대한원격탐사학회 2003년도 Proceedings of ACRS 2003 ISRS
• /
• pp.159-161
• /
• 2003

Fitting the projected wire-frame model to the detected edge pixels on images by using least-squares approach, called Least-squares Model-image Fitting (LSMIF), is the key of the Model-based Building Extraction (MBBE). It is implemented by iteratively adjusting the model parameters to minimize the squares sum of distances from the extracted edge pixels to the projected wire-frame. This paper describes a series of experiments and studies on various factors affect the fitting results, including the edge detectors, the weighting rules, the initial value of parameters, and the number of overlapped images. The experimental result is not only helpful to clarify the influences of each factor, but is also able to enhance the robustness of the LSMIF algorithm.

• ETRI Journal
• /
• 제30권5호
• /
• pp.723-728
• /
• 2008

In this paper, a simple transmission line model for an edge-coupled patch antenna is presented. The coupled section is modeled with a lump network which represents the mutual admittance between patches and from patch to ground. Theoretical analysis of two edge-coupled patch antenna models are compared by simulation and experiment in antennas designed to operate at the 2 GHz band. The proposed model predicts the return loss of the antenna accurately.

• 대한임베디드공학회논문지
• /
• 제18권3호
• /
• pp.89-100
• /
• 2023

Recently, on-device artificial intelligence (AI) solutions using mobile devices and embedded edge devices have emerged in various fields, such as computer vision, to address network traffic burdens, low-energy operations, and security problems. Although vision transformer deep learning models have outperformed conventional convolutional neural network (CNN) models in computer vision, they require more computations and parameters than CNN models. Thus, they are not directly applicable to embedded edge devices with limited hardware resources. Many researchers have proposed various model compression methods or lightweight architectures for vision transformers; however, there are only a few studies evaluating the effects of model compression techniques of vision transformers on performance. Regarding this problem, this paper presents a performance evaluation of vision transformers on embedded platforms. We investigated the behaviors of three vision transformers: DeiT, LeViT, and MobileViT. Each model performance was evaluated by accuracy and inference time on edge devices using the ImageNet dataset. We assessed the effects of the quantization method applied to the models on latency enhancement and accuracy degradation by profiling the proportion of response time occupied by major operations. In addition, we evaluated the performance of each model on GPU and EdgeTPU-based edge devices. In our experimental results, LeViT showed the best performance in CPU-based edge devices, and DeiT-small showed the highest performance improvement in GPU-based edge devices. In addition, only MobileViT models showed performance improvement on EdgeTPU. Summarizing the analysis results through profiling, the degree of performance improvement of each vision transformer model was highly dependent on the proportion of parts that could be optimized in the target edge device. In summary, to apply vision transformers to on-device AI solutions, either proper operation composition and optimizations specific to target edge devices must be considered.

• 인터넷정보학회논문지
• /
• 제4권6호
• /
• pp.33-42
• /
• 2003

본 논문에서는 Edge-Labeled Graph에 기반하여 XML 인스턴스들을 관계형 데이터베이스내에 저장하는 모델을 제안하고 구현한다. 저장 모델은 저장되는 XMI 인스턴스들을 Edge-Labeled Graph에 기반하여 데이터 그래프로 표현하며, 표현한 데이터 그래프상의 정보를 저장하기 위해 데이터베이스 스키마로 제시된 데이터 경로, 요소, 속성, 테이블 인덱스 테이블의 구조에 따라 정의된 값들을 추출하고 Mapper 모듈을 이용하여 저장하며 질의를 지원하기 위해, XPATH를 따르는 질의 언어인 XQL을 SQL로 변환하는 모듈, 또한 저장된 XML 인스턴스를 복원하는 DBtoXML 모듈을 갖도록 하였다. 구현 결과, XML 인스턴스들과 제안한 저장 모델 구조로의 저장 관계가 그래프 기반의 경로를 이용한 표현으로 가능했으며, 동시에, 특정 요소 또는 속성들의 정보들을 쉽게 검색할 수 있는 가능성을 보였다.

• 한국소음진동공학회:학술대회논문집
• /
• 한국소음진동공학회 2012년도 춘계학술대회 논문집
• /
• pp.45-54
• /
• 2012

In order to predict trailing edge noise from a flat plate more effectively and accurately, the prediction algorithm based on semi-analytic model for point pressure spectrum is proposed. The semi-analytic model consists of empirical models for point pressure spectra and theoretical model to determine the boundary layer characteristics needed for the empirical models. The proposed methods are applied to predict the trailing edge noise of the flat plate located in the mean flow of speed 38 m/s, for which the measured data are available. In present study, six empirical models for point pressure spectra are utilized for the predictions of trailing edge noise and their prediction results are compared to the measured data. Through the analysis of these comparisons, it is revealed that the present method based on non-frozen formula using Efimtsov model and Smol'yakov-Tkachenko model can provide more accurate and efficient predictions of trailing edge noise.

• 한국소음진동공학회논문집
• /
• 제22권6호
• /
• pp.524-534
• /
• 2012

In order to predict trailing edge noise from a flat plate more effectively and accurately, the prediction algorithm based on semi-analytic model for point pressure spectrum is proposed. The semi-analytic model consists of empirical models for point pressure spectra and theoretical model to determine the boundary layer characteristics needed for the empirical models. The proposed methods are applied to predict the trailing edge noise of the flat plate located in the mean flow of speed 38 m/s, for which the measured data are available. In present study, six empirical models for point pressure spectra are utilized for the predictions of trailing edge noise and their prediction results are compared to the measured data. Through the analysis of these comparisons, it is revealed that the present method based on non-frozen formula using Efimtsov model and Smol'yakov-Tkachenko model can provide more accurate and efficient predictions of trailing edge noise.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2003년도 하계종합학술대회 논문집 Ⅳ
• /
• pp.2120-2123
• /
• 2003

This paper presents the moving object detection and tracking algorithm using edge information base on human perceptual system The human visual system recognizes shapes and objects easily and rapidly. It's believed that perceptual organization plays on important role in human perception. It presents edge model(GCS) base on extracted feature by perceptual organization principal and extract edge information by definition of the edge model. Through such human perception system I have introduced the technique in which the computers would recognize the moving object from the edge information just like humans would recognize the moving object precisely.