• 한국지능시스템학회논문지
• /
• 제12권5호
• /
• pp.467-472
• /
• 2002

본 논문에서, 영상에서 임펄스 잡음을 효과적으로 제거하고, 연산 속도를 개선하기 위해 Fuzzy Cellular Neural Network(FCNN)구조에 Hausdorff distance(HD)를 적용한 $\alpha$-Least Trimmed Square HD($\alpha$-LTSHD) 기반 FCNN 구조를 제안한다. FCNN는 Cellular Neural Network(CNN) 구조에 퍼지 이론을 적용한 것이고, HD는 특징 대상의 대응 없이 이진 영상의 두 픽셀 집합 사이의 거리를 구하는 척도로 물체의 정합에 널리 사용한다. 성능 평가를 위해, 제안된 방법을 MSE와 SNR을 이용하여 기존 FCNN, Opening-Closing(OC) 그리고 LTSHD 연산자를 적용한 FCNN과 비교 분석하였다. 그 결과, 본 논문에서 제안된 망(network) 구조의 성능이 다른 필터보다 임펄스 잡음 제거에 우수함을 확인하였다.

• 전자공학회논문지S
• /
• 제35S권11호
• /
• pp.144-151
• /
• 1998

논문에서는 템플릿과 입력/출력사이에 퍼지논리를 결합한 CNN의 구조를 갖는 FCNN에 미리 정의된 가중구조적 요소의 조합으로 최소/최대 연산을 수행하기 위하여 수학적 형태학의 분해특성을 적용하였다. 본 논문에서는 잡음이 포함된 영상에 퍼지 estimator를 적용함으로써 잡음을 제거함과 동시에 에지 검출이 가능함을 컴퓨터 모의 실험을 통해 확인하였다. 형태학의 erosion 연산을 적용한 FCNN과 CNN, 메디안필터를 이용한 에지 검출 결과를 비교분석 하였으며, 양자화 퍼지 함수를 이용한 FCNN에 고정 템플릿과 가변적 템플릿을 적용한 결과 FCNN의 성능이 향상됨을 컴퓨터 모의실험을 통하여 확인하였다.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 1999년도 하계종합학술대회 논문집
• /
• pp.1025-1028
• /
• 1999

After detecting the edge which is applying the morphological operators to the hybrid FCNN, we could analyze and compare. The hybrid FCNN is completely removed to the noise in the image, and worked in order to obtain the result image which is closest to the original image. Also, the morphological operator is applied to the image as the method in order to detect more good the edge than the conventional edge. FCNN which is the pipeline type is completely suitable to detecting the image processing as well as the hardware size. In this paper. we would make the structure elements of the morphological operator the variable template and the static template, and compare with the edge enhancement of two images. After being the result which is applying the variable template morphological operator and the static template morphological operator to the image, we could know that the edge images applying the variable template is superior in a edge enhancement side.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2000년도 추계종합학술대회 논문집(3)
• /
• pp.13-16
• /
• 2000

This paper is to compare DTCNN(Discrete-time Cellular Neural Networks) applying the fuzzy morphology operators with the conventional FCNN(Fuzzy CNN) using the general morphology operators. These methods are to the image filtering, and are compared as MSE. Also the main goal of this paper is to compare the fuzzy morphology operators with the general morphology operators through image input. In a result of computer simulation, we could know that the error of DTCNN applying the fuzzy morphology operators is less about 6.1809 than FCNN using the general morphology operators in the image included 10% noise, also the error of the former is less about 5.5922 than the latter in the image included 20% noise. And the image of DTCNN applying the fuzzy morphology operators is superior to FCNN using the general morphology operators.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2000년도 ITC-CSCC -1
• /
• pp.50-53
• /
• 2000

It is effectively removed noise in the image using FCNN(Fuzzy Cellular Neural Network) applying fuzzy theory to CNN(Cellular Neural Network) structure and HD(Hausdorff Distance) commonly used measures for object matching. HD calculates the distance between two point set of pixels in two-dimensional binary images without establishing correspondence. Also, this method is proposed in order to improve the operation speed. In this paper, $\alpha$-LTSHD(Least Trimmed Square HD) operator applying $\alpha$-Trimmed to LTSHD, one field of HD, is applied to FCNN structure, and it is proposed as the modified method in order to remove noise in the image. Also, it is made a comparison with the other filters by using MSE and SNR after removing noise using the FCNNS which are applied $\alpha$-LTSHD operator through the computer simulation. In a result, FCNN performance which is applied the proposed $\alpha$-LTSHD demonstrated the superiority to the other filters in the noise removal.

• Current Optics and Photonics
• /
• 제8권3호
• /
• pp.239-245
• /
• 2024

This paper proposes single logarithmic amplification (single-LA) and deep learning (DL)-based fixed-threshold on-off keying (OOK) detection for free-space optical (FSO) communication. Multilevel LAs (MLAs) can be used to mitigate intensity fluctuations in the received OOK signal by their nonlinear gain characteristics; however, it is ineffective in the case of high scintillation, owing to degradation of the OOK signal's extinction ratio. Therefore, a DL technique is applied to realize effective scintillation compensation in single-LA applications. Fully connected (FC) networks and fully connected neural networks (FCNN), which have nonlinear modeling characteristics, are deployed in this work. The performance of the proposed method is evaluated through simulations under various scintillation effects. Simulation results show that the proposed method outperforms the conventional adaptive-threshold-decision, single-LA-based, MLA-based, FC-based, and FCNN-based OOK detection techniques.

• 한국항공우주학회지
• /
• 제49권3호
• /
• pp.197-204
• /
• 2021

유도탄의 비행 시험 중 고장 또는 비정상적인 기동이 발생하는 경우 비행을 계속하지 않도록 의도적으로 자폭한다. 이때 파편이 발생하며 안전 지역을 벗어났는지 여부를 실시간으로 추정하는 것이 중요하다. 본 논문에서는 Fully-Connected Neural Network(FCNN)를 이용하여 실시간으로 파편의 예상 낙하 영역 및 낙하 시간을 추정하는 방법을 제안한다. 많은 양의 학습 데이터 생성을 위해 Unscented Transform(UT)를 적용하였으며 신뢰도 확보를 위해 Monte-Carlo(MC) 시뮬레이션과 비교하여 파라미터를 선정하였다. 또한 제안한 방법의 추정 결과를 MC와 비교하여 성능을 분석하였다.

• 전기전자학회논문지
• /
• 제28권2호
• /
• pp.158-167
• /
• 2024

본 논문에서는, 시뮬레이션 상에서 반도체 후단 공정의 프로세스를 구현하고 파이프 내부 상황을 모니터링하기 위해 전기 정전용량을 기반으로 한 U-net 모델을 적용하였다. 배관에 부착된 전극에서 측정한 정전용량 값은 U-net 네트워크 모델의 입력 데이터로 사용되며, 모델을 통해 추정한 유전율 분포를 가지고 파이프 단면을 이미지화하였다. 성능 평가를 위해 수치 시뮬레이션 얀에서 U-net 모델, FCNN(Fully-connected neural network) 모델, Newton-Raphson 방법으로 재구성한 이미지를 비교한 결과, U-net이 다른 이미지 복원 방식보다 좋은 복원 성능을 보였다.

• 국제학술발표논문집
• /
• The 9th International Conference on Construction Engineering and Project Management
• /
• pp.814-822
• /
• 2022

The deployment of sensors for Structural Health Monitoring requires a complicated network arrangement, ground truthing, and calibration for validating sensor performance periodically. Any conventional sensor on a structural element is also subjected to static and dynamic vertical loadings in conjunction with other environmental factors, such as brightness, noise, temperature, and humidity. A structural model with strain gauges was built and tested to get realistic sensory information. This paper investigates different deep learning architectures and algorithms, including unsupervised, autoencoder, and supervised methods, to benchmark blind drift calibration methods using deep learning. It involves a fully connected neural network (FCNN), a long short-term memory (LSTM), and a gated recurrent unit (GRU) to address the blind drift calibration problem (i.e., performing calibrations of installed sensors when ground truth is not available). The results show that the supervised methods perform much better than unsupervised methods, such as an autoencoder, when ground truths are available. Furthermore, taking advantage of time-series information, the GRU model generates the most precise predictions to remove the drift overall.

• Wind and Structures
• /
• 제38권4호
• /
• pp.231-244
• /
• 2024

As wind farms expand into low wind speed areas, an increasing number are being established in mountainous regions. To fully utilize wind energy resources, it is essential to understand the details of mountain flow fields. Reconstructing the wind speed field in complex terrain is crucial for planning, designing, operation of wind farms, which impacts the wind farm's profits throughout its life cycle. Currently, wind speed reconstruction is primarily achieved through physical and machine learning methods. However, physical methods often require significant computational costs. Therefore, we propose a Full Convolutional Neural Network (FCNN)-based reconstruction method for mountain wind velocity fields to evaluate wind resources more accurately and efficiently. This method establishes the mapping relation between terrain, wind angle, height, and corresponding velocity fields of three velocity components within a specific terrain range. Guided by this mapping relation, wind velocity fields of three components at different terrains, wind angles, and heights can be generated. The effectiveness of this method was demonstrated by reconstructing the wind speed field of complex terrain in Beijing.