http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
패턴 인식을 위한 Dynamic Weight Warping 알고리즘에 관한 연구
문성룡 圓光大學校 大學院 1996 論文集 Vol.16 No.-
본 논문에서는 임의의 패턴이 입력되면 일정한 패턴이 저장된 기준패턴과 입력패턴을 같은 면적 크기로 정규화하여 두 패턴을 이루는 픽셀의 가중치(weight)를 거리(distance)로 환산하여 이 거리에 의해 패턴간의 유사성을 찾아 가장 유사한 패턴을 선택하는 새로운 Dynamic Weight Warping (DWW)알고리즘을 제안하였다. 입력패턴과 기준패턴의 무게를 제안된 DWW 알고리즘을 이용하여 구하게되며, 얻어진 가중치는 기준패턴중에서 MAXNET에 의해서 입력패턴과 가장 유사한 기준패턴을 선택하게 된다. 인식하려는 패턴의 크기를 5 ×7로 면적 정규화하여 처리마며 가중치가 가장 큰 기중패턴을 선택하여 Global Winner에 저장하고, 저장된 7행의 Global Winner중에서 가장 입력패턴과 유사한 패턴을 선택한다. 제안된 DWW 알고리즘은 COMPASS 틀을 이용하여 구현하였고, 이에 대한 동작특성은 모의 실험 결과를 통하여 검증하였다. In this Paper, we tried to improve the pattern recognition in processing the pattern with the noise and speed to make the real-time processing possible by the proposed Cynamic Weight Warping algorithm using digital system. The DWW algorithm circuit designed to have hierachical structure according to a kind of refernce patterns and identified operating charateristic of designed DWW algorithm by the computer simulation and then it is designed by COMPASS Tool.
Mathematical Morphology를 이용한 이미지 해석에 관한 연구
문성룡,유창규,김종교 전북대학교 전자산업개발연구소 1991 전자산업연구 Vol.2 No.-
Mathematical morphology is very efficient for analysis of geometrical characteristics of signals and systems, and a predominant tool for smoothing data with noise. In this study the several opertions using binary and gray scale morphologies (Erosion, Dilation, Opening, Closing, Closing-Opening(CO), Opening-Closing(OC), etc)are discussed and the good results for an image with noise are obtained through computer simulation.
그레이 스케일 형상학을 이용한 DTCNN의 하드웨어 구현에 관한 연구
문성룡,윤여근 원광대학교 공업기술개발연구소 1997 工業技術開發硏究誌 Vol.17 No.-
The application of morphological methods in image analysis has often shown its efficiency by providing a geometrical approach to signal processing while other techniques offer a spatial frequency approach. However, the morphology operators are highly computation intensive because of their iterative, sequential and irregular behaviour and therefore, it is difficult to real time processing. In order to improve the performances, parallel processing is a necessity. Morphological expressions can be transformed by algebraic rules to provide an efficient DTCNN (Dlscrete-Time Cellular Neural Networks) implementation for complex operations on images. In this paper, The circuit of DTCNN was designed using dilaion and erosion of basic morphology operators, also each cell designed PE to strengthen using local connectivity. Each cell makes routing easier and make suitable for VLSI implementation, operation of designed PE implemented circuit to parallel processing and strengthen. As a result of simulations, the proposed method verified to improve high speed operation in comparision with the sequential data operation, parallel processing DTCNN was implemented in a 0.8μm CMOS technology using COMPASS Tool.
계층적 구조를 갖는 DBNN 의 하드웨어 구현에 관한 연구
문성룡 원광대학교 공업기술개발연구소 1995 工業技術開發硏究誌 Vol.15 No.-
In this paper, We tried to improve the pattern recognition in processing the image signal with the noise and speed to make the real-time processing possible by implementing hardware of DBHIN algorithm using digital system. The DBNN algorithm circuit designed to have a enlargement according to a kind of reference patterns and identified operating characteristic of designed hardware DBNN by the computer simulation and then it is designed by Verilog-HDL simulator.
Opening-Closing 연산을 이용한 하이브리드 가중메디안 필터에 관한 연구
문성룡 원광대학교 공업기술개발연구소 1994 工業技術開發硏究誌 Vol.14 No.-
Morphological filter is generally composed of several morphological operation. which are the diverse structuring element. Morphological operation apply to design of opening, closing filter, are defined based on erosion and dilation of operation. Performance of niter to applied morphological filter and weighted median filter is better excellent than different image qualify to reduce noise of image data with salt-and-pepper noise, In this paper, to use the opening-closing filter and weighted median filter in precessing the image. is shown that hybrid weighted median niter has better results image quality to analyze by computer simulation.
국부적 디지탈 로직을 갖는 프로그래머블 CNN 구현에 관한 연구
문성룡,윤여근 원광대학교 공업기술개발연구소 1996 工業技術開發硏究誌 Vol.16 No.-
Neural network can potentially be used for practical applications such as nonlinear signal processing in real time processing. According to the learning method, system is severally characteristic. In the circuit implementation of the neural system, system is complex. An approach to the voltage - controlled current source in areas neural networks, the key features of CNN is locally connected only to its neighbors. The architecture of the interconnection elements between cells is very simple and space invariant. CNNs are suitable for VLSI implementation. In this paper, a processing element of CNN with digital programmable synapses is designed using the current mode circuit. Gain control technique can be used to search for the optimal solutions in the CNN output. After identified operation of cell through the simulation, processing element is designed in a VLSI technology.
IoT 기반 건물 내부 지도 및 화재 안내 시스템에 관한 연구
문성룡,조준호 한국디지털정책학회 2019 디지털융복합연구 Vol.17 No.1
This paper is a study on IoT based map inside the building and fire perception system using microprocessor and LABVIEW program. The smart control system implemented in this paper is designed to identify the location of fire by using microprocessor, flame detection sensor, carbon monoxide sensor and temperature sensor, and to guide the optimal travel route through Zigbee communication. And the proposed system uses QR code to interoperate with smartphone. The coordinator control verified that the sensor value of the smart control system installed through the LABVIEW software was confirmed. The IoT based control system studied in this paper was implemented with Arduino mega board and LABVIEW software, and the operation status was confirmed by display device and coordination. 본 논문은 마이크로프로세서와 랩뷰 프로그램을 이용한 IoT 기반의 건물 내부 지도 및 화재 안내 시스템에 관현 연구 이다. 지금까지의 건물 내부의 비상구를 안내하는 장치는 화재 발생시 화재장소와 상관없이 항상 일정한 방향만을 나타내어 화재로부터 건물 밖까지 최적의 비상구를 나타내어 주지 않고 있다. 논문에서 구현된 스마트 제어 시스템은 마이크로프로세서와 불꽃 감지센서, 일산화탄소센서 및 온도센서를 이용하여 화재 발생 장소를 파악하였고, Zigbee 통신을 통하여 최적의 이동경로를 안내하도록 구성되어 있다. 그리고 제안된 시스템은 QR코드를 활용하여 스마트폰과 연동되게 하였다. 코디네이터 제어는 LABVIEW 소프트웨어를 통하여 설치된 스마트제어 시스템의 센서 값이 확인되도록 검증하였다. 본 논문에서 연구된 IoT 기반 제어시스템은 Arduino mega 보드와 LABVIEW 소프트웨어로 구현하였고, 디스플레이 장치와 코디네이션으로 동작 상태를 확인 할 수 있었다.