http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
오병주(Byung-Joo Oh) 한국정보기술학회 2005 한국정보기술학회논문지 Vol.3 No.2
This paper presents a face recognition method based on the combination of the Principal Component Analysis(PCA) with the Multi-Layer Neural Networks(MLNN). The face images are preprocessed by PCA technique and produces features of the face images. The features are then taken as the input of the Multi-Layer Neural Network. The classification is carried out by using MLNN. The proposed approaches has been tested on the ORL face database. The experimental results have been demonstrated, and the recognition rate of 95% has been achieved.
Thelma M. Legaspi,오병주(Byung-Joo Oh) 한국정보기술학회 2005 한국정보기술학회논문지 Vol.3 No.3
We propose a student information service model based on the Short Messaging Service(SMS) technology that would provide basic information needs of the students. This is to further explore the use of SMS technology as a medium to deliver information on student and university administration effectively and timely. The experimental model of the system is designed and developed on the GSM mobile system using Java Development Kit (JDK) 1.4, MYSQL Server 3.23, Visual Basic 6.0 and Nokia PC Connectivity SDK 2.1. A GSM mobile phone was attached to the computer through Infrared Ray port to serve as a telecommunication link of short messages between the system and the mobile users. Respondents evaluated the system and positively recognized the importance of communications technology with greater investment in sophisticated SMS-based system services.
양근화(KeunWha Yang),오병주(Byung-Joo Oh) 한국정보기술학회 2007 한국정보기술학회논문지 Vol.5 No.1
This paper analyzes and compares a few methodologies for face recognition based on the combination of two well-known statistical representations of face images: Principal Component Analysis(PCA) and Linear Discriminant Analysis (LDA) with neural networks. PCA and LDA features of the face image are taken as the input of the Multiple Layer Neural Network (MLNN) and Radial Basis Function Network (RBFN). The classification is carried out by using MLNN and RBFN respectively. The performance of the face recognition methods are compared with each other. The proposed approaches have been tested on the ORL face image database. The experimental results have been demonstrated, and analyzed, The performance of PCA with MLNN is superior to that of the LDA with RBFN. A recognition rate of more than 95% has been achieved.
양근녕(KeunNyoung Yang),오병주(Byung-Joo Oh) 한국정보기술학회 2006 한국정보기술학회논문지 Vol.4 No.1
This paper presents a method on the face detection using moving energy and color information. The moving body image is detected by analyzing the difference of the concatenated input images. The position of the moving body is found by analyzing the edge and minimum moving area boundary blocks. The face is detected based on the relative position and boundary size of the detected person.
윤환기(Yoon Hwan-Ki),오병주(Oh Byung-Joo) 한국태양에너지학회 1994 한국태양에너지학회 논문집 Vol.14 No.1
본 논문은 기존의 추적방식인 아날로그 방식의 하드웨어적 제어에서 탈피하여 디지탈 방식의 소프트웨어적 제어방식인 적응제어 이론을 태양 추적장치에 적용하여 정확하고 간단하게 추적하는 제어장치를 설계 제작하고, 이에 대한 성능을 평가하였다.<br/> 추적회로에 적용할 제어이론은 최근 로보트 제어에 적용하여 각광을 받고있는 Adaptive independent control 이론을 도입하였다. 이 제어이론을 태양 주적장치에 적용했을 경우의 효과에 대해 모델링에 의한 컴퓨터 시뮬레이션을 행하였고, 그 결과, 제어이론을 적용한 경우가 좀더 정확한 추적을 행하며, 추적오차를 약 1/2 정도 줄일 수 있음을 알았다. 추적 제어장치의 제작은 DALLAS 사의 DS5000T를 마이크로 프로세터로 사용하여 회로를 간소화 시켰으며, 제작된 추적장치를 실제로 동작시켜 추적이 제대로 행해지는지를 엔코더와 Data Acquisition System을 설치하여 측정하였다. 측정 결과는 시뮬레이션의 경우와 근접했으며, 적응제어 이론을 적용한 추적 제어장치가 태양을 더 정확하게 추적함을 알 수 있었다. This study aims at developing the control device of solar tracking system which can accurately track the sun with a simple operation. Actual control device is designed and manufactured to evaluate its performance. The digital control method base on recently developed adaptive control theory is adopted for main control algorithm instead of a traditional analog control method.<br/> Adaptive independent control theory is introduced for the control algorithm of tracking circuit, which is prevailing in the field of robot control nowadays. Computer simulation of solar tracking model reveals the new device operated by control theory could reduce the error to one half with the comparison of a existing device. Tracking circuit can be simplified by adopting the DS5000T ?-processor made by the DALLAS company. Actual measurement is conducted to verify the correct operation of manufactured control device with a encorder and a data acquisition system. The experimental results coincided with the simulation results, and our new tracking control device operated by adaptive control theory can track the sun more accuratly.
서브윈도우 회전 아다부스팅 알고리즘을 이용한 얼굴 영상 검출
양근녕(Keun-Nyoung Yang),오병주(Byung-Joo Oh) 한국정보기술학회 2006 한국정보기술학회논문지 Vol.4 No.2
This paper proposes a rotation invariant face detection using sub-window rotation in adaboosting algorithm. The sub-window used in the algorithm is rotated in stead of the wavelet rotation. The proposed rotation is used to detect rotated face from -90 to +60 degree. The result shows a better performace in the error detection and speed compared to the previously presented algorithm.. The algorithm is implemented and tested using MATLab program.
PCA와 Back-Propagation의 결합에 의한 얼굴인식
양근화(Keun-Hwa Yang),오병주(Byung-Joo Oh) 한국정보기술학회 2004 Proceedings of KIIT Conference Vol.2004 No.-
본 논문은 Principal Component Analysis(PCA)와 Backpropagation(BP)을 이용하여 얼굴을 인식하는 기법을 제안한다. 원래의 얼굴영상에 PCA기법을 적용하여 저차원의 고유벡터(eigen-vector)와 고유얼굴(eigen-face)를 생성하여 BP에 적용하여 가중치를 산출하는 방식이다. 훈련DB에 있는 얼굴영상들 에 대해 PCA를 적용하여 얻어진 고유벡터를 BP의 입력으로 하여 BP 가중치를 일정한 오차범위까지 학습한다. 다음에 테스트DB에서 PCA와 BP를 통해 얻어진 새로운 가중치를 훈련DB의 학습과정에서 얻어진 가중치와 비교함으로써 가장 적합한 얼굴을 찾아내게 된다. 본 연구에서는 PCA와 PCA+BP의 인식률 비교를 해본 결과 신경망을 통한 인식률이 더 뛰어나다는 결과를 얻을 수 있었다. This paper proposes a method of face recognition based on the PCA and Back - Propagation. PCA method produces a low - dimensional eigenvector and eigenface for the original face images. Then the back-propagation's weight is trained using the produced low - dimensional vectors. The proposed result shows better performance compared with those of PCA or BP methods.