http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
연역데이타베이스에서의 무결성 규칙의 컴파일링에 관한 연구
진성일 충남대학교 자연과학연구소 1989 忠南科學硏究誌 Vol.16 No.2
We must consider implicit insertions and deletions affected by update tuple when we check the integrity constraints in deductive database. In this study, we propose a method of integrity constraint checking that we prepare compiled formulas for each EDB predicate affecting the integrity constraints in advance and at the time of update we test if compiled formulas instantiated by the constant of update tuple be true or not using relational operations.
Gauss 함수 근사식에 의한 α Profile 단일모드 광섬유의 전파특성 해석
진성일 경북대학교 공과대학 1982 工大硏究誌 Vol.11 No.-
Formulae that calculate the beam width and the propagation constant of HE_11 mode of a profile fibers are derived under the Gaussian approximation and solved by the numerical method. The same calculation is performed for the case, in which the dip in the central core of the fiber is modelled by the appropriate equation.
진성일,배건성,박현화,강해동 경북대학교 전자기술연구소 1991 電子技術硏究誌 Vol.12 No.1
Taking advantage of the property that Korean digit is a monosyllable word, we proposed a spoken Korean-digit recognition scheme using the multi-layer perception. Dynamic spectral features obtained from the linear regression of cepstrum were used as an input vector of neural network. The effects of the number of nodes in the hidden layer to the recognition rate and learning rate were investigated.
진성일,강봉수,정동현 경북대학교 전자기술연구소 1999 電子技術硏究誌 Vol.20 No.1
In this paper, we developed a CCD camera-based serial number recognition system which can recognize the serial number printed on the electronic products. Much noise is expected to be sometimes included in the acquired images due to the external environments such as the lighting or the texture of the products. So, we needed to propose some image processing algorithms that are robust to such environmental changes. We employed the gray-level normalization for image enhancement, mode-selective binarization method for thresholding, and median filtering for noise elimination And neural network was adapted for our recognition system Predefined character information(character width, height, aspect ratio, etc.) was introduced as knowledge database. For enhancing the reliability of the recognition system we proposed co-working of the character segmentation-recognition. Using this system, we could acquire the 98.2% recognition performance for the characters extracted from about 200 input images.
진성일,김종우,Chien, Sung-Il,Kim, Jong-Woo 대한전자공학회 1990 전자공학회논문지 Vol. No.
본 논문은 2-D 영상의 운동 변화량(크기, 회전, 이동)을 추정할 수 있는 새로운 방법을 제안하였다. 이 방법은 입력 영상내의 직선 성분을 Hough 영역에서 한 점으로 대응시키는 Hough 변환 알고리듬을 사용하여 입력 영상에서 크기 변화가 있는 물체의 운동과 그 Hough space상의 특징점들의 변화를 나타내는 운동 변화량들에 대한 관계식을 유도한다. 유도한 관계식은 크기, 회전, 이동을 가지고 모든 입력 영상에 대해서 운동 변화량을 효율적으로 계산할 수 있다. 제안된 관계식을 항공기 영상에 적용하여 운동 변화량을 비교적 정확히 측정하였다. 또한 제안된 방식이 잡음이 섞인 영상에 대해서도 큰 오차 없이 추정 결과를 나타내고 있음을 보인다. A new method for determining the motion parameters (scale, rotation, translation) of 2-D image is introduced. It employs Hough transform that maps the straight lines in the input image to the points in the Hough space (HS). This method makes use of the relations between the motion of an object in input image and the translations of peak points in the HS and thus derives relating equations about motion parameters especially when scale changes are involved. The derived equations make is efficient and simple to estimate motion parameters of input image, even if the scale parameter of input image is varied. Performance of this approach on an aircraft image is provided in detail in the presence of noise.