http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
진경찬,조진호,Jin, Kyung-Chan,Cho, Jin-Ho The Institute of Electronics and Information Engin 2001 電子工學會論文誌-SC (System and control) Vol.38 No.1
PISA 방법은 주로 승모판에서 역류하는 혈류량을 측정하기 위해 사용되고 있다. 이 방법은 PISA isotach의 기하학적 모양에 대한 모델링에 관한 것이다. PISA의 일반적인 유동모델은 isotach의 표면이 수식적으로 반구이거나 비반구임을 가정하여 계산된 것이지만, 본 논문에서는 영역기반방법으로 isotach를 추정한 후, 타원체의 높이에 기초한 실제적인 표면분할 유동모델을 이용하여 유체량을 추정하였다. 제안한 밥법을 평가하기 위해, $30cm^3/sec-60\;cm^3/sec$의 실제 유량을 가지는 동적인 180개의 유동영상에 대해서 기존 방법들과 비교하였다. 실험한 결과, 반구 유동모델의 유체량 평균이 $29\;cm^3/sec$로 실제 유체량 평균보다 35%정도 적게 추정을 하였고, 제안한 방법의 평균은 $45\;cm^3/sec$으로 비반구 유동모델의 평균과 같았고, 유체량 변화파형도 유사한 결과를 가짐을 알 수 있었다. The proximal isovelocity surface area (PISA) method is an effective way of measuring the regurgitant blood flow rate in the mitral valve. This method defines the modelling required to describe the geometry of the isotach of the PISA. In the normal PISA flow model, the flow rate is calculated assuming that the surface of the isotach is either hemispherical or non-hemispherical numerically. However, this paper evaluated the estimate flow rate using a direct surface subdivision flow model based on the height field after isotach extraction using a region-based scheme. To validate the proposed method, the various PISA flow models were compared using pusatile color Doppler images with flow rates ranging from $30\;cm^3/sec\;to\;60\;cm^3/sec$ flow rate. Whereas the hemispherical flow model had a mean value of $29\;cm^3/sec$ and underestimated the measured flow rate by 35%, the proposed model and non-hemispherical model produced a c;ame mean value of $45\;cm^3/sec$, moreover, both flow models produced a similar pulsatile flow rate.
SVD 알고리즘 및 HMM을 이용한 얼굴 및 눈 패턴 검출
진경찬,김명남,신장규,손병기,조진호 경북대학교 센서기술연구소 1998 센서技術學術大會論文集 Vol.9 No.1
The studies about automatic pattern detection of the eye and face from the human image acquired by the CCD image sensor have good applicabilities in the industry, home automation, and data communication field. In general, pattern detection method consists of feature based matching and template matching. In feature based matching, the feature vector is extracted with DLM(dynamic linking matching), EBGM(elastic bunch graph matching), HMM(hidden markov model) matching and knowledge based matching using statistical characteristics. In template matching, in general, the template vector is extracted with PCA(principal component analysis). When these method applied in the face and eye detection, each method has its own merits and some disadvantage. Therefore, by combined utilization of SVD(singular value decomposition) and HMM algorithm, is expected that we can selectively make use of each methods advantage and it result in improved detection accuracy. In this paper, we proposed the method for face and eye detection, which was combined by the two algorithms, to be suitable for the high speed image processing using DSP chip or microprocessor. In the beginning, template matching was followed by a template extraction using batch SVD algorithm and then face pattern was classified and recognized by HMM algorithm which is one of feature based matching technique. Finally, eye pattern detection was performed by pattern search neural network utilizing eigeneye image.
빠른 피쳐변위수렴을 위한 BMA을 이용한 STK 피쳐 추적
진경찬,조진호,Jin, Kyung-Chan,Cho, Jin-Ho 대한전자공학회 1999 電子工學會論文誌, S Vol.s36 No.8
일반거인 피쳐검출 및 추적 알고리즘에는 Garbor-jet를 이용한 elastic bunch graph matching (EBGM), rotation normalized cross-correlation (NCC-R) 및 화소의 고유치를 이용한 Shi-Tomasi-Kanade(STK) 알고리즘 등이 있다. 이들 중에서 EBGM, NCC-R은 피쳐모델에 의해 피쳐를 검출하지만 STK 알고리즘은 피쳐를 자동적으로 검출하는 특징을 가진다. 본 논문에서는 STK알고리즘인 Newton-Raphson (NR) 추적의 초기화 문제를 해결하기 위해서 모델링된 피쳐영역에서 STK 알고리즘으로 피쳐를 검출한 후, NR 방법으로 피쳐를 추적할 때, NR 방법에 의한 피쳐추적의 정확성을 개선시키기 위해 block matching agorithm (BMA)-NR 방법을 제안하였다. NR 방법에 의한 피쳐변위수렴시 BMA-NR 방법이 NBMA-NR (no BMA-NR)방법보다 피쳐추적의 정확성이 향상되었는데, 이는 NR의 서치영역크기로 인한 국소 최소치(local minimum) 문제를 해결하였기 때문이다. In general, feature detection and tracking algorithms is classified by EBGM using Garbor-jet, NNC-R and STK algorithm using pixel eigenvalue. In those algorithms, EBGM and NCC-R detect features with feature model, but STK algorithm has a characteristics of an automatic feature selection. In this paper, to solve the initial problem of NR tracking in STK algorithm, we detected features using STK algorithm in modelled feature region and tracked features with NR method. In tracking, to improve the tracking accuracy for features by NR method, we proposed BMA-NR method. We evaluated that BMA-NR method was superior to NBMA-NR in that feature tracking accuracy, since BMA-NR method was able to solve the local minimum problem due to search window size of NR.