• 한국멀티미디어학회논문지
• /
• 제13권6호
• /
• pp.817-824
• /
• 2010

Linear Discriminant Analysis (LDA) has been successfully applied for dimension reduction in face recognition. However, LDA requires the transformation of a face image to a one-dimensional vector and this process can cause the correlation information among neighboring pixels to be disregarded. On the other hand, 2D-LDA uses 2D images directly without a transformation process and it has been shown to be superior to the traditional LDA. Nevertheless, there are some problems in 2D-LDA. First, it is difficult to determine the optimal number of feature vectors in a reduced dimensional space. Second, the size of rectangular windows used in 2D-LDA makes strong impacts on classification accuracies but there is no reliable way to determine an optimal window size. In this paper, we propose a new algorithm to overcome those problems in 2D-LDA. We adopt an ensemble approach which combines several classifiers obtained by utilizing various window sizes. And a practical method to determine the number of feature vectors is also presented. Experimental results demonstrate that the proposed method can overcome the difficulties with choosing an optimal window size and the number of feature vectors.

• 한국멀티미디어학회논문지
• /
• 제17권12호
• /
• pp.1446-1452
• /
• 2014

Existing LDA uses the transform matrix that maximizes distance between classes. So we have to convert from an image to one-dimensional vector as training vector. However, in 2D-LDA, we can directly use two-dimensional image itself as training matrix, so that the classification performance can be enhanced about 20% comparing LDA, since the training matrix preserves the spatial information of two-dimensional image. However 2D-LDA uses same calculation schema for transformation matrix and therefore both LDA and 2D-LDA has the heteroscedastic problem which means that the class classification cannot obtain beneficial information of spatial distances of class clusters since LDA uses only data correlation-based covariance matrix of the training data without any reference to distances between classes. In this paper, we propose a new method to apply training matrix of 2D-LDA by using WPS-LDA idea that calculates the reciprocal of distance between classes and apply this weight to between class scatter matrix. The experimental result shows that the discriminating power of proposed 2D-LDA with weighted between class scatter has been improved up to 2% than original 2D-LDA. This method has good performance, especially when the distance between two classes is very close and the dimension of projection axis is low.

• 한국콘텐츠학회논문지
• /
• 제10권8호
• /
• pp.18-25
• /
• 2010

본 논문에서는 열방향 2차원 PCA(Column-directional 2 Dimensional PCA, C2DPCA) 와 행방향 2차원 LDA(Row-directional 2 Dimensional LDA, R2DLDA)를 사용하여 얻은 각각의 투영 행렬을 동시에 사용하는 방법을 제안하였다. 제안 방법은 얼굴의 가로 특징과 세로 특징을 모두 포함한 저 차원의 특징 행렬을 얻음으로써, 훈련 영상의 수에 관계없이 안정적이고 높은 인식률을 얻을 수 있다. 또한, 같은 알고리즘으로 가로 방향과 세로 방향에 PCA와 LDA를 각각 달리 적용한 실험(C2DPCA & R2DLDA, C2DLDA & R2DPCA)에서 가로 방향의 특징에 2차원 LDA를 적용한 시스템(C2DPCA & R2DLDA)이 그 반대의 경우보다 저차원으로 높은 인식률을 얻을 수 있음을 확인할 수 있었다. 실험 결과 제안한 방법이 2DPCA와 2DLDA 등 의 기존 방법보다 인식율이 높은 99.4%를 얻었다. 또한 제안 방법의 인식 처리속도도 기존의 2DPCA와 2DLDA 방법보다 3배 이상 빠름을 확인하였다.

• 한국통신학회논문지
• /
• 제30권12C호
• /
• pp.1162-1166
• /
• 2005

본 논문에서는 얼굴 인식을 위한 새로운 저차원 특징 표현 기법을 제안하였다. 선형판별기법(LDA)는 인기있는 특징추출 기법이다. 하지만 고차원 데이터의 경우에 계산적인 복잡도가 높고 샘플의 개수가 적은 경우 역행렬을 구할 수 없는 특이행렬문제에 직면한다. 이러한 문제들을 해결하기 위해 일반적인 선형판별기법과 다르게 우리는 이차원 이미지 공분산 행렬을 구한 다음 직접선형판별기법(dirct LDA)을 적용하였으며 이것을 2D-DLDA라고 부른다. ORL 얼굴데이터베이스를 사용하여 실험한 결과 기존의 직접선형판별기법보다 성능이 우수함을 확인하였다.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제16권9호
• /
• pp.2991-3007
• /
• 2022

Two dimensional locality preserving projections (2D-LPP) is an improved algorithm of 2D image to solve the small sample size (SSS) problems which locality preserving projections (LPP) meets. It's able to find the low dimension manifold mapping that not only preserves local information but also detects manifold embedded in original data spaces. However, 2D-LPP is simple and elegant. So, inspired by the comparison experiments between two dimensional linear discriminant analysis (2D-LDA) and linear discriminant analysis (LDA) which indicated that matrix based methods don't always perform better even when training samples are limited, we surmise 2D-LPP may meet the same limitation as 2D-LDA and propose a novel matrix exponential method to enhance the performance of 2D-LPP. 2D-MELPP is equivalent to employing distance diffusion mapping to transform original images into a new space, and margins between labels are broadened, which is beneficial for solving classification problems. Nonetheless, the computational time complexity of 2D-MELPP is extremely high. In this paper, we replace some of matrix multiplications with multiple multiplications to save the memory cost and provide an efficient way for solving 2D-MELPP. We test it on public databases: random 3D data set, ORL, AR face database and Polyu Palmprint database and compare it with other 2D methods like 2D-LDA, 2D-LPP and 1D methods like LPP and exponential locality preserving projections (ELPP), finding it outperforms than others in recognition accuracy. We also compare different dimensions of projection vector and record the cost time on the ORL, AR face database and Polyu Palmprint database. The experiment results above proves that our advanced algorithm has a better performance on 3 independent public databases.

• 대한기계학회:학술대회논문집
• /
• 대한기계학회 2001년도 추계학술대회논문집B
• /
• pp.371-376
• /
• 2001

The operating point of a small-sized axial fan for refrigerator is strongly dependent upon the system resistance. Therefore, the turbulent flow characteristics around a small-sized axial fan may change significantly according to the operating point. This study represents three-dimensional turbulent flow characteristics around a small-sized axial fan measured at the four operating points such as $\varphi=0.1$, 0.18, 0.25 and 0.32 by using fiber-optic type LDA system. This LDA system is composed of a 5 W Argon-ion laser, two optics in back-scatter mode, three BSA's, a PC, and a three-dimensional automatic traversing system. A kind of paraffin fluid is utilized for supplying particles by means of fog generator. Mean velocity profiles downstream of a small-sized axial fan along the radial distance show that both the streamwise and the tangential components exist predominantly in downstream except $\varphi=0.1$ and have a maximum value at the radial distance ratio of about 0.8, but the radial component, which its velocity is relatively small, is acting role that only turns flow direction to the outside or the central part of axial fan. Moreover, all of the velocity components downstream at $\varphi=0.1$ show much smaller than those upstream due to the static pressure rise at the low-flowrate region.

• 동력기계공학회지
• /
• 제4권3호
• /
• pp.25-33
• /
• 2000

This study represents three-dimensional turbulent flow characteristics around an axial fan measured at the operating point ${\varphi}=0.32$, which is equivalent to the maximum flowrate region, by using three-dimensional fiber-optic type LDA system. This LDA system is composed of a 5 W Argon-ion laser, two optics in back-scatter mode, three BSA's, a PC, and a three-dimensional automatic traversing system. A kind of paraffin fog is used for laser particles in this study. Mean velocity profiles around an axial fan along the downstream radial distance show that the streamwise and the tangential components exist as a predominant velocity and have the maximum value at the radial distance ratio 0.8, while the radial component has a small scale distribution and its flow direction is inward except a part of blade tip. The turbulent intensity profiles show that the radial component exists the most greatly. And also the turbulent kinetic energy shows about 60% as a maximum value at the radial distance ratio 0.9. Moreover, the Reynolds shear stresses do not exist at upstream flow, but the streamwise and the radial components of them show about 20% as a maximum value at the radial distance ratio 0.9 at downstream flow.

• 대한기계학회논문집B
• /
• 제25권6호
• /
• pp.819-828
• /
• 2001

The operating point of a small-sized axial fan is strongly dependent upon the system resistance. Therefore, the turbulent flow characteristics around a small-sized axial fan may change significantly according to the operating point. This study represents three-dimensional turbulent flow characteristics around a small-sized axial fan measured at the ideal design point $\phi$=0.25, which is equivalent to the maximum total efficiency point, by using three dimensional fiber-optic type LDA system. This LDA system is composed of a 5 W Argon-ion laser, two optics in back-scatter mode, three BSAs, a PC, and a three-dimensional automatic traversing system. A kind of paraffin fluid is used to supply particles by means of fog generator. Mean velocity profiles downstream of a small-sized axial fan along the radial distance show that the streamwise and the tangential components exist in a predominant manner, while the radial component has a small scale distribution and shows the inflection which its flow direction is inward or outward. Moreover, the turbulent intensity profiles show that the radial component exists the most greatly among turbulent energies.

• 한국정보통신학회논문지
• /
• 제9권4호
• /
• pp.897-904
• /
• 2005

영상 인식은 영상획득이 용이하다는 것과 실생활에서 광범위하게 사용될 수 있다는 것으로 인해 활발하게 연구되고 있는 분야이다. 그러나 얼굴영상은 높은 차원의 영상공간으로 인해 이미지 처리가 쉽지 않다. 본 논문은 얼굴 영상 데이터의 차원을 특징적인 벡터로 표현하고 이러한 특징벡터를 통해 얼굴 영상을 인식하는 방법은 제안한다. 제안되는 알고리즘은 두 부분으로 나뉜다. 첫째로는 칼라 영상을 그레이 영상으로 변환할 때 RGB 세 개의 플레인의 평균이 아닌 세 플레인의 주성분을 사용하는 PCA(Principal Component Analysis)를 적용한다. PCA는 칼라 영상을 그레이 영상으로 변환하는 과정과 인식률을 높이기 위한 영상 대비 개선 과정이 동시에 수행한다. 두 번째로는 PCA와 LDA(Linear Discriminant Analysis) 방식을 하나의 과정으로 통합하는 개선된 통합 LDA 방법이다. 두 과정을 통합함으로서 간결한 알고리즘 표현이 가능하며 분리된 단계에서 있을 수 있는 정보 손실을 방지할 수 있다. 제안된 알고리즘은 잘 제어된 대용량 얼굴 데이터베이스에서 개인을 확인하는 분야에 적용되어 성능을 향상시키고 있음을 보여주었고, 추후에는 실시간 상황에서 특정 개인을 확인하는 분야의 기초 알고리즘으로 적용될 수 있다.

• 한국통신학회논문지
• /
• 제35권9C호
• /
• pp.748-755
• /
• 2010

본 논문은 가버 텐서(Gabor tensor)를 이용한 얼굴인식 시스템을 제안하였다. 가버 변환은 얼굴 고유의 특징을 잘 나타내주며 외부적인 영향을 줄일 수 있어 인식률 향상에 기여한다. 이러한 특징을 이용한 3차원의 텐서를 구성하여 얼굴인식을 수행하는 방법을 제안한다. 3차원의 가버 텐서를 입력으로 하여 기존의 1차원이나 2차원 주성분 분석법(PCA)보다 다양한 특징을 이용할 수 있는 다중선형 주성분 분석법(Multilinear PCA)를 수행한 다음 선형 판별법(LDA)을 수행하는 얼굴인식 방법을 제안하였다. 이러한 방법들은 표정이나 조명등의 변화에 강인한 특성을 가진다. 제안한 방법은 매트랩(Matlab)을 이용하여 실험하였다. ORL과 Yale 데이터베이스를 이용한 실험 결과를 기존의 방법들과 비교하였을 경우 제안한 방법이 기본적인 1차원 주성분 분석법보다 최대 9~27% 향상된 우수한 인식성능을 나타냄을 확인할 수 있었다.