Yujie Zheng

Driver Fatigue Detection: A Survey

Driver fatigue is an important factor in a large number of accidents. There has been much work done in driver fatigue detection. This paper presents a comprehensive survey of research on driver fatigue detection and provides structural categories for the methods which have been proposed. The methods of fatigue detection mainly focused on measures of the drivers state, driver performance and the combination of the drivers state and performance. The measures of drivers state included PERCLOS, mouth shape and head position; the measures of driver performance included lane tracking and tracking of distance between vehicles. These approaches are presented and discussed in detail. Some typical driver monitoring systems are also introduced in this paper. Finally, summary and conclusions are presented

A complete fuzzy discriminant analysis approach for face recognition

In this paper, some studies have been made on the essence of fuzzy linear discriminant analysis (F-LDA) algorithm and fuzzy support vector machine (FSVM) classifier, respectively. As a kernel-based learning machine, FSVM is represented with the fuzzy membership function while realizing the same classification results with that of the conventional pair-wise classification. It outperforms other learning machines especially when unclassifiable regions still remain in those conventional classifiers. However, a serious drawback of FSVM is that the computation requirement increases rapidly with the increase of the number of classes and training sample size. To address this problem, an improved FSVM method that combines the advantages of FSVM and decision tree, called DT-FSVM, is proposed firstly. Furthermore, in the process of feature extraction, a reformative F-LDA algorithm based on the fuzzy k-nearest neighbors (FKNN) is implemented to achieve the distribution information of each original sample represented with fuzzy membership grade, which is incorporated into the redefinition of the scatter matrices. In particular, considering the fact that the outlier samples in the patterns may have some adverse influence on the classification result, we developed a novel F-LDA algorithm using a relaxed normalized condition in the definition of fuzzy membership function. Thus, the classification limitation from the outlier samples is effectively alleviated. Finally, by making full use of the fuzzy set theory, a complete F-LDA (CF-LDA) framework is developed by combining the reformative F-LDA (RF-LDA) feature extraction method and DT-FSVM classifier. This hybrid fuzzy algorithm is applied to the face recognition problem, extensive experimental studies conducted on the ORL and NUST603 face images databases demonstrate the effectiveness of the proposed algorithm.

Fuzzy Kernel Fisher Discriminant Algorithm with Application to Face Recognition

In this paper, a new kernel Fisher discriminant (KFD) algorithm with fuzzy set theory is studied. KFD algorithm is effective to extract nonlinear discriminative features of input samples with kernel trick. While conventional KFD algorithm assumes the same level of relevance of each sample to the corresponding class. In this paper, a novel KFD algorithm named fuzzy kernel Fisher discriminant (FKFD) is proposed. Distribution information of samples is represented with fuzzy membership degree in this paper. Furthermore, this information is utilized to redefine the corresponding scatter matrices, which are different to the conventional KFD algorithm and effective to extract discriminative features from overlapping (outlier) samples. Experimental results on ORL face database demonstrate the effectiveness of the proposed method

Letters: A reformative kernel Fisher discriminant algorithm and its application to face recognition

In this paper, a reformative kernel Fisher discriminant (KFD) algorithm with fuzzy set theory is studied. The KFD algorithm is effective to extract nonlinear discriminative features of input samples using the kernel trick. However, the conventional KFD algorithm assumes the same level of relevance of each sample to the corresponding class. In this paper, a fuzzy kernel Fisher discriminant (FKFD) algorithm is proposed. Distribution information of samples is represented with fuzzy membership degree and this information is utilized to redefine corresponding scatter matrices which are different to the conventional KFD algorithm and effective to extract discriminative features from overlapping (outlier) samples. Experimental results on the ORL face database demonstrate the effectiveness of the proposed method.

Discriminant analysis approach using fuzzy fourfold subspaces model

In this paper, some studies have been made on the essence of a novel fuzzy discriminant analysis (FDA) on the fourfold-objective model (FOM). First, a fourfold-objective model on the discriminant analysis is developed, by which a set of integrated subspaces derived from within-class and between-class scatter matrices are constructed, respectively. Second, an improved FDA (IFDA) algorithm based on the relaxed normalized condition is proposed to achieve the distribution information of each sample represented with fuzzy membership grade, which is incorporated into the redefinition of Fishers scatter matrices. Therefore, the presented algorithm has the potential to outperform the traditional subspace learning algorithms, especially in the cases of small sample size. Experimental results conducted on the ORL, NUST603, FERET and Yale face image databases demonstrate the effectiveness of the proposed method.

Effective classification image space which can solve small sample size problem

Linear discriminant analysis (LDA) is one of the most popular methods in feature extraction and dimension reduction. However, in many real applications, particularly in image recognition applications such as face recognition, conventional LDA algorithm will often encounter small sample size problem. In this paper, an effective classification image space is defined and optimal features are extracted from this space. With the proposed method, an effective classification image space of each original image is first obtained. Then, optimal features are extracted from this space. The small sample size problem is solved effectively with the proposed method. Experimental results on XM2VTS face database demonstrate the effectiveness of the proposed method

A Complete and Rapid Feature Extraction Method for Face Recognition

Feature extraction is one of the key steps in face recognition. In this paper, common vector is used to extract features from null space of within-class scatter matrix, which is independent of the sample index in the same class and accelerates the speed of feature extraction. Furthermore, effective features in regular space are extracted to enhance the performance of face recognition. The proposed method not only solves the small sample size problem, but also extracts more effective features from face images. Experimental results on two popular databases demonstrate the effectiveness of the proposed method

Fuzzy kernel discriminant analysis (FKDA) and its application to face recognition

A fuzzy kernel diccriminant analysis algorithm (FKDA) is proposed in this paper, which is the kernel version of the fuzzy fisherface method. First, KPCA is performed on the training data. Then fuzzy k-nearest neighbor (FKNN) is introduced to find the mean vectors of each class. Fuzzy scatter matrices are derived for fuzzy LDA in the kernel space. The results of experiments conducted on ORL database show that the proposed method is better than fuzzy fisherface method in terms of accurate recognition rate.

A Study on An Improved Algorithm of Self-Adaptive Clustering Network

A study has been made on the algorithm of adaptive clustering network. The fact that different feature component has different function has not been considered in the algorithm of adaptive clustering network. The weight of every feature component has been considered in obtaining winning node and vigilance test of it. The weighted distance has been introduced for the patterns. An improved algorithm of adaptive clustering network has been proposed based on the above considerations. We have made experiments on Andersons data and singular value features of ORL image base respectively. The experimental results show that both effectiveness and adaptiveness of the proposed algorithm has been improved

Feature extraction by structured stepwise nonparametric maximum margin criterion

In this paper, a new feature extraction method named structured stepwise nonparametric maximum margin criterion (SSNMMC) is proposed. Previous nonparametric discriminant analysis methods only use the point-to-point distance to measure class difference. In the proposed method, point-to-line distance with nearest neighbor line (NNL) theory is adopted and more intrinsic structure information of training samples is preserved in the feature space. Furthermore, the proposed method does not assume that the class densities belong to any particular parametric family and does not depend on the nonsigularity of the within-class scatter matrix, which are shortcomings of conventional linear discrimiant analysis based algorithms. Besides, limitation of feature number is overcome with the proposed method. Experiments on the ORL face database demonstrate the effectiveness of our proposed method.