Two-class Linear Discriminant Analysis for Face Recognition

Ekenel, Hazım Kemal; Stiefelhagen, Rainer

doi:10.1109/siu.2007.4298761

Cited by 7 publications

(3 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The most common classification algorithms include linear discriminate classifier [22], nearest‐neighbour classifiers [23], neural networks (NNs) [24] and support vector machines (SVMs) [25]. Our experiments show the considerable efficiency of using some well‐known classifiers such as SVM in gender classification of adult faces.…”

Section: Proposed Methodsmentioning

confidence: 93%

Gender classification based on fuzzy clustering and principal component analysis

Hassanpour

Zehtabian

Nazari

et al. 2016

IET Computer Vision

View full text Add to dashboard Cite

Gender classification is one of the most challenging problems in computer vision. Facial gender detection of neonates and children is also known as a highly demanding issue for human observers. This study proposes a novel gender classification method using frontal facial images of people. The proposed approach employs principal component analysis (PCA) and fuzzy clustering technique, respectively, for feature extraction and classification steps. In other words, PCA is applied to extract the most appropriate features from images as well as reducing the dimensionality of data. The extracted features are then used to assign the new images to appropriate classes – male or female – based on fuzzy clustering. The computational time and accuracy of the proposed method are examined together and the prominence of the proposed approach compared to most of the other well‐known competing methods is proved, especially for younger faces. Experimental results indicate the considerable classification accuracies which have been acquired for FG‐Net, Stanford and FERET databases. Meanwhile, since the proposed algorithm is relatively straightforward, its computational time is reasonable and often less than the other state‐of‐the‐art gender classification methods.

show abstract

Section: Proposed Methodsmentioning

confidence: 93%

Gender classification based on fuzzy clustering and principal component analysis

Hassanpour

Zehtabian

Nazari

et al. 2016

IET Computer Vision

View full text Add to dashboard Cite

show abstract

“…Hence, the improvement on basic LDA improves the performance in facial recognition system. Nevertheless, the work of [28] proved otherwise. They reported that other two classes of LDA outperformed the multi-class LDA.…”

Section: Discussion Based On Performance Metricsmentioning

confidence: 99%

A Comparative Analysis of Selected Fisher Linear Discriminant Based Algorithms in Human Faces

Okediran

Ashaolu

Omidiora

2019

JAMCS

View full text Add to dashboard Cite

One of the most reliable biometrics when issues of access control and security is been considered is face recognition. An integral part of a face recognition system is the feature extraction stage, which becomes a critical problem where is a need to obtain the best feature with minimum classification error and low running time. Many of the existing face recognition systems have adopted different linear discriminantbased algorithms independently for feature extraction in which excellent performance were achieved, but identifying the best most suitable of these variants of linear discriminant-based algorithms for face recognition systems remains a subject open for research. Therefore, this paper carried out a comparative analysis of the performance of the basic Linear Discriminant Algorithm (LDA) and two of its variants which are Kernel Linear Discriminant Analysis (KLDA) and Multiclass Linear Discriminant Analysis (MLDA) in face recognition application for access control. Three Hundred and forty (340) face images were locally acquired with default size of 1200 x 1200. Two hundred and forty (240) images were used for training while the remaining hundred (100) images were used for testing purpose. The image enhancement involves converting into grayscale and normalizing the acquired images using histogram equalization method. Feature extraction and dimension reduction of the Original Research ArticleArticle Okediran et al.; JAMCS,33(4): 1-19, 2019; Article no.JAMCS.50647 2 images were done using each of LDA, KLDA and MLDA algorithms individually. The extracted feature subsets of the images from each of LDA, KLDA and MLDA algorithm were individually classified using Euclidian distance. This technique was implemented using Matrix Laboratory (R2015a). The performance of LDA, KLDA and MLDA was evaluated and compared at 200 x 200 pixel resolution and 0.57 threshold value using recognition accuracy, sensitivity, specificity, false positive rate, training time and recognition time. The evaluation result shows that the LDA algorithm yielded recognition accuracy, sensitivity, specificity, false positive rate, training time and recognition time of 93.00%, 92.86%, 93.33%, 6.67%, 1311.76 seconds and 67.98 seconds respectively. Also, KLDA recorded recognition accuracy, sensitivity, specificity, false positive rate, training time and recognition time of 95.00%, 95.71%, 93.33%, 6.67%, 1393.24 seconds and 63.67 seconds respectively. Furthermore, MLDA algorithm yielded recognition accuracy, sensitivity, specificity, false positive rate, training time and recognition time of 97.00%, 97.14%, 96.67%, 3.33%, 1191.55 seconds and 58.65 seconds respectively. The t-test measured betweenthe accuracies of MLDA algorithm and KLDA reveals that MLDA algorithm was statistically significant at < 0.05; = 0.014 = 1.50 . Also, the t-test measured between the accuracies of MLDA algorithm and LDA reveals that MLDA algorithm was statistically significant at < 0.01; = 0.001 = 3.75.

show abstract

“…proposed two-class LDA method[10], denoted as 2CLDA_1. It classifies test samples by confidence score.…”

mentioning

confidence: 99%

Image features optimizing for content-based image retrieval

Shi

Liu

et al. 2009

2009 IEEE International Conference on Intelligent Computing and Intelligent Systems

View full text Add to dashboard Cite

Developing low-dimensional semantics-sensitive features is crucial for content-based image retrieval (CBIR). In this paper, we present a method called M2CLDA (merging 2class linear discriminant analysis) to capture low-dimensional optimal discriminative features in the projection space. M2CLDA calculates discriminant vectors with respect to each class in the one-vs-all classification scenario and then merges all the discriminant vectors to form a projection matrix. The dimensionality of the M2CLDA space fits in with the number of classes involved. Moreover, when a new class is added, the new M2CLDA space can be approximated by only calculating a new discriminant vector for the new class. The features in the M2CLDA space have better semantic discrimination than those in traditional LDA space. Our experiments show that the proposed approach improves the performance of image retrieval and image classification dramatically.

show abstract

Two-class Linear Discriminant Analysis for Face Recognition

Cited by 7 publications

References 8 publications

Gender classification based on fuzzy clustering and principal component analysis

Gender classification based on fuzzy clustering and principal component analysis

A Comparative Analysis of Selected Fisher Linear Discriminant Based Algorithms in Human Faces

Image features optimizing for content-based image retrieval

Contact Info

Product

Resources

About