Face recognition via Weighted Sparse Representation

Lu, Canyi; Min, Hai; Gui, Jie; Zhu, Lin; Lei, Yingke

doi:10.1016/j.jvcir.2012.05.003

Cited by 218 publications

(95 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…The SRC method is based on solving a P 0 problem. In order to have better performance [9] suggests to solve the following P 1,w problem (an 1 relaxation of P 0,w )…”

Section: An Application Examplementioning

confidence: 99%

“…Since [9] uses the term WSRC for classification based on weighted 1 norm minimization, we use the term WSRC0 for classification based on weighted 0 norm minimization. In this section, we propose a new weight vector (which has some negative components) and we solve the P 0,w problem with this new weight vector in order to show its superiority with respect to the SRC and WSRC0 methods with positive weights.…”

Section: An Application Examplementioning

confidence: 99%

See 1 more Smart Citation

A generalization of weighted sparse decomposition to negative weights

Delfi

Aziznejad

Amani

et al. 2017

2017 25th European Signal Processing Conference (EUSIPCO)

View full text Add to dashboard Cite

Abstract-Sparse solutions of underdetermined linear systems of equations are widely used in different fields of signal processing. This problem can also be seen as a sparse decomposition problem. Traditional sparse decomposition gives the same priority to all atoms for being included in the decomposition or not. However, in some applications, one may want to assign different priorities to different atoms for being included in the decomposition. This results to the so called "weighted sparse decomposition" problem [Babaie-Zadeh et al. 2012]. However, Babaie-Zadeh et al. studied this problem only for positive weights; but in some applications (e.g. classification) better performance can be obtained if some weights become negative. In this paper, we consider "weighted sparse decomposition" problem in its general form (positive and negative weights). A tight uniqueness condition and some applications for the general case will be presented.

show abstract

“…The SRC method is based on solving a P 0 problem. In order to have better performance [9] suggests to solve the following P 1,w problem (an 1 relaxation of P 0,w )…”

Section: An Application Examplementioning

confidence: 99%

Section: An Application Examplementioning

confidence: 99%

A generalization of weighted sparse decomposition to negative weights

Delfi

Aziznejad

Amani

et al. 2017

2017 25th European Signal Processing Conference (EUSIPCO)

View full text Add to dashboard Cite

show abstract

“…The shared dictionary learning method usually uses all training samples to obtain a classification dictionary. Lu et al [18] proposed a locality weighted sparse representation based classification (WSRC) method which utilizes both data locality and linearity to train a classification dictionary. Yang et al [19] proposed a novel dictionary learning method based on the Fisher discrimination criterion to improve the pattern classification performance.…”

Section: Introductionmentioning

confidence: 99%

Laplace Graph Embedding Class Specific Dictionary Learning for Face Recognition

Wang

Liu

2018

Journal of Electrical and Computer Engineering

View full text Add to dashboard Cite

The sparse representation based classification (SRC) method and collaborative representation based classification (CRC) method have attracted more and more attention in recent years due to their promising results and robustness. However, both SRC and CRC algorithms directly use the training samples as the dictionary, which leads to a large fitting error. In this paper, we propose the Laplace graph embedding class specific dictionary learning (LGECSDL) algorithm, which trains a weight matrix and embeds a Laplace graph to reconstruct the dictionary. Firstly, it can increase the dimension of the dictionary matrix, which can be used to classify the small sample database. Secondly, it gives different dictionary atoms with different weights to improve classification accuracy. Additionally, in each class dictionary training process, the LGECSDL algorithm introduces the Laplace graph embedding method to the objective function in order to keep the local structure of each class, and the proposed method is capable of improving the performance of face recognition according to the class specific dictionary learning and Laplace graph embedding regularizer. Moreover, we also extend the proposed method to an arbitrary kernel space. Extensive experimental results on several face recognition benchmark databases demonstrate the superior performance of our proposed algorithm.

show abstract

“…[1][2][3] In the past decades, various face recognition methods have been proposed, in which the representation based classification (RC) methods have attracted a great amount of attention owing to its good performance in reducing the negative influence of noises and occlusions. [4][5][6][7][8][9][10] Sparse representation based classification (SRC), [11][12][13][14] linear regression classification (LRC), [15] and collaborative representation classification (CRC) [16] are the most well-known RC methods. These RC methods are all based on the assumption that each sample can be represented well by the joint linear combination of few samples from its own subspace and all use the representation residuals of every class for classification.…”

Section: Introductionmentioning

confidence: 99%

The fusion of original and symmetric virtual images for image preprocessing in face recognition and collaborative representation based classification

Liu

2017

AIR

View full text Add to dashboard Cite

Various poses, facial expressions and illuminations are the biggest challenges in the fields of face recognition. To overcome these challenges, we propose a simple yet novel method in this paper by using the approximately symmetrical virtual face. Firstly, based on the symmetrical characteristics of faces, we present the method to generate the virtual faces for all samples in detail. Secondly, the collaborative representation based classification method is performed on both of the original set and virtual set individually. In this way, two kinds of representation residuals of every class can be obtained. Thirdly, an adaptive weighted fusion approach is presented and utilized to integrate those two kinds of representation residuals for face recognition. Lastly, we can obtain the label of the test sample by assigning it to the class with the minimum fused residual. Several experiments are conducted which show that the proposed method not only can greatly improve the classification accuracy, but also can effectively reduce the negative influence of various poses, illuminations, and facial expressions.

show abstract

Face recognition via Weighted Sparse Representation

Cited by 218 publications

References 19 publications

A generalization of weighted sparse decomposition to negative weights

A generalization of weighted sparse decomposition to negative weights

Laplace Graph Embedding Class Specific Dictionary Learning for Face Recognition

The fusion of original and symmetric virtual images for image preprocessing in face recognition and collaborative representation based classification

Contact Info

Product

Resources

About