Document Type : Research Article


1 School of Computing and Engineering, University of Huddersfield, Huddersfield, England Department of information Technology, University of Human Development, Sulaymaniyah, Iraq

2 School of Computing and Engineering, University of Huddersfield, Huddersfield, England

3 Department of information Technology, University of Human Development, Sulaymaniyah, Iraq


Standard face recognition algorithms that use standard feature extraction techniques always suffer from image performance degradation. Recently, singular value decomposition and low-rank matrix are applied in many applications,
including pattern recognition and feature extraction. The main objective of this research is to design an efficient face recognition approach by combining many techniques to generate efficient recognition results. The implemented face
recognition approach is concentrated on obtaining significant rank matrix via applying a singular value decomposition technique. Measures of dispersion are used to indicate the distribution of data. According to the applied ranks, there
is an adequate reasonable rank that is important to reach via the implemented procedure. Interquartile range, mean absolute deviation, range, variance, and standard deviation are applied to select the appropriate rank. Rank 24, 12, and 6
reached an excellent 100% recognition rate with data reduction up to 2 : 1, 4 : 1 and 8 : 1 respectively. In addition, properly selecting the adequate rank matrix is achieved based on the dispersion measures. Obtained results on standard face databases verify the efficiency and effectiveness of the implemented approach.


[1] Gurjit Singh Walia, Tarandeep Singh, Kuldeep Singh, and Neelam Verma. Robust multimodal biometric system based on optimal score level (a) Standard Deviation Values (b) Standard Deviation Percentages Figure 16. Standard Deviation Measures fusion model. Expert Systems with Applications,116:364–376, 2019.
[2] Ahmed Mahfouz, Tarek M Mahmoud, and Ahmed Sharaf Eldin. A survey on behavioral biometric authentication on smartphones. Journal of information security and applications, 37:28–37, 2017.
[3] Ajita Rattani and Reza Derakhshani. A survey of mobile face biometrics. Computers & Electrical Engineering, 72:39–52, 2018.
[4] Pedro Tome, Ruben Vera-Rodriguez, Julian Fierrez, and Javier Ortega-Garcia. Facial soft biometric features for forensic face recognition. Forensic science international, 257:271–284, 2015.
[5] Abdullah Mohammad Awad and Muzhir Shaban AL-Ani. An efficient approach for image hiding using hybrid technique. Journal of Theoretical & Applied Information Technology, 96(11), 2018.
[6] Zhiming Liu and Chengjun Liu. Fusion of color, local spatial and global frequency information for face recognition. Pattern Recognition, 43(8):2882–2890, 2010.
[7] R Satya Prasad, Muzhir Shaban Al-Ani, and Salwa Mohammed Nejres. An efficient approach for fingerprint recognition. Image, 4(5), 2015.
[8] Alaa Hussein Al-Hamami. Handbook of Research on Threat Detection and Countermeasures in Network Security. IGI Global, 2014.
[9] Haifeng Hu. Variable lighting face recognition using discrete wavelet transform. Pattern Recognition Letters, 32(13):1526–1534, 2011.
[10] Guangwei Gao, Zangyi Hu, Pu Huang, Meng Yang, Quan Zhou, Songsong Wu, and Dong Yue. Robust low-resolution face recognition via lowrank representation and locality-constrained regression. Computers & Electrical Engineering,
70:968–977, 2018.
[11] Nan Zhang, Yi Chen, Maolong Xi, Fangqin Wang, and Yanwen Qu. Feature extraction based on low-rank affinity matrix for biological recognition. Journal of computational science, 27:199–205, 2018.
[12] Baiyu Chen, Zi Yang, and Zhouwang Yang. An algorithm for low-rank matrix factorization and its applications. Neurocomputing, 275:1012–1020,2018.
[13] Jicong Fan and Tommy W.S. Chow. Matrix completion by least-square, low-rank, and sparse self-representations. Pattern Recognition, 71:290– 305, 2017.
[14] Hengmin Zhang, Jian Yang, Jianchun Xie, Jianjun Qian, and Bob Zhang. Weighted sparse coding regularized nonconvex matrix regression for robust face recognition. Information Sciences,394:1–17, 2017.
[15] Lunke Fei, Yong Xu, Bob Zhang, Xiaozhao Fang, and JieWen. Low-rank representation integrated with principal line distance for contactless palmprint recognition. Neurocomputing, 218:264–275,2016.
[16] LiÅcÄC, LÄCcrÄCmioara, and Elena Pelican. A low-rank tensor-based algorithm for face recognition. Applied Mathematical Modelling,39(3):1266 – 1274, 2015.
[17] Jian Zhang, Jian Yang, Jianjun Qian, and Jiawei Xu. Nearest orthogonal matrix representation for face recognition. Neurocomputing, 151:471 –480, 2015.
[18] Haipeng Shen and Jianhua Z. Huang. Sparse principal component analysis via regularized low rank matrix approximation. Journal of Multivariate Analysis, 99(6):1015 – 1034, 2008.
[19] Nojun Kwak. Feature extraction for classification problems and its application to face recognition. Pattern Recognition, 41(5):1701 – 1717, 2008.
[20] Frank B. ter Haar and Remco C. Veltkamp. Expression modeling for expression-invariant face recognition. Computers & Graphics, 34(3):231 –241, 2010. Shape Modelling International (SMI)Conference 2010.
[21] Semih Ergin, Serdar ÃCâAaakir, ÃCâAS. Nezih Gerek, and M. Bilginer GÃC¼lmezoÃDÅÿlu. A new implementation of common matrix approach using third-order tensors for face recognition. Expert Systems with Applications, 38(4):3246 –
3251, 2011.
[22] M.ÃC¼ge ÃCâAaarÃDÂskÃCÂgÃDÂs and Figen ÃCâASzen. A face recognition system based on eigenfaces method. Procedia Technology, 1:118– 123, 2012. First World Conference on Innovation and Computer Sciences (INSODE 2011).
[23] Eslam Mostafa, Riad Hammoud, Asem Ali, and Aly Farag. Face recognition in low resolution thermal images. Computer Vision and Image Understanding, 117(12):1689 – 1694, 2013.
[24] Si-Bao Chen, Chris H.Q. Ding, and Bin Luo. Extended linear regression for undersampled face recognition. Journal of Visual Communication and Image Representation, 25(7):1800 – 1809,2014.
[25] Afshin Dehghan, Omar Oreifej, and Mubarak Shah. Complex event recognition using constrained low-rank representation. Image and Vision Computing, 42:13 – 21, 2015.
[26] Chun-Hou Zheng, Yi-Fu Hou, and Jun Zhang. Improved sparse representation with low-rank representation for robust face recognition. Neurocomputing, 198:114 – 124, 2016. Advances in Neural Networks, Intelligent Control and Information Processing.
[27] Guangwei Gao, Jian Yang, Xiao-Yuan Jing, Fumin Shen, Wankou Yang, and Dong Yue. Learning robust and discriminative low-rank representations for face recognition with occlusion. Pattern Recognition, 66:129 – 143, 2017.
[28] Cho Ying Wu and Jian Jiun Ding. Occluded face recognition using low-rank regression with generalized gradient direction. Pattern Recognition, 80:256 – 268, 2018.
[29] Peiguang Jing, Yuting Su, Zhengnan Li, Jing Liu, and Liqiang Nie. Low-rank regularized tensor discriminant representation for image set classification. Signal Processing, 156:62 – 70, 2019.
[30] Yegang Hu, Yuping Wang, and Jicong Zhang. Low-rank matrix recovery for source imaging with magnetoencephalography. Optics & Laser Technology, 110:99 – 104, 2019. Special Issue: Optical Imaging for Extreme Environment.
[31] Shanhua Zhan, JigangWu, Na Han, JieWen, and Xiaozhao Fang. Unsupervised feature extraction by low-rank and sparsity preserving embedding. Neural Networks, 109:56 – 66, 2019.
[32] Duc-Thuan Vo and Ebrahim Bagheri. Featureenriched matrix factorization for relation extraction. Information Processing & Management, 56(3):424 – 444, 2019.
[33] Qiegen Liu, Sanqian Li, Jing Xiao, and Minghui Zhang. Multi-filters guided low-rank tensor coding for image inpainting. Signal Processing: Image Communication, 73:70 – 83, 2019. Tensor Image Processing.
[34] Jianchun Xie, Jian Yang, Jianjun Qian, and Lei Luo. Bi-weighted robust matrix regression for face recognition. Neurocomputing, 237:375 – 387, 2017.
[35] Yixin Yang, Jianqi Zhang, Delian Liu, and Xin Wu. Low-rank and sparse matrix decomposition with background position estimation for hyperspectral anomaly detection. Infrared Physics & Technology, 96:213 – 227, 2019.
[36] Pranay Dighe, Afsaneh Asaei, and HervÃCÂl Bourlard. Low-rank and sparse subspace modeling of speech for dnn based acoustic modeling. Speech Communication, 109:34 – 45, 2019.
[37] Hengmin Zhang, Jian Yang, Jianjun Qian, and Wei Luo. Nonconvex relaxation based matrix regression for face recognition with structural noise and mixed noise. Neurocomputing, 269:188– 198, 2017.
[38] Zhen Dong, Mingtao Pei, and Yunde Jia. Orthonormal dictionary learning and its application to face recognition. Image and Vision Computing, 51:13 – 21, 2016.
[39] Aihua Zheng, Tian Zou, Yumiao Zhao, Bo Jiang, Jin Tang, and Chenglong Li. Background subtraction with multi-scale structured low-rank and sparse factorization. Neurocomputing, 328:113– 121, 2019. Chinese Conference on Computer
Vision 2017.
[40] Qiang Zhang, Zhen Huo, Yi Liu, Yunhui Pan, Caifeng Shan, and Jungong Han. Salient object detection employing a local tree-structured lowrank representation and foreground consistency. Pattern Recognition, 92:119 – 134, 2019.
[41] Yunfang Fu, Qiuqi Ruan, Ziyan Luo, Yi Jin, Gaoyun An, and Jun Wan. Ferlrtc: 2d+3d facial expression recognition via low-rank tensor completion. Signal Processing, 161:74 – 88, 2019.
[42] Taisong Jin, Jun Yu, Jane You, Kun Zeng, Cuihua Li, and Zhengtao Yu. Low-rank matrix factorization with multiple hypergraph regularizer. Pattern Recognition, 48(3):1011 – 1022, 2015.
[43] Junhong Lin and Song Li. Convergence of projected landweber iteration for matrix rank minimization. Applied and Computational Harmonic Analysis, 36(2):316 – 325, 2014.
[44] Xiao Luan, Bin Fang, Linghui Liu, Weibin Yang, and Jiye Qian. Extracting sparse error of robust pca for face recognition in the presence of varying illumination and occlusion. Pattern Recognition, 47(2):495 – 508, 2014.
[45] Miaoyun Zhao, Licheng Jiao, Jie Feng, and Tianyu Liu. A simplified low rank and sparse graph for semi-supervised learning. Neurocomputing, 140:84 – 96, 2014.
[46] Wei Han, Christian Sorg, Changgang Zheng, Qinli Yang, Xiaosong Zhang, Arvid Ternblom, Cobbinah Bernard Mawuli, Lianli Gao, Cheng Luo, Dezhong Yao, Tao Li, Sugai Liang, and Junming Shao. Low-rank network signatures in the triple network separate schizophrenia and major depressive disorder. NeuroImage: Clinical, 22:101725, 2019.
[47] Shengxiang Gao, Zhengtao Yu, Taisong Jin, and Ming Yin. Multi-view low-rank matrix factorization using multiple manifold regularization. Neurocomputing, 335:143 – 152, 2019.
[48] Jinghua Wang, Jane You, Qin Li, and Yong Xu. Orthogonal discriminant vector for face recognition across pose. Pattern Recogn., 45(12):4069–4079, December 2012.
[49] Binbin Pan, Jianhuang Lai, and Pong C. Yuen. Learning low-rank mercer kernels with fastdecaying spectrum. Neurocomput., 74(17):3028–3035, October 2011.