Elsevier

Optics & Laser Technology

Volume 110, February 2019, Pages 13-23
Optics & Laser Technology

Full length article
Iris feature extraction through wavelet mel-frequency cepstrum coefficients

https://doi.org/10.1016/j.optlastec.2018.03.002Get rights and content

Highlights

  • A wavelet cepstrum based feature extraction technique is proposed for iris recognition.

  • The technique helps in emphasizing the high frequency components for achieving better recognition accuracy.

  • The wavelet cepstrum feature helps in dimensionality reduction of the feature set.

  • The validity of the concept is verified by detailed experiments.

Abstract

In this paper, a novel technique based on wavelet cepstrum feature is discussed for iris recognition system. The proposed method is based on the wavelet derived from the popular biorthogonal Cohen-Daubechies-Feauveau 9/7 filter bank. Moreover, being biorthogonal in nature it has superior frequency selectivity, symmetric, and better time-frequency localization. The suggested scheme deals with computing the two level detail coefficients from the normalized iris template. Then these detailed coefficients are then divided into non-uniform bins in a logarithmic manner. This helps in reducing the dimension of the wavelet coefficients followed by assigning non-uniform weights to the different frequency components. Then the discrete cosine transform of the same is computed, from which the energy feature is extracted. The proposed technique is experimentally validated with publicly available databases: CASIAv3, UBIRISv1, and IITD. The performance of the proposed approach is found be superior to that of the state-of-the-art methods.

Introduction

Biometrics is the science that deals with recognition of an individual by considering its different traits. These traits can be either behavioral or physical. In cybernetics, authentication is a fundamental and important activity. Biometrics has found its place ranging from modest to expansive applications. Since automation of many day-to-day activities in public places are achieved through softwares, biometrics based person identification has provided a platform for ensuring security in airports, making financial transactions in banks, and surveillance [1]. The level of difficulty in identity verification increases manifold, when it is designed with high accuracy. Person authorization and authentication since the last two decades has been predominantly carried out by: (i) token-based system (e.g., passport, smart card, etc.); (ii) knowledge-based system(e.g., PIN, password, etc.); and (iii) biometric based system (e.g., face, signature, etc.). To deal with spoofing effectively, a combination of two or more of these, referred to as multifactor authentication, has become the preferred way of authenticating users in recent years.

Password/PIN is the frequently used authentication/authorization technique, whose usage is widely found in personal computers, automated teller machines (ATM) for making financial transactions. The need to physically possess the token/piece of information makes the biggest disadvantage of the token/knowledge based authentication system. Hence, both are prone to spoofing attack. On the contrary, biometrics has a unique advantage, where the user authenticates himself by using his/her own physical traits. These unique biometric traits include fingerprint, iris, face, palmprint, etc. The fingerprint/palmprint is prone to changes due to external injury or change in pattern due to rough work. Similarly, face feature is not stable as it changes over time. On the contrary, the iris pattern is not exposed to the outside environment. It is covered by a very transparent material called cornea leaving the iris protected (see Fig. 1). Apart from that the iris pattern remain stable over the years. Also it is not affected by cataract surgery or any other disease. Hence, iris is known to be highly reliable due to the above said reasons among all these biometric features. A block diagram representation of iris recognition system is shown in Fig. 2.

The rest of the paper is organized as follows: Section 2 describes about some notable contributions towards iris recognition followed by Section 3, which discusses about the proposed technique. Section 4 deals with the application of wavelet cepstrum feature extraction from iris images. Result and discussion is given in Section 5 followed by concluding remark in Section 6.

Section snippets

Related work

Numerous contributions on iris segmentation have been reported in the last decade. Iris localization carried out by employing integrodifferential operator is a widely adopted method, which is invented by Daugman [2]. This technique globally searches for the circle center and radius. The time complexity of this technique amounts to cubic order. Subsequently Wildes has developed another landmark method by using primitive edge detection followed by Circular Hough Transform (CHT) [3] for achieving

Proposed wavelet cepstrum feature for iris recognition

Mel-cepstral analysis is one of the most widely used feature extraction technique in speech processing applications including sound recognition and speaker identification. Two-dimensional (2D) cepstrum Cepstrum is also used in image registration and filtering applications. To the best of our knowledge 2D mel-cepstrum which is a variant of 2D cepstrum is rarely studied in image feature extraction and classification problems [51]. The goal of this contribution is to study the wavelet 2D

Proposed WCF for iris recognition

In the literature, various successful applications of cepstrum based features are found [52]. Mostly, it is applied widely in speech processing techniques. Apart from that, it has been used to achieve recognition in face biometrics with high accuracy. A unique representation, known as mel-frequency cepstrum (MFC), is adopted for speech processing. To convert a spectrum of f Hz into m mel, the following equation is used:m=1127.01048ln1+f700

The mel-frequency cepstrum coefficients (MFCC), which

Results and discussion

The following section discusses the results obtained applying the proposed scheme on few publicly available datasets. This section initially briefs the properties of the databases, marks the parameters used for evaluation of system. Subsequently this section analyses and compares the proposed method against few landmark methods to establish the claim of suitability of the proposed method.

Conclusion

Feature extraction plays a vital role in iris recognition. This paper suggests the application of wavelet mel-cepstrum for iris feature extraction. The MFCCs have been successfully used to characterize the iris texture. To validate the efficacy of the proposed method, simulation has ben carried out on three standard different databases: CASIAv3, UBIRISv1, and IITD. The performance of the same is compared with some existing techniques. The limitation of the proposed technique is that, with

Soubhagya Sankar Barpanda is an Assistant Professor at the Department of Computer Science & Engineering, VIT Amaravati University, India. He received his doctoral degree from Department of Computer Science & Engineering, National Institute of Technology Rourkela, India. He has completed his M. Tech. degree from the same institute. His research interests include biometric security and classical image processing.

References (59)

  • S. Bakshi, T. Tuglular, Security through human-factors and biometrics, in: Proceedings of the 6th International...
  • R. Wildes

    Iris recognition: an emerging biometric technology

    Proc. IEEE

    (1997)
  • Y.-P. Huang, S.-W. Luo, E.-Y. Chen, An efficient iris recognition system, in: International Conference on Machine...
  • Y. Liu, S. Yuan, X. Zhu, Q. Cui, A practical iris acquisition system and a fast edges locating algorithm in iris...
  • H. Sung, J. Lim, J. Park, Y. Lee, Iris recognition using collarette boundary localization, in: 17th International...
  • X. Liu, K. Bowyer, P. Flynn, Experiments with an improved iris segmentation algorithm, in: Fourth IEEE Workshop on...
  • X. Feng, C. Fang, X. Ding, Y. Wu, Iris localization with dual coarse-to-fine strategy, in: 18th International...
  • J. Koh, V. Govindaraju, V. Chaudhary, A robust iris localization method using an active contour model and Hough...
  • R. Sundaram, B. Dhara, B. Chanda, A fast method for iris localization, in: Second International Conference on Emerging...
  • R. Chen et al.

    Iris segmentation for non-cooperative recognition systems

    IET Image Proc.

    (2011)
  • S. Bakshi, H. Mehrotra, B. Majhi, Real-time iris segmentation based on image morphology, in: Proceedings of the 2011...
  • S.-J. Baek et al.

    Eyeball model-based iris center localization for visible image-based eye-gaze tracking systems

    IEEE Trans. Consum. Electron.

    (2013)
  • Y. Han et al.

    Efficient iris localisation using a guided filter

    IET Image Proc.

    (2015)
  • N. Reddy, A. Rattani, R. Derakhshani, A robust scheme for iris segmentation in mobile environment, in: Proceedings of...
  • J. Daugman

    High confidence visual recognition of persons by a test of statistical independence

    IEEE Trans. Pattern Anal. Mach. Intell.

    (1993)
  • T. Camus, R. Wildes, Reliable and fast eye finding in close-up images, in: 16th International Conference on Pattern...
  • B. Bonney, R. Ives, D. Etter, D. Yingzi, Iris pattern extraction using bit planes and standard deviations, in:...
  • H. Proenca et al.

    Iris segmentation methodology for non-cooperative recognition

    IEE Proc. Vision Image Signal Process.

    (2006)
  • S. Pundlik, D. Woodard, S. Birchfield, Non-ideal iris segmentation using graph cuts, in: IEEE Computer Society...
  • Cited by (25)

    • Dynamic iris inner edge location method based on distance-regularized level set

      2023, Optik
      Citation Excerpt :

      In order to roughly locate the pupils, the binarized and inverted images are subjected to vertical and horizontal projection operations, respectively. Barpanda applied this method in iris localization [22], but this method is only applicable to one iris bank, and the parameters need to be manually modified in different iris bank images. For this problem, the secondary projection method proposed in this paper is applied to this paper.

    • Iris feature extraction using three-level Haar wavelet transform and modified local binary pattern

      2022, Applications of Computational Intelligence in Multi-Disciplinary Research
    • Efficient iris recognition through curvelet transform and polynomial fitting

      2019, Optik
      Citation Excerpt :

      Vyas et al. [22] used difference of variance (DoV) features extracted from different partitions of the Gabor filtered iris images. In another recent work, Barpanda et al. [23] utilized mel-cepstral analysis of the iris images to extract two dimensional (2D) mel-cepstrum features, which have reduced dimension and preserved image structure, simultaneously. Kaur et al. [24] used discrete orthogonal moments like Tchebichef, Krawt-chouk and Dual-Hahn moments, to extract real-values features from iris images.

    • Wavelet-based protection strategy for series arc faults interfered by multicomponent noise signals in grid-connected photovoltaic systems

      2019, Solar Energy
      Citation Excerpt :

      The Dmey and Db family only have the orthogonality so that there are serious overlaps between arc fault and normal cycles in some frequency bands. The biorthogonal wavelets are established to solve the incompatibility between the symmetry and signal reconstruction precision (Barpanda et al., 2019). Unlike the Haar wavelet, the Rbio family are biorthogonal wavelets with the further lifting ability, which is benefit to improve the vanishing moment and regularity.

    View all citing articles on Scopus

    Soubhagya Sankar Barpanda is an Assistant Professor at the Department of Computer Science & Engineering, VIT Amaravati University, India. He received his doctoral degree from Department of Computer Science & Engineering, National Institute of Technology Rourkela, India. He has completed his M. Tech. degree from the same institute. His research interests include biometric security and classical image processing.

    Banshidhar Majhi is a Professor with the Department of Computer Science and Engineering, National Institute of Technology Rourkela, India. He is presently serving as Director of Indian Institute of Information Technology Design & Manufacturing Kancheepuram, India. He has successfully executed various Research and Development projects being funded by agencies such as MHRD, ISRO, DRDO, and DeitY. He has authored hundreds of articles in reputed journals and conferences. His current research interests include image processing, computer vision, biometric security, and pattern recognition. He has been conferred with prestigious awards and honors for his contribution towards scientific research and academic excellence.

    Pankaj Kumar Sa received the Ph.D. degree in computer science from NIT Rourkela in 2010. He is currently an Assistant Professor with the CSE Department, NIT Rourkela. He has authored or co-authored a number of research articles in various journals, conferences, and book chapters. His research interests include computer vision, biometrics, and visual surveillance. He is a member of the CSI. He has co-investigated some R&D projects funded by SERB, PXE, DeitY, and ISRO. He has been conferred with various prestigious awards and honors. Apart from research and teaching, he is also actively involved with the automation of NIT Rourkela, where he conceptualizes and engineers the automation process.

    Arun Kumar Sangaiah has received his Master of Engineering (ME) degree in Computer Science and Engineering from the Government College of Engineering, Tirunelveli, Anna University, India. He had received his Doctor of Philosophy (PhD) degree in Computer Science and Engineering from the VIT University, Vellore, India. He is presently working as an Associate Professor in School of Computer Science and Engineering, VIT University, India. His area of interest includes software engineering, computational intelligence, wireless networks, bio-informatics, and embedded systems. He has authored more than 100 publications in different journals and conference of national and international repute. His current research work includes global software development, wireless ad hoc and sensor networks, machine learning, cognitive networks and advances in mobile computing and communications. Also, he was registered a one Indian patent in the area of Computational Intelligence. Besides, Prof. Sangaiah is responsible for Editorial Board member/Associate Editor of various international journals.

    Sambit Bakshi is currently with Centre for Computer Vision and Pattern Recognition of National Institute of Technology Rourkela, India. He also serves as Assistant Professor in Department of Computer Science & Engineering of the institute. He earned his PhD degree in Computer Science & Engineering. He serves as associate editor of International Journal of Biometrics (2013 -), IEEE Access (2016 -), Innovations in Systems and Software Engineering (2016 -), Plos One (2017 -), and Expert Systems, Wiley (2018 -). He is technical committee member of IEEE Computer Society Technical Committee on Pattern Analysis and Machine Intelligence. He received the prestigious Innovative Student Projects Award–2011 from Indian National Academy of Engineering (INAE) for his master’s thesis. He has more than 50 publications in journals, reports, conferences.

    View full text