Adaptive Neuro-Fuzzy Inference System Based ...

Copyright © 2014 American Scientific Publishers All rights reserved Printed in the United States of America

Journal of Computational Intelligence and Electronic Systems Vol. 2, 1–7, 2014

Adaptive Neuro-Fuzzy Inference System Based Fingerprint and Retina Recognition Tripti Rani Borah1 ∗ , Kandarpa Kumar Sarma2 , and Pran Hari Talukdar3 1

2

Department of Computer Science, Gauhati University, Guwahati 781014, Assam, India Department of Electronics and Communication Technology, Gauhati University, Guwahati 781014, Assam, India 3 Department of Instrumentation and USIC, Gauhati University, Guwahati 781014, Assam, India

Fuzzy logic (FL) is a powerful problem solving methodology receiving wide spread acceptance for a range of applications. It provides a simple and easy way to draw a definite conclusion from ambiguous, imprecise or vague information. Like Artificial Neural Network (ANN) models, some fuzzy inference system (FIS)s have the capability of universal approximation. The adaptive neurofuzzy inference system (ANFIS) belongs to the class of systems commonly known as neuro-fuzzy systems (NF). NF combines the advantages of ANN with those of fuzzy systems. Here, an ANFIS based identification system is described which uses fingerprint and retina as inputs. Experiments are carried out using a number of samples. Obtained results show that the system is reliable enough for considering it as a part of a verification mechanism.

Keywords: ANFIS (Adaptive Neuro-Fuzzy Inference System), Fingerprint, Ridge, Minutiae, Fuzzy Logic, Retina, Blood Vessel.

Biometrics or biometric authentication refers to the identification of humans by their characteristics or traits. Biometrics is used in computer science as a form of identification and access control. Biometric identifiers are the distinctive, measurable characteristics used to label and describe individuals.1 They are often categorized as physiological versus behavioral characteristics. A physiological biometric would identify by iris scan, DNA or fingerprint. Behavioral biometrics is related to the behavior of a person, including but not limited to: typing rhythm, gait and voice. Fingerprint identification is the mature biometric technique used for criminal investigations. Major representations of the finger are based on the entire image, finger ridges or salient features derived from the ridges (minutiae). These characteristics are used to generate an orientation field of the fingerprint, which subsequently provides the discriminating details for authentication of persons. Fingerprint identification is a popular technique because of their easy access, low price of fingerprint sensors, non-intrusive scanning and relatively good performance. In recent years, significant performance improvements have been achieved in commercial automatic fingerprint recognition systems (FRS). The fingerprint of an individual is unique and remains unchanged ∗

Author to whom correspondence should be addressed.

J. Comput. Intell. Electron. Syst. 2014, Vol. 2, No. 2

over a lifetime.2 Retina is another unique biometric pattern that can be used as a part of a verification system. Retina identification is an automatic method that provides true identification of the person by acquiring an internal body image which is difficult to counterfeit.3 Retina identification has found application in security environments of all types. Here, we propose a FRS and Retina Recognition System (RRS) based on ANFIS. Experimental results show that the proposed approach is fast, reliable and robust for a range of conditions. An ANFIS can be configured and trained to handle range of variations in the texture of the fingerprint and retina. The specialty of the work is associated with the fact that if the ANFIS is configured properly in terms of number and types of membership functions (MF) it can tackle the variations in the fingerprint and retinal images. This way the approach provides the insights for developing a system which requires these samples for verification and authorization. A system designed to provide authentication decision using fingerprint and retina as inputs and an ANFIS system as the decision support mechanism can be a reliable means of verification. Various works on biometric identification based on fingerprint and retina recognition using fuzzy have been going on all over the world. The work4 by Hsieh and Hu reports a method which uses ridge bifurcation as fingerprints minutiae and a fuzzy encoder with BPNN as recognizer for fingerprint recognition. Again in Ref. [5], authors designed a technique of fingerprint image enhancement

2326-3008/2014/2/001/007

doi:10.1166/jcies.2014.1052

1

RESEARCH ARTICLE

1. INTRODUCTION

RESEARCH ARTICLE

ANFIS Based Fingerprint and Retina Recognition

and minutiae extraction. Here, we noticed that the major literature have been restricted to the preprocessing techniques with no focus on the variations of the ridge pattern. Again we noticed that the literature have been restricted to the popular NFS and FNS learning and decision making arena with no focus on performance variations that may be derived from the application of fuzzy based hybrid systems with varying fuzzification norms and inference rules. In Ref. [6] Xu et al., obtained vector curve of blood vessel’s skeleton using the green channel gray-scale retina images. They defined a set of feature vectors for each image including feature points, directions and scaling factor. Although they have reached a good recognition result, but major drawbacks of their method is its computational cost since a number of rigid motion parameters should be computed for all possible correspondences between the query and enrolled images in the database. Another approach is reported in Ref. [7] by Ortega et al. They used a fuzzy circular Hough transform to localize the optical disk in the retina image. This algorithm is computationally more efficient with respect to the algorithm presented in Ref. [6]. But, the performance of the algorithm has been evaluated using a very small database including only 14 subjects. The rest of the paper is organized as follows: Section 2 provides the brief theoretical background. Section 3 provides the working principle of the proposed system model. All experimental results and related discussions are provided in Section 4–5. This paper is concluded by summing up the work in Section 6. Some of the relevant literatures are cited between.22 23 24

2. BASIC THEORETICAL ASPECTS RELATED TO THE PROPOSED SYSTEM Here, we briefly cover the basic theoretical aspects related to the work. 2.1. Fingerprint These are graphical flow-like ridges and valleys present on the surface of human fingers. Typically, there are two prominent types of minutiae (ridge endings and ridge bifurcations) that constitute a fingerprint pattern. A ridge ending is defined as the ridge point where it ends abruptly. A ridge bifurcation is defined as the ridge point where a ridge diverges into branch ridges. A fingerprint can be represented by the minutiae locations, types and attributes like orientation. A proper quality fingerprint image typically has about 40 to 100 minutiae, but a dozen of minutiae are considered sufficient to identify a fingerprint pattern.5 2.2. Fingerprint Recognition It is one of the popular biometric techniques. It refers to the automated method of verifying a match between two fingerprint images.5 2

Borah et al.

2.3. Artificial Neural Network (ANN) An ANN is a soft-computing or mathematical tool based on the analogy of biological nervous systems consisting of interconnected group of artificial neurons and informationprocessing units using connectionist approach8 An ANN can be a single or multilayered approach. The knowledge gained during the training is stored in the interconnecting neurons for subsequent uses. 2.4. Fuzzy System Fuzzy sets were introduced by Zadeh in 19659 as a means of representing and manipulating data that was not precise, but rather fuzzy. Fuzzy logic (FL) starts with and builds on a set of user-supplied human language rules. It provides a simple way to arrive at a definite conclusion based upon vague, ambiguous, imprecise, noisy or missing input information. FL provides a framework that addresses the issue of uncertainty and lexical imprecision. FL is a set of mathematical principles for knowledge representation and reasoning based on degrees of membership.10 The primary components of a fuzzy based system are fuzzification, fuzzy inference system (FIS) consisting of MFs and fuzzy rules and defuzzification stages as shown in Figure 1. In FIS, fuzzifier takes the crisp inputs to a fuzzy system and converts them into fuzzy inputs. Fuzzy rule base consists of fuzzy IF-THEN rules that form the heart of a FIS. FIS makes use of FL principles to combine the fuzzy IF-THEN rules. Defuzzifier extracts a crisp value from a fuzzy set. There are three types of FISs that have been widely employed in various applications. These are Mamdani FIS (1975), Tsukamoto FIS (1979) and Sugeno FIS (TSK, 1985). The Mamdani FIS11 was proposed as the first attempt to control a steam engine and boiler combination by a set of linguistic control rules obtained from experienced human operators. It is the most commonly used fuzzy models. Tsukamoto fuzzy model12 is not used often since it is not as transparent as either the Mamdani or Sugeno fuzzy model. Sugeno or TSK fuzzy model was proposed by Takagi, Sugeno and Kang13 in an effort to develop a systematic approach to generating fuzzy rules from a given inputoutput dataset. It is almost similar to the Mamdani model.

Inputs

Fuzzification

DeOutput fuzzification

Inference engine

Membership function

Fuzzy rules

Knowledge base Fig. 1. Block diagram of a fuzzy system.

J. Comput. Intell. Electron. Syst. 2, 1–7, 2014

Borah et al.


The main difference is that Mamdani’s model expects the output MFs to be fuzzy sets whereas Sugeno output MFs are either linear or constant.

Rule 1 If x is A1 and y is B1 then f1 = p1 x + q1 y + r1 Rule 2 If x is A2 and y is B2 then f2 = p2 x + q2 y + r2 Where Ai and Bi are the fuzzy sets, fi are the outputs within the fuzzy region specified by the fuzzy rule, pi , qi and ri are the design parameters that are determined during the training process. Figure 2, shows the Type-3 fuzzy reasoning mechanism for a first order Sugeno fuzzy model.

Fig. 2.

TSK type-3 fuzzy reasoning.


Fig. 3. Equivalent ANFIS.

The corresponding equivalent ANFIS architecture to implement these two rules is shown in Figure 3, in which a circle indicates a fixed node, whereas a square indicates an adaptive node.16 The general architecture of ANFIS, which is primarily a fuzzy system having five layers as described below: Layer 1: Every node i in this layer is an adaptive node with a node output defined by Oi1 = Ai x Oi1

for i = 1 2

= Bi−2 y

or

for i = 3 4

where x or y is the input to the node and Ai or Bi−2 is a fuzzy set associated with this node. The outputs of this layer are the membership values of the premise part. Here, the MFs for Ai and Bi can be any appropriate parameterized MFs. For example, if the bell shaped MF is employed, Ai x is given by Ai x =

1 1 + x − ci /ai 2bi

where ai , bi and ci are the parameters of the MFs, governing the bell shaped functions accordingly and are referred to as premise parameters. Layer 2: Every node in this layer is a fixed node labeled , which multiplies the incoming signals and gives the product as output. For instance, Oi2 = wi = Ai x × Bi y

i = 1 2

Each node output represents the firing strength of a rule. Layer 3: Every node in this layer is a fixed node labeled N, indicating that they play a normalization role to the firing strengths from the previous layer. The output of this layer can be represented as Oi3 = w¯ i =

wi w1 + w 2

i = 1 2

which are the so called normalized firing strengths. Here w1 , w2 are the firing strengths to the rules 1 and 2. Layer 4: Every node in this layer is an adaptive node. The output of each node in this layer is simply the product of the normalized firing strength and a first order polynomial (for a first order Sugeno Model). Thus the output of this layer is given by: Oi4 = w¯ i fi = w¯ i pi x + qi y + ri

i = 1 2 3

RESEARCH ARTICLE

2.5. Adaptive Neuro-Fuzzy Inference System (ANFIS) Fuzzy systems provide as expert level decision making system by following the variations in the texture of the input data. Artificial Neural Network (ANN) and Fuzzy System (FS) are the important components of soft computing. They have received large acceptance in the diverse fields of uncertainty. ANNs are comfortable with the situations that have sufficient model free measurement data14 whereas FSs work where sufficient process data are available and expert knowledge is necessary.15 Thus ANN and FS having the numeric quantitative capability and symbolic-qualitative capacity respectively can be applied to solve problems where higher precision of estimation is difficult. Hybrid systems formed by combinations of fuzzy and ANN methods have adaptability, parallelism, non-linear processing, robustness and learning in data rich environment for modeling uncertainty.10 Fuzzy and ANN can be combined to form either Neuro-Fuzzy System (NFS) or Fuzzy-Neural System (FNS) based units. Here, we are using NFS based approach. The ANFIS combines the advantages of ANN with those of fuzzy systems. It has found numerous applications in a variety of fields. Here, the fuzzy model can either be Mamdani fuzzy model or TSK fuzzy model. However, TSK model is smoother and computationally efficient representation than Mamdani’s model. Let, the considered FIS is given two inputs ‘x’ and ‘y’ and one output ‘f .’ For a first-order Sugeno fuzzy model, a common rule set with two fuzzy if-then rules is as follows:


parameters in this layer will be referred to as consequent parameters. Layer 5: In thislayer, there is only one single fixed node labeled with that performs the summation of all incoming signals. Hence, the overall output of the model is given by: wf 5 Oi = w¯ i fi = i i i = overall output i wi i=1 Thus, ANFIS uses a hybrid learning procedure for estimation of the premise and consequent parameters.9

RESEARCH ARTICLE

3. PROPOSED SYSTEM MODEL The proposed biometric identification system using anfis in block diagram form is shown in Figure 4. It involves the following modules: • Image acquisition: It is required to capture a sequence of input images.Image preprocessing: It includes various stages which should be taken for making an image suitable for manipulation and interpretation by subsequent stages. The steps include removal of noise and variation of intensity recorded, sharpening, improving the contrast and strengthening the texture of the image. Another important aspect is image restoration which extracts image information from a degraded form to make it suitable for subsequent processing and interpretation.17 • Feature extraction: It is a process through which certain vital information and details of an image section is captured for subsequent interpretation. • Classification: This is the key component of the system and determines the system’s performance to a large extent. ANFIS is used as classifier and it produces the correct result by classifying the feature extracted templates and matching these features with known patterns in the feature database.

4. DESIGN AND IMPLEMENTATION OF THE PROPOSED SYSTEM

Borah et al.

In this proposed model, a multi stage approach is used. The decision obtained from the system is used to generate the response. During image acquisition the operations are performed separately. Fingerprint image can be captured by a digital scanner, digital camera and sensor etc. for subsequent manipulation. Fingerprint image preprocessing includes resizing the original images to required size, image normalization, segmentation and thinning etc. Normalization is done so that the gray level values lies within a given set of values. It is required as the image usually has distorted levels of gray values. It allows standardizing the distorted levels of variation in the gray scale values. It involves pixel-wise operations and does not change the structures. Retina images captured by the Fundus camera are preprocessed for subsequent manipulation. Retina image preprocessing includes gray image conversion and resizing the original images to required size. The next stage is feature extraction. In case of fingerprint, the thinned images are next considered for the minutiae feature extraction. The minutiae feature extraction algorithm extracts the main minutiae features required for matching two fingerprints. Here, Crossing Number (CN) method is used for minutiae extraction of fingerprints.18 Retina features include patterns of blood vessels.19 In this proposed model, we are using Principal Component Analysis (PCA) to extract the retina features. 4.1. Feature Extraction of Fingerprint Using CN Algorithm The ridge pixel can be divided into bifurcation, ridge ending and non-minutiae point based on it. The CN algorithm is working on pixel representation ‘1’ or ‘0’, but the decision of minutiae point can be selected for each pixel value. CN method extracts the ridge endings and bifurcations from the skeleton image by examining the local neighborhood of each ridge pixel using a 3 × 3 window. The CN for a ridge pixel P is given by:

In this work, the focus is to study the performance of the fingerprint recognition system that provides reliability, accuracy and reduced overall match speed. Image acquisition

Image preprocessing

Actual output

Image enhancement

Feature extraction

ANFIS output Decision (result) Fig. 4.

4

Matching

System model of the complete system.

Classification using ANFIS

CN = 05

8 i=1

Pi − Pi+1

P9 = P1

(1)

where Pi is the pixel value in the neighborhood of P . For a pixel P , its eight neighboring pixels are scanned in an anti-clockwise direction as follows: P4 P5 P6

P3 P P7

P2 P1 P8

After the CN for a ridge pixel has been computed, the pixel can then be classified according to the property of its CN value. With this formula, if CN = 1 it corresponds to the end point and if CN = 3, it corresponds to bifurcation point of minutiae. Other properties of CN are described in Table I. J. Comput. Intell. Electron. Syst. 2, 1–7, 2014

Borah et al.


Table I. Properties of CN. CN

Properties

0 1 2 3 4

Isolated point Ending point Connective point Bifurcation point Crossing point

In applying this algorithm, border area may be ignored, since there is no need to extract minutiae point on border area of the image that will gives more false minutiae points.

After a successful extraction of minutiae, they are stored in a template, which may contain the position, direction, type and quality of the minutiae. In our proposed model, ANFIS is used as classifier for recognition. It applies a combination of the least-squares method and the back propagation gradient descent method for training FIS membership function parameters to emulate a given training data set. Fingerprint with CN features and retina with PCA feature of length 101 are used for the training of ANFIS for 10 to 60 epochs. The results obtained are average values of atleast ten trails for the epochs considered.

where c determines the center of the curve, a is the half width and b (together with a) controls the slopes at the crossover points (where MF value is 0.5) and the parameter b is usually positive. A desired bell MF can be obtained by a proper selection of the parameter set a b c . Specifically, we can adjust c and a to vary the center and width of the MF and then use to control the slopes at the crossover points. Figure 5, shows a generalized bell MFs. The effects of change of a, b, c parameters are shown in Figure 6. We have also considered the Gaussian MF. Gaussian MF is specified by two parameters c s and is expressed as: 1 x − c m gaussianx c s m = exp − 2 s where c represents the center of the MF’s, width of the MF’s and m is the fuzzification factor (e.g., m = 2). Because of their smoothness and concise notation, gauss MF’s and bell MF’s are becoming increasingly popular methods for specifying fuzzy sets. Gaussian functions are well known in the fields of probability and statistics and they possess useful properties such as invariance under multiplication and Fourier transform. The bell MF has one more parameter than the Gaussian MF, so it can approach a non fuzzy set if b → .15 A detailed set of experiments

4.3. Fuzzification One of the primary blocks of our proposed approach is a fuzzfier. In our ANFIS based FRS and RRS, we have used Gaussian and generalized bell MFs. The related details and basic principles are described in Section 2. E. It has the advantage of being smooth and nonzero at all points. A generalized bell MF is specified by three parameters a b c and is expressed as: 1 bellx a b c = 1 + x − c/a2b J. Comput. Intell. Electron. Syst. 2, 1–7, 2014

Fig. 6. The effects of changing parameters in bell function: (a) changing ‘a’; (b) changing ‘b’; (c) changing ‘c’; (d) changing ‘a’ and ‘b’ simultaneously but keeping their ratio constant.

5

RESEARCH ARTICLE

4.2. Feature Extraction of Retina Using PCA Algorithm Principal Component Analysis (PCA) is a mathematical procedure that uses linear transformations to map data from high dimensional space to low dimensional space.20 The low dimensional space can be determined by Eigen vectors of the covariance matrix. The following steps involved in PCA include: Step 1: Get the mean value S¯ of the given dataset S. Step 2: Subtract the mean value say from S. From these values a new matrix is obtained. Let say A. Step 3: Covariance is obtained from the matrix i.e., C = AAT Eigen values are obtained from the covariance matrices that are V1 V2 V3 V4 VN Step 4: Finally, Eigen vectors and Eigen values are calculated for covariance matrix C. Step 5: Only the largest Eigen values are kept to form lower dimensional dataset.

Fig. 5. Physical meaning of parameters in a gbell MF.


Borah et al.

are carried out. Respective performances observed derived by considering these two MFs and their impact in ascertaining the system performance.

Table IV.

Input name

5. EXPERIMENTAL DETAILS AND RESULTS

RESEARCH ARTICLE

The performance of FRS and RRS are analyzed in terms of computational speed and reliability. The overall computational time taken by the system is reduced to a greater level. ANFIS parameters are adjusted so as to reduce the error measure defined by the sum of the square of the difference between the actual and desired output. The Root Mean Square Error (RMSE) is calculated using the following expression: N 1 RMSE = A − Ft 2 (2) N t=1 t where At and Ft are actual and fitted values respectively and N is the number of training or testing sample. The RMSE is used as the primary criterion to ascertain the extent of learning acquired by the ANFIS. The obtained results are compared with21 and found that the ANFIS generate comparable RMSE values but the computational speed is significantly better. The ANFIS shows at least 50% increase in computational speed which establishes its usefulness. Further, the RMSE convergence is completed in less number of epochs. Thus the proposed approach is suitable for verification and authorization for

Fingerprint

Retina

Time required for 50 epochs for different inputs. Time required to reach 50 epochs by (in sec.)

Number of membership functions (MFs)

Gauss mf

Gbell mf

2 4 6 8 10 2 4 6 8 10

102 101 101 101 10 15 15 16 16 16

109 108 106 105 101 15 16 16 16 16

the design of FRS and RRS. A total of 40 identical fingerprint as well as retina images have been provided to the system for training, validation and testing. After extensive training, the system is subjected to certain variations with signal to noise ratio (SNR) range between 0 to 3 dB to achieve robustness and proper recognition. The ANFIS considered is configured using the specifications shown in Table II. The convergence RMSE and time required due to change in MF are shown in Tables III and IV for both the two inputs. Again, we have also considered the performance for both the Gaussian MF and Bell shaped MF. Here, we see that the best average RMSE is obtained for Gauss MF with 6 and 10 numbers of MFs with an average

Table II. ANFIS specification. Input data size

For fingerprint—CN features of length 101 For retina—PCA features of length 101. 0–3 dB ANFIS with five layers Least-squares with the back propagation gradient descent method 10–60 2, 4, 6, 8 and 10 functions Gbell MF, Gaussian MF functions

SNR ANFIS type ANFIS training method Average training epochs Total no. of membership Type of membership

Table III. Calculation of average RMSE for 50 epochs for different inputs.

Input name Fingerprint

Retina

6

Number of membership functions (MFs) 2 4 6 8 10 2 4 6 8 10

Fig. 7. RMSE plot for fingerprint image.

Minimum RMSE attained by after 50 epochs Gauss mf −2

4 × 10 23 × 10−2 4 × 10−3 8 × 10−3 6 × 10−3 37 × 10−3 36 × 10−3 35 × 10−3 34 × 10−3 16 × 10−3

Gbell mf 9 × 10−2 28 × 10−2 56 × 10−2 65 × 10−2 1 × 10−2 37 × 10−3 36 × 10−3 36 × 10−3 34 × 10−3 16 × 10−3

Fig. 8. RMSE plot for retina image.


Borah et al.


RMSE of 4 × 10−3 and 16 × 10−3 obtained within 10.1 and 1.6 seconds respectively. The RMSE curves for both the FRS and RRS with 6 and 10 numbers of MFs are shown in Figures 7 and 8. Also with Gaussian MFs, the computational time required is a bit lower compared to Bell MF. The advantage of using the ANFIS for FRS and RRS thus is obvious. With proper selection of MF type and numbers higher efficiency and lower computational time can be obtained for a range of patterns.

11.

6. CONCLUSIONS

12.

Here, we described a FRS and RRS where the ANFIS forms a critical decision support system. The experimental results show that the proposed approach is at least 50% more efficient compared to ANN based approaches reported earlier.21 The strength of the proposed system is its speed, computational efficiency, robustness and high precision which shall make it suitable for certain application. The future scope of this work is to improve the ANFIS architecture to achieve high classification accuracy. The overall performance of the system can be enhanced further by considering more number of samples and variations and by using better preprocessing techniques with ANFIS based blocks.

1. A. Jain, L. Hong, and S. Pankanti, Biometric Identification. Communications of the ACM 43, 91 (2000). 2. A. Jain, L. Hong, S. Pankanti, and R. Bolle, An identity authentication system using fingerprints, in Proceedings of the IEEE 85 September (1997), pp. 1365–1388. 3. R. B. Hill, Apparatus and method for identifying individuals through their retinal vasculature patterns, US Patent No. 4109237 (1978). 4. C. T. Hsieh and C. S. Hu, An Application of Fuzzy Logic and Neural network to Fingerprint Recognition, Tamkang University. 5. R. Thai, Fingerprint image enhancement and minutiae extraction, M. Eng. Thesis, The University of Western Australia (2003). 6. Z. W. Xu, X. X. Guo, X. Y. Hu, and X. Cheng, The blood vessel recognition of ocular fundus, Proceedings of the 4th International

8. 9. 10.

13.

14.

15. 16. 17. 18.

19. 20.

21.

22. 23.

24.

Received: 16 July 2013. Accepted: 14 August 2013.


7

RESEARCH ARTICLE

References

7.

Conference on Machine Learning and Cybernetics (ICMLC05), Guangzhou, China, August (2005), pp. 4493–4498. M. Ortega, C. Marino, M. G. Penedo, M. Blanco, and F. Gonzalez, Biometric authentication using digital retinal images, Proceedings of the 5th WSEAS International Conference on Applied Computer Science (ACOS 06), Hangzhou, China, April (2006), pp. 422–427. S. Haykin, Neural Networks-A Comprehensive Foundation, Pearson Education, 2nd edn., New Delhi (2003). L. A. Zadeh, Fuzzy sets. Information and Control 8, 338 (1965). T. J. Ross, Fuzzy Logic with Engineering Applications, 2nd edn., Wiley, New Delhi, India (2008). E. H. Mamdani and S. Assilian, An experiment in linguistic synthesis with a fuzzy logic controller in International Journal of ManMachine Studies 7, 1 (1975). Y. Tsukamoto, An approach to fuzzy reasoning method, Advances in fuzzy set theory and applications, edited by Madan M. Gupta, Rammohan K. Ragade, and Ronald R. Yager, North-Holland, Amsterdam (1979), pp. 137–149. T. Takagi and M. Sugeno, Fuzzy identification of systems and its applications to modeling and control. IEEE transactions on Systems, Man and Cybernetics 15, 116 (1985). S. Mitra and Y. Hayashi, Neuro-fuzzy rule generation: Survey in soft computing framework. IEEE Transactions on Neural Networks 3, 748. J. S. R. Jang and C. T. Sun, Neuro fuzzy modeling and control. The Proceedings of the IEEE 83, 378 (1995) R. J. Jang, ANFIS: Adaptive-Network Based Fuzzy Inference System. IEEE Transaction on Systems, Man and Cybernetics 23 (1993). K. K. Sarma, MATLAB, Demystified, Basic Concepts and Applications, 1st edn., Vikas Publication, New Delhi (2010). D. Maltoni, D. Maio, A. K. Jain, and S. Prabhakar, Handbook of Fingerprint Recognition, Springer Professional Computing, 2nd edn., Berlin (2009). R. B. Hill, Retina Identification, Biometrics, Springer (2003). M. Suganthy and P. Ramamoorthy, Principal component analysis based feature extraction, morphological edge detection and localization for fast iris recognition. Journal of Computer Science 8, 1428 (2012). T. R. Borah, K. K. Sarma, and P. H. Talukdar, Retina and fingerprint based biometric identification system. International Journal of Computer Applications (IJCA) 74 (2013). R. C. Gonzalez and R. E. Woods, Digital Image Processing, 3rd edn., Pearson Education (2009). P. Manju and V. Subbiah, Membership functions in the operation of fuzzy based unified power flow controller in a power system network. J. Comput. Intell. Electron. Syst. 2, 60 (2013). J.-H. Wang and K. Liu, Feature-based fuzzy neural network approach for intrusion data classification. J. Comput. Intell. Electron. Syst. 1, 99 (2012).