Iris Recognition using Four Level Haar Wavelet · PDF fileIris Recognition using Four Level...

Iris Recognition using Four Level Haar Wavelet Transform

Anjali Soni1, Prashant Jain2 M.E. Scholar, Dept. of Electronics and Telecommunication Engineering, Jabalpur Engineering

College, Jabalpur, Madhya Pradesh, India1

Professor, Dept. of Information and Technology Engineering, Jabalpur Engineering College, Jabalpur, Madhya Pradesh, India2

Abstract

There is considerable rise in the research of iris recognition system over a period of time. Most of the researchers has been focused on the development of new iris pre-processing and recognition algorithms for good quail iris images. In this paper, iris recognition system using Haar wavelet packet is presented. Wavelet Packet Transform (WPT ) which is extension of discrete wavelet transform has multi-resolution approach. In this iris information is encoded based on energy of wavelet packets. And then matching of this iris code with the stored one is performed using hamming distance . Our proposed work significantly decreases FAR and FRR values as compared to previous work. Experimental results are demonstrating significant improvements in iris verification process

1. Introduction For the last decade, biometric-based personal verification and identification methods have gained much interest with an increasing emphasis on security. Iris recognition is a fast, accurate and secure biometric technique that can operate in both verification and identification modes since the iris texture pattern has no links with the genetic structure of an individual and since it is generated by chaotic processes [1]. Consequently, a variety of iris recognition approaches were proposed that can be broadly classified in three categories depending on the method used to extract features from the texture. These categories are: (1) texture-based, (2) appearance-based and (3) featurebased extraction. Texture-based techniques [2-9] make use of filters for image processing and extract iris features from the filtered images. Daugman [2] used Gabor wavelets to extract texture phase structure

information of the iris to compute a 256-byte binary iriscode. The Hamming distance measure was used to compare the difference between a pair of iris representations. In Wildes’ approach [4], the Laplacian pyramid constructed with four different resolution levels is used as iris features. Then, the normalized correlation is used to compare both the input and the template iris images. Wang et al [7,8] have proposed recognition techniques using multichannel Gabor filtering. Hence, they have used 8- directional Gabor filters with multiple frequencies to capture both global and local details in an iris image.

The mean and variance of these Gabor filtered images are used as features for the matching process. Other approaches that exploit other wavelet categories such as the 2D Haar wavelet and zero-crossing representation of 1D wavelet Transform, were proposed in [9] to characterize an iris texture. Appearance-based approaches [10,11], such as Independent Component Analysis (ICA), direct Linear Discriminant Analysis (LDA) and Principal Component Analysis (PCA), attempt to use classical statistical approaches to extract iris features. The PCA is superior in image construction, because it can control construction errors by selecting the cumulative variance. Euclidean-

Anjali Soni et al, International Journal of Computer Technology & Applications,Vol 7(4),568-575

IJCTA | July-August 2016 Available [email protected]

568

ISSN:2229-6093

Figure 1.Iris recognition process

distance and Nearest- Neighborhood (NN) classifier are adopted in these approaches.

Feature-based extraction techniques [7,11,12] use local variations, which are characterized by the appearance and disappearance of an important image structure. A bank of spatial filters is constructed for efficient feature extraction and for matching process. However, if the original iris images are of poor quality, the wavelet-packet approaches are more efficient for iriscode generation comparatively to the classical wavelet approaches without a greater risk of iris recognition errors. For this reason, some authors have proposed the iris recognition based on wavelet packet [13].

In this paper, we present a method for iris recognition based on a wavelet packet decomposition of iris images. Each iris image is described by a subset of band-filtered images (subimages) containing wavelet coefficients. From these coefficients, which characterize the iris texture, we compute a compact iris feature code using the appropriate energies of these sub images to generate binary iris codes according to an adapted threshold. Thereafter, we show how an efficient and reliable Hamming distance can be used in order to classify iris codes. Results are presented that demonstrate significant improvements in iris recognition accuracy.

2. IRIS RECOGNITION SYSTEM

Figure 1 summarizes the steps to be followed when doing iris recognition.

Step 1: Image acquisition, the first phase, is one of the major challenges of automated iris recognition since we need to capture a high-quality image of the iris while remaining non-invasive to the human operator. that a comparison between Step 2: Iris localization takes place to detect the edge of the iris as well as that of the pupil; thus extracting the iris region.

Step 3: Normalization is used to be able to transform the iris region to have fixed dimensions, and hence removing the dimensional inconsistencies between eye images due to the stretching of the iris caused by the pupil dilation from varying levels of illumination.

Step 4: The normalized iris region is unwrapped into a rectangular region.

Step 5: Finally, it is time to extract the most discriminating feature in the iris pattern so templates can be done. Therefore, the obtained iris region is encoded using wavelets to construct the iris code.

As a result, a decision can be made in the matching step.



569

ISSN:2229-6093

3. PROPOSED METHODOLOGY Figure.2 shows the flow of the proposed method. It consists of various stages like segmentation ,normalization, feature extraction and matching.

3.1.Iris image pre-processing

3.1.1. Iris image localisation

The iris is an annular part between the pupil (inner boundary) and the sclera (outer boundary). Different iris localization approaches have been reported in the literature: Integro-Differential, Circular Hough, Hough Transform and edge and contour detection. In our approach, this task is performed by hough transform. The Hough transform is a standard computer vision algorithm that can be used to determine the parameters of simple geometric objects, such as lines and circles, present in an image. The circular Hough transform can be employed to deduce the radius and centre coordinates of the pupil and

Firstly, an edge map is generated by calculating the first derivatives of intensity values in an eye image and then thresholding the result .

Figure.2. Proposed method of iris recognition

3.1.2. Iris image normalization The Once the iris region is segmented, the next stage is to normalize this part, to enable generation of the iris code and their comparisons. Since variations in the eye, like optical size of the iris, position of pupil in the iris, and the iris orientation change person to person, it is required to normalize the iris image, so that the representation is common to all, with similar dimensions.

Normalization process involves unwrapping the iris and converting it into its polar equivalent. It is done using Daugman’s Rubber sheet model. The centre of the pupil is considered as the reference point and a Remapping formula is used to convert the points on the Cartesian scale to the polar scale.

The radial resolution was set to 20 and the angular resolution to 240 pixels.

(a)Original image

(b)iris localisation

(c)normalization

Figure. 3. Iris image segmentation



570

ISSN:2229-6093

3.2 . Feature extraction and encoding

The majority of researchers have proposed wavelets approaches in order to capture local iris features at different scales.

After the iris image segmentation process is completed, the iris code is performed using Haar wavelet packets as well as the energy of the packets sub-images to extract texture phase structure information of the iris and to compute the iris 256-bits codes.

The Iris code generation process can be summarized

in the following steps:

3.2.1. Wavelet packets decomposition: We have used the Haar wavelet in a 4-level wavelet packet decomposition to extract the texture features of the unwrapped images . This generates 256 wavelet packets (output iris subimages), numbered 0 to 255.

In the Haar wavelet transformation method, low-pass filtering is conducted by averaging two adjacent pixel values, whereas the difference between two adjacent pixel values is figured out for high-pass filtering.

The Haar wavelet applies a pair of low-pass and high-pass filters to image decomposition first in image columns and then in image rows independently. As a result, it produces four sub-bands as the output of the first level Haar wavelet. The four sub-bands are LL1, HL1, LH1, and HH1. Upto four levels of decomposition are done to get the detail image [14]. The WT separates an image into a lower resolution approximation image (LL) as well as horizontal (HL), vertical (LH) and diagonal (HH) detail components. The process can then be repeated to compute multiple scale wavelet decomposition, as in the four scales WT is shown in fig.4.

Figure.4 .four level wavelet decomposition

3.2.2. Wavelet packets energy computation: In order to obtain the most texture information in packet subimages, we have used an energy measure. The mean energy distribution allows evaluating which packets are used to compute the normalized adapted threshold for our iriscode generation. The energy measure Ei for a wavelet packet subimage Wi can be computed as follows :

We use the appropriate wavelet packet energies of each iris image to compute the adapted threshold to encode the 256 subimages Let E1…Eλ be the appropriate wavelet packet

energies of the packets 1…λ respectively. We define the normalized adapted threshold S as follows:

where represents the mean wavelet peak energy value, Coeff as a constant and λ is the number of the appropriate energies.



571

ISSN:2229-6093

3.2.3. Iris feature coding: After determination the appropriate wavelet packets energies and the normalized adapted threshold, we can carry out the coding of the 256 wavelet packets energies to generate a compact iriscode by quantizing these energies into one bit according to each appropriate energy. Let Eλ be the appropriate energy of the peak λ. Then the iriscode Cλ.

computed according to E λwe define by the following:

Where j=0...............255.

In our approach, we have used significant wavelet coefficients of the iris subimage. Each used

appropriate energy resulting in a total of 256 bits which correspond to the 256 subimages of the iris wavelet decomposition. Therefore, we obtain one iriscode according to each energy.

3.3. Iris code matching

The iris codes matching task is performed by pairing the iriscodes extracted from the input and the template iris images. The most common comparison method of iris signatures is the Hamming Distance (HD) [12]. This comparison is to be made with the user’s iris template, which will be calculated depending on the binary matching algorithm which is implemented by the Boolean exclusive-OR operator ( ) applied to the 256 bit vectors that encode any two iriscodes.

be the iris binary codes extracted from the template and input iris images. The binary Hamming distance

HD i,k, between these iris binary codes is measured according to the following

The preceding distance is computed according to the appropriate energy (maximal value) between two iris signatures in order to decide if both come from images of the same iris (authentic distance) or from Images of different iris (impostor distance). Therefore, the combination of two or several packets is likely to hold more discriminating information than what each of them does separately as mentioned in. So, the preceding distance can be extended for two appropriate energies of each iris image as follow [14]:

where Ci and Ciˈ are the iris codes computed according to the two appropriate energies of the template iris image and Ci and Ciˈ of the input one.

4. Comparison of haar wavelet and gabor filter

Daugman [13] uses convolution with 2-dimensional Gabor filters to extract the texture from the normalized iris image. In his system, the filters are “multiplied by the raw image pixel data and integrated over their domain of support to generate coefficients which describe, extract, and encode image texture information”. After the texture in the image is analyzed and represented, it is matched against the stored representation of other irises. If iris recognition were to be implemented on a large scale, the comparison between two images would have to be very fast. Thus, Daugman chose to quantize each filter’s phase response into a pair of bits in the texture representation. Each complex coefficient was transformed into a two-bit code:

the first bit was equal to 1 if the real part of the coefficient was positive, and the second bit was equal to 1 if the imaginary part of the coefficient was positive.



572

ISSN:2229-6093

Thus after analyzing the texture of the image using the Gabor filters, the information from the iris image was summarized in a 256byte (2048 bit) binary code. The output of applying the wavelets process is then encoded in order to provide a compact and discriminating representation of the iris pattern. In order to perform wavelets process, a feature vector is used to describe an image or some measures obtained from wavelet coefficients.

Gabor filters are used to provide optimum conjoint representation of a signal in space and spatial frequency. A Gabor filter is constructed by modulating a sine/cosine wave with a Gaussian which able to provide the optimum conjoint localization in both space and frequency, since a sine wave is perfectly localized in frequency, but not localized in space. Modulation of the sine with a Gaussian provides localization in space, though with loss of localization in frequency. Decomposition of a signal is accomplished using a quadrature pair of Gabor filters, with a real part specified by a cosine modulated by a Gaussian, and an imaginary part specified by a sine modulated by a Gaussian. The real and imaginary filters are also known as the even symmetric and odd symmetric components respectively.

The centre frequency of the filter is specified by the frequency of the sine/cosine wave, and the bandwidth of the filter is specified by the width of the Gaussian. Daugman designed a 2D version of Gabor filters in order to use for encoding iris pattern data. Daugman was suggested to demodulate the output of the Gabor filters in order to compress the data. This can be performed by quantizing the information into four levels, for each possible quadrant in the complex plane.

These four levels of phase information are represented by using two bits of data, which each pixel in the normalized iris pattern corresponds to these two bits of data in the iris template. A total of 2,048 bits are calculated for iris template, and an equal number of masking bits are generated in order to mask out the corrupted regions within the iris image. This creates a compact 256-byte template, which allows for efficient storage and comparison of irises.

We uses the wavelet transform to extract features from the iris region. For most recent studies both the Gabor transform and the Haar wavelet are considered as the mother wavelet methods. From multi-dimensionally filtering, a feature vector with 87 dimensions is computed. Since each dimension has a real value ranging from -1.0 to +1.0, the feature vector is sign quantized so that any positive value is represented by 1 and negative value as 0. Lim et al. compared the use of Gabor transform and Haar wavelet transform and showed that the recognition rate of Haar wavelet transform is slightly better than Gabor transform by 0.9%.

5 . Result

In this section, we analyze the performance of the proposed iris verification algorithm. So, to evaluate the performance of our algorithm, different experiments on the Institute of Automation, Chinese Academy of Science, CASIA database have been carried out. This experiment was motivated by an iris code computation using only two appropriate energies according to the equation (2 and 3) to compute the adapted threshold and one iriscode according to the maximal energy.

In biometric systems, reference threshold is defined as a value that can decide the authenticity of a person. Authenticity means whether the person is genuine or intruder. The statistical calculation of various values like reference threshold, FAR (False Acceptance Rate), FRR (False Rejection Rate) are required for real- time automated biometric authentication system because the measurement of biometric features are statistical values.

FAR can be defined as the rate at which a non- authorized person is authorized as genuine.

Depending on the choice of the classification threshold, between all and none of the impostor patterns are falsely accepted by the system. The threshold depending fraction of the falsely accepted patterns divided by the number of all impostor patterns is called False Acceptance Rate (FAR).Its value is one, if all impostor patterns are falsely accepted and zero, if none of the impostor patterns is accepted. The FRR is defined as the rate of a genuine person getting rejected.



573

ISSN:2229-6093

If a classification threshold that is too high is applied to the classification scores, some of the client patterns are falsely rejected. Depending on the value of the threshold, between none and all of the client patterns will be falsely rejected. The fraction of the number of rejected client patterns divided by the total number of client patterns is called False Rejection Rate (FRR). According to the FAR, its value lies in between zero and one.

So , values of FAR and FRR is calculated based on different threshold.These are shown in table no.1

Results: FRR and FAR values for the selected threshold values

As shown in table no. 1 ,it is clear that threshold value 0.35 gives more appropriate FARand FRR values.Now we compare our result with the approach used in . The accuracy of the authentication system is given by the following equation:

Accuracyy(%)=(100−(FAR(%)+FRR(%))/2) (6)

where, FAR is False Acceptance Rate FRR is False Rejection Rate

Table no .2

METHOD USED

FRR(%) FAR(%) ACCURACY(%)

Haar wavelet in 4 level

0.09 0.009 99.955

So ,our approach gives more accurate recognition with accuracy 99.955%. which means that our system is more appropriate for very high security environments

6.CONCLUSION

A proposal algorithm for iris recognition has been presented. Hough transform is useful for segmentation of the iris because of efficient localization. The Haar wavelet transform has a number of Advantages , it is conceptually simple, fast, memory efficient and reversible when compared with other wavelets. A fixed length feature vector is obtained. Experimental results show that proposal algorithm can effectively distinguish different persons by identifying their irises. It is also computationally efficient and insensitive to illumination and noise. Our future work will focus on more robust iris features as well as iris recognition from image sequence.

7.REFERENCES

1] Ales Muron and Jaroslav Pospisil, “The human iris structure and its usages,” Acta Univ. Palacki. Olomuc. Fac. Rerum Nat. Phys., vol. 39, 2000, pp. 87–95.

[2] J.G. Daugman, “High confidence visual recognition of persons by a test of statistical independence,” IEEE Trans. on Pattern Analysis and Machine Intelligence, vol. 15, no. 11, Nov 1993, pp. 1148–1160.

[3] J.G. Daugman, “Demodulation by Complex-Valued Wavelets for Stochastic Pattern Recognition,” Int’l J. Wavelets, Multiresolution and Information Processing, vol.1, no. 1, 2003, pp. 1-17.

Haar wavelets

Gabor filter

Threshold

FRR(%)

FAR(%)

Threshold FRR(%)

FAR(%)

0.25 4.2 0.0 0.35 NA NA

0.30 1.3 0.001 0.40 13.6 0.6

0.35 0.09 0.009 0.45 9.2 1.2

0.40 0.0 0.03 0.50 4.3 5.2

0.45 0.0 3.3 0.55 0.1 11.2



574

ISSN:2229-6093

[4] R. Wildes, “Iris Recognition: An Emerging Biometric Technology,” Proc. IEEE, vol. 85, 1997, pp. 1348-1363.

[5] S. Lim, K. Lee, O. Byeon, and T. Kim, “Efficient Iris Recognition through Improvement of Feature Vector and Classifier,” ETRI J, vol. 23, 2001, no. 2, pp. 61-70.

[6] Y. Zhu, T. Tan, and Y. Wang, “Biometric Personal Identification Based on Iris Patterns,” Proc. Int’l Conf. Pattern Recognition, vol. 2, 2000, pp. 805-808.

[7] L. Ma, Y. Wang, and T. Tan, “Personal identification based on iris texture analysis,” IEEE Trans. on Pattern Analysis and Machine Intelligence, vol. 25, no. 12, Dec 2003, pp. 1519–1533.

[8] W. Boles and B. Boashash, “A human identification technique using images of the iris and wavelet transform,” IEEE Trans. on Signal Processing, vol. 46, no. 4, Apr 1998, pp. 1185–1188.

[9] C. Sanchez-Avilaa, R. Sanchez-Reillob, “Two different approaches for iris recognition using Gabor filters and multiscale zero-crossing representation,” Pattern Recognition, vol. 38, 2005, pp. 231 – 240.

[10] K N Pushpalatha 1, Aravind Kumar Gautham2 D.R.Shashikumar3 and K.B ShivaKumar4 “ Iris Recognition System with Frequency Domain Features optimized with PCA and SVM Classifier” 5, No 1, September 2012.

[11] L. Ma, T. Tan, Y. Wang, and D. Zhang, “Efficient iris recognition by characterizing key local variations,” IEEE Trans. Image Process, vol. 13, 2004, pp. 739–750.

[12] L. Ma, T. Tan,D. Zhang, andY.Wang, “Local intensity variation analysis for iris recognition,” Pattern Recognition, vol. 37, no. 6, , 2005, pp. 1287–1298.

[13] Farid Benhammadi & Nassima Kihal “Personal Authentication Based on Iris Texture Analysis “2008 IEEE. [14] Tamililakkiya, V; Vani, K.; Lavanya, A.; Micheal A.; "Linear and non-linear feature extraction Algorithms for lunar images"; (SIPIj); 2011; IVSL.

[15] Gatheejathul Kubra.J, Rajesh.P , “ Iris Recognition and its Protection Overtone using Cryptographic Hash Function” SSRG International Journal of Computer Science and Engineering (SSRG-IJCSE) – volume 3 Issue 5–May2016.



575

ISSN:2229-6093

Iris Recognition using Four Level Haar Wavelet · PDF fileIris Recognition using Four Level...

Documents

Transcript of Iris Recognition using Four Level Haar Wavelet · PDF fileIris Recognition using Four Level...