A Survey On Iris Recognition Methods

Abstract'Biometric features provides the distinct characteristics to individual, biometric systems used these unique features to provide personal authentication. The inexpensive and fast digital circuits are capable of performing the complex mathematical functions at very high speed due to which this area of research rapidly to the maturity level .In this paper the comparative analysis of different iris recognition techniques in terms of performance and error rate is given.

Key words-Biometric systems, Independent component analysis, Phase based method, Hamming distance, Correlation, Euclidean distance, Gabor wavelet.



IOMETRIC recognition deals with the identification of a person by features of person's body which are different from each other. Iris recognition is a method of recognition like face, fingerprint, ear, hand geometry and vein recognition. Iris is part of eyes between sclera and pupil. the structure of iris is made of many layers. its outer surface is made of epithelial cell which are thickly pigmented. The uniqueness of Iris pattern was first proposed by the French ophthalmologist, Alphonse Bertillon [1] .Iris patterns has few properties which other biometric features do not have like; Iris patterns are stable of a person throughout the lifetime. They are unique as iris of any two person cannot be same, even iris pattern of twins are also different from each others [2].the iris pattern of both eyes the same person are also different [3]. Anti-falsification as iris pattern of a person cannot be steal. Dead iris can easily be identified during matching process. This method of recognition does not require physical interaction with the equipment. Comparing with other biometric system like fingerprint, voice, handprint, Iris is known to be the method of identification of person most accurately and it is reliable [4]. This method received great extension in last decade[5][6][7][8][9][10].On the other hand fingerprint recognition can easily requires physical contact with scanner and by this biometric recognition technique system can be fooled as its difficult to know that person is dead

Muhammad Husnain

B.E. Telecommunication

Iqra University

Karachi, Pakistan

[email protected]

or alive. For face recognition method, face of a person changes with age , hair growth on face can make problem for matching .


Image Acquisition

In first step, image acquisition is done by using cameras and sensor. Image captured should be clear this captured image consists of whole eye including. The occlusion, lighting, number of pixels on the iris are factors that affect the quality of image [11].


Since captured eye image have noise from environment and captured devices. Some preprocessing is done to remove noise and make eye image more detail defining. Pre-processing is done to detect iris aliveness. since image is of complete eye including eyelids and eye lashes. Detection of Features of iris are then calculated by using any extraction methods and are stored in database for future matching purpose.


In matching process, when ever recognition is required an input iris is taken from camera and after apply pre-processing steps matching is done by using any of the following method i.e Hamming Distance, Euclidean Distance, Normalized Correlation Coefficient. Recognition achieves result by comparison of features with stored patterns [12].

II. Recognition Methods

Hamming Distance

J. Daugman, used hamming distance for matching of iris template by using the integro-differential method to find the iris boundary he estimate the iris and pupil center, radius normalize the iris portion to polar form by using the function I(x(r, ??), y(r, ??)) ' I(r, ??) where r lies on the unit interval [0,1] and ??. He used 2D Gabor filter for representing the phase response of iris texture and 2048 bit vector is calculated for template the equal number of masking bits are generated and saved in the database for matching the iris code of the two different vectors is calculated by finding the hamming distance.

HD = (code A code B) mask A mask B (1)

Mask A mask B

Euclidean Distance.

Finding the hamming distance between the template is not a the very much effective approach because if matching template is shifted by any error (rotation ,translation) the resulting the drastic change on matching score. Another approach is proposed by Ya-Ping Huang [27] he used independent component analysis (ICA) for extracting the texture feature he find the average Euclidian distance for classification of iris pattern.

ED= '_(i=1)^N'(('(f_i-'f_i'^k)'^2)/'('??_i'^k)'^2 (2)

Normalized Correlation

Wildes, for the authentication of a person, used normalized correlation method for matching the acquired iris image with the images stored in database.

('_(i=1)^n''_(j=1)^m''(p_1 [i,j]-??_1)(p_2 [i,j]-??_2)')/(nm??_1 ??_2 ) (3)

Where p_1 is the image which is to be compared with database of images. p_2 is the image with which p1 is being compared, both images having size n*m. ??_1,??_2 are the mean of the image and ??_1,??_2 are standard deviation of images p_1 ' and p'_2 respectively. This matching technique gives better results as it is able to account local intensity variations in image pixel's values.

The average Euclidean Distance is used to recognize iris patterns for specific person from database.


Before these techniques of recognitions, individuals are provided with a username and password for authentication. If username and password is stolen intruder can easily access the personal information. Organization's sensitive information which is only for that person.Iris recognition system is being used to authorize entry through doors of secure areas. Iris recognition is used to identify person to get his or her complete data from database. It can also be used for attendance management system in educational institutes and companies Dutta, proposes that iris code of a person can be used in an audio file as a watermark to prove ownership of the audio file [13]. It is being used in airports to access sensitive area.


Iris is part of eye exists between the sclera and pupil. Iris has few features which differ it from other biometric, iris get stable at about two years of age [14]. and remains same throughout the lifetime. Statistical independence of iris can easily be found by using Boolean XOR, AND and Exclusive-OR of iris phase bits of two patterns [15]. Integro-differential operator method was proposed by Daugman to deduct the eyelid noise[6] [9].

In 1987, Flom and Ara proposed the the idea of recognition of person by his/her iris pattern to uniquely identify the person from the database [16].

The first working iris recognition system was implemented by John Daugman [17][18]. It was the most successful and well known system known, after that many system are developed.

Wildes used Hough transform to extract iris features [19].

1-D wavelet transform applied by Boles and Boashah to extract iris feature [20].

Sanchez-Avila and Sanchez-Reillo extended the work of boles and used 2-D Haar wavelet transform to extract features of iris patterns [21].

Hu invariant moment based algorithm was proposed by Ramli [22]. He used image dataset of CASIA which had 108 classes. Only 100 iris images were used. Histogram equalization is the first step of proposed algorithm, after a filtering step is done, then canny edge detection us donr to detect the edges , invariant movement and template matching, These are the five steps of algorithm proposed by Ramli.

Lagree and Bowye used log Gabor filter for segmentation. Segmentation is done to get the desired portion of the iris image and remove those parts of the image which are not required like eyelid, eyelashes. A 240*40 pixel normalized iris image with bitmask of eyelid. Feature vector of iris was built by using spot detector and line detector filters. Feature vector for all iris images was calculated[23].

Rose defines five steps in iris recognition system, which are image acquisition, image segmentation, normalization of image, encoding for storage in dataset, and matching . In first step image acquisition is done more than one image are captures and only one image is chosen which has good quality and have maximum iris information. After image acquisition segmentation is done. In this step boundaries of iris in eye image is detected and image parts like eye lashes and eyelids which are undesired are removed from eye image. In normalization step, same region of iris is obtained for matching it with other iris images. Due to pupil dilation, many iris images of same iris are captured from different angle. in last step encoding is done. Encoding is done after enhanced iris image is obtained, first 2D Gabor Filter is used to extract textural features. These textures are encoded as 2D binary

code 'Iris Code'. Hamming distance is used to compare two iris

codes [24].

Ziauddin and Dailey proposed hybrid method of localizing the iris which uses multiple techniques at a single time; intensity thresholding, edge detection and Hough transform [25]. Iris segmentation is done by using circular Hough transform, many improvements are required to be made to get accuracy and better performance.

Gupta and Saini evaluated the performance of existing iris recognition system by using Matlab image processing toolbox. Steps of the method are image acquisition, segmentation (for detection of pupil circle and iris boundary by Daugman's filter), normalization (to create rectangular block of fixed size by using rubber sheet model), image enhancement to convert low contras image into high contrast to make image more detail defining and image matching to perform iris code matching by Hamming distance [26].

A new method instead of traditional method proposed by Xu. In this, first global features are extracted by using wavelet filter and then by applying SIFT method local features of iris are obtained. To get the similarity distance between them different weights are applied [27].

Annapoorani proposed a robust method of iris segmentation for fast and accurate results. Elimination of specularities from input eye image, detection of pupil, localization of iris and detection of eyelid and eyelashes [28].

Sgroi investigate the age of a person (old / young) from his/her iris pattern. He achieved 64% accuracy. He concluded that using iris pattern, age of person can be found,. He applied nine different filters to get texture feature, which are small spot, large spot, thin vertical line, thick vertical line, thin horizontal line, thick horizontal line, S5S5, R5R5, E5E5 [29].

Key Findings

Before the method of biometric recognition techniques individuals are provided with there personal username and password. They use them as a identification on a large dataset to get their information. Password can be stolen black hats. Biometric recognition made it complicated but somehow impossible to steal password as they are the orangs of human body.in case of iris everyone in the world has iris pattern. Additional features of iris recognition makes it different from other biometrics like aliveness detection which fingerprint, face, ear don't possess. Stability in about two years of age is another great quality of iris.


The bars showing the trend of publications on iris recognition methods.


Iris recognition has proved that it is a very useful and better security measure as compared to other biometric systems. It is a quick and accurate way to authorize the correct person rather than intruder. Since aliveness detection is also a part of iris system, so it is nearly impossible that intruder can steal eye of authorized person and get access to secured area or system.


iris biometric is one of the most emerging biometric technology in last decade due to its stability over time, its uniqueness and noninvasive nature its is become popular for personal authentication in highly secure area ,mass transit ,air ports driving license etc. it is almost impossible to cheat the iris biometric system ,As the iris is commercially used by many private and Government organizations this is showing that the iris this area of research has reached to the maturity level. But to deal with the noisy, blurred, moving iris and distant iris images is still a big issue. More over as the cataract(white cloud) on the pupil and iris occluded the iris image and the after cataract surgery subject correct recognition is still issue. As the mobility increases the iris biometric is becoming prominent in the security of mobiles and laptops such system demands the strong illumination ,contrast and also rotation invariant system. the demand of iris biometric is also increasing in the area of cryptography for the security of data. The researcher are still finding new ways to reduced the computational cost and to increase the Correct Recognition Rate (CCR) for searching in the Large data bases such as the data bases of country people now using hybrid approaches (use multiple technique for iris recognition and fusing the score of different technique to reach to the desire level of recognition) to increase the CCR.

Source: Essay UK - http://www.essay.uk.com/free-essays/information-technology/survey-iris-recognition.php

About this resource

This Information Technology essay was submitted to us by a student in order to help you with your studies.

Search our content:

  • Download this page
  • Print this page
  • Search again

  • Word count:

    This page has approximately words.



    If you use part of this page in your own work, you need to provide a citation, as follows:

    Essay UK, A Survey On Iris Recognition Methods. Available from: <https://www.essay.uk.com/free-essays/information-technology/survey-iris-recognition.php> [28-05-20].

    More information:

    If you are the original author of this content and no longer wish to have it published on our website then please click on the link below to request removal: