BIOMETRIC SECURE ACCESS TECHNOLOGIES 1 Mr.Nitin Khachane 2 Mr.Ashish Khachane Introduction: The keyword Biometric only deals with individuality. Bio word itself indicates individual entity. The entity remains unmatched with any other entity in the world. The biometric information is secure from any third party interference. This Greek word biometric means to measure life which means analysing statistical of biological parameters. In talking about this all methodologies this term is now modified to another level of technology and now featured in to the technology. In this research work we take into account different biometric authentication techniques e..g fingerprint, iris scan. Literature Survey: Several research papers regarding the trends with modern authentication mechanisms have been published. David Evans Bailey authored a fascinating book, Virternity: The quest for a virtual eternity. The book aims to further conceptualize the aims and goals of the Virternity Project [1]. He discussed how iris is very accurate in the form of biometric. It has ability to fast scanning than finger print scanner. Overall Virternity elaborates the secure methodology of iris recognition. With efficient, right and accurate team which helps for proper implementation of the technology which is also right path for selection of Virternity system. Chapter Five of the book explicitly discusses the proposal of the Virternity project to use iris scan technology to enable secure access to the Virternity systems and assets. Current developments and the technology itself are examined as well as future potentials of this technology. Virternity: The quest for a virtual eternity by David Evans Bailey also states how the technology itself is not in its infancy and has reached a high level of sophistication. The drawbacks and potential threats to this kind of security system are also examined. According to the book scanning happens in two parts which first is identification of unique traits in the iris as well as it does analysis subsequently. To understand properly we will take example of Samsung mobiles. In Samsung note 7
infrared as well as normal cameras is used where its scanning and identification rate is at least 240 frames per second. According to Rattani s research vasculature is effective data source for eyes scan rather than iris scanning, this system uses vessel which carries our blood in the white of the eye whichh is exact opposite of iris also which is movement dependent capture is done. Vessel of eye is also unique for every individual. This could be a replacement to the iris and can be treated as the best method for recognition and authentication. As a part of the pilot project, San Bernadino Sherrif collected nearly 189 scans per day in 2016. The best investing agencies across the world like FBI,CBI have different techniques to scan faster and potentially which is easier than finger print scanner. The implementation matters on long distant iris scanning have been progressing since the beginning of the year 2015. The communication of two fellows suggest on this idea of different scanning which can scan from 12 meters to 40 meters away while scientist of Carnegie Mellon in the united states demonstrate the iris scanning from long distantt driver via vehicles mirror. Some of countries are now able to scan the eye via surveillance camera system. History of iris scanning is not too much old. Three years ago a handbook on iris reorganisation is published of researcher Kevin Bowyer and Mark Burge, they noted that roughly 15 years of research would list approximately 180 publications. Professor John Daugman presented a paper on algorithm can detect the eye when a person have contact lenses. Also eye tracking through VR headset has also begun according to CNET. They noted on eyes scanning and tracking through VR headset. Without spoofing topic iris scanning topic is incomplete. Bowyer and Burge have included the topic on pattern matching which spoof can be done by matching the patterns of different person s iris so that it can be identifies. This could be done using Gabor parameters which gives similar verification and identification.
Security scientist Jan Krissler clams that iris can be hacked using 4k resolution images which can be available from internet. In some of papers by doing eye duplication hacking have been done easily. Other way to detection includes photos, contact lens and eye duplication which are researched by Javier Galbally and Marta Gomez-Barrero and this technique are used for spoofing.biometric security gives a strong and ethical security solution and convenience methodologies to detect people identification easily. In today s scenario security is the basic need of any organisation. The biometric secure access technologies offer a strong authentication mechanism and approach towards the security application, obviously having good advantages over classical technologies like PIN password, PATTERN matching, face recognition etc. These classical technologies are of some knowledge based means that you have to do remember it. Now it depends how people will memorise it. However, there is use of mind that only a person knows how to restore that password if it is lost. But in the case of Biometric there is no need of mind or memorising; only you have to carry yourself for identification or matching. Biometric need only individual appearance, antique physical appearance for identification, commonly used biometrics are palm scanning iris scanning, finger print, face scanning, speech reorganisation etc[2]. Method ID Cards, Passports, UID s, token based identification etc. Memory codes, Numeric alpha signature) based(pin patterns, password, passwords, Benefits 1. If lost new can be issued. 2. Standardized globally and works in different nation. 1. Economic and simple type. 2. If found a problem can be changeable. Deficiencies 1. Can be easily stolen. 2. Can be distributed or shared. 3. One can have many identities. 1. Hacking is easily possible. 2. Strong pass. very hard to remember. 3. Can be shared. 4. Can be copied.
Biometric 1. Unique in their own 1. Very few cases of fake way. identity. 2. Cannot be hacked. 2. Cannot be replaced. 3. Easy to handle. 3. Due to physical 4. Multiple appearances appearancee this will not can be identify easily. secret. 5. Tough security level. Table 1. Benefits and deficiencies of the three main methods of biometric security approaches. Now it depends on the application of technology, how it demands for privacy equivalent methods is used. For high and tough security level mainly seven criteria of security is followed. Acceptability Permanence Performanc e Universabili ty Parameters Collectabilit y Uniqueness Circumventi on Figure 1: Fundamental basic parameters of Biometric secure system. As shown in Figure 1 the fundamentals of basic parameters lead to variation in the security. Uniqueness means the requirement of antique biometric secure system. It will differentiate the system to another system that how this system is unique than another system. For better explanation take example of human DNA which every human and beings have their own unique DNA code. This parameter in particular system is unique so this can be used for accessing technologies. Acceptability means choosing the field
of secure system also the Permanence means collections of different parameters and characteristics which the system will work on it. Performance parameters lead to the system for better work handling performance how our system will work. Universality means which the platform will globally handled and easily accessed by anyone. Collectability means system must have each and every characteristic parameters that will collectively verify uniqueness of the system. And at last circumvention which will decide how many time user get failure during identification process. For example DNA have strong characteristics lead to failure of verification process [3]. Now problem arise which technology have to choose for better reorganisation, to select that one that must follow these properties as shown in fig 1. Distinctiveness: Two people must have their own personal characteristics so that they can easily distinguish. Universality: Each people have their own property of characteristics. Permanence: Characteristics must be permanent and stable that in future fake identities are not possible. Collectability: These characteristics should be measurable and easily acquirable. Acceptability: Person should have tendency of acceptance on biometric system. Performance: Speed and efficiency of accuracy should be good for biometric system so that it perform well. Circumvention: fraudulent and faking people and tricky technique which can hack the system should be avoidable. Biometrics has two categories:- 1. Physical biometrics: This trait is based on direct analysis of human body. Which body parameters is measurement directly. 2. Behavioral biometrics: signature, gesture body parameters are measured indirectly. Biometric Secure Access Methodologies:- There are several techniques which are used in various fields and industries. From which verification and identification are the most famous modes. Identification: Matching is i performed in this mode. Previously entered data is
Figure 1b:General Block diagram for Secure Access Authentication compared with the new one so that identification is done. It is open type of identification means providing none of the above type of option will leads to open set of identification. Verification: In this mode, after founding the perfect matching the accuracy is measured so that exact answer is obtain. Many times it is also called detection as well as authentication where several time verification is done after many failing perfect answer are extracted. Several biometric techniques uses following flow for verification and identification. Every system has similar type of design flow. Most mandatory thing is sensor which receives the information from outside. After that entity characteristics and properties are extracted. Now these values are transferred to matching part in which comparison is carried out. This comparison is in between data base entity and received entity. After perfect matching the decision maker will verify it.
Figure 2:Generalized Process Flow These are the generalize flow of every technology. In traditional way of security in which people have to remember the password, pin or pattern and things have to keep in mind. If someone is suffering from any diseases like Alzheimer or having short term memory loss they may not to recall what they have set password. Now we will discuss advantage and drawback on each methodology for secure access technology depending on their need. 1. Finger Print Technology
Figure 3: Fingerprint Recognition This methodology is most popular in the world. Many of the mobiles and industries are using this finger print technology. This technology has good speed of reorganisation of people. Some decades before people uses ink paper method for identification but this method is not recommended for many of the organisation. After few years optical technique come which have wide range of scanning and easy to use. It has wider range than capacitive technique but due to optical phenomenon it gives some distortions. Capacitive is low cost and efficient technique and easy to integrate in to small scale but hard to handle than optical. Using ultrasound and detect the nodes of finger is good and accurate technique. This technique has ability to penetrate depth than other one[5].
It depend the finger pattern of people which will store in the data base of scanner as shown in figure 3. If a person figures get stretch mark on his/her tip of finger. This matching would get failed. This is the biggest disadvantage of this system. 2. Face Detection Face detection is done generally by photo camera or video camera. For this camera must have high resolutionss so that fake cannot access the system. Most of the mobile company have this type of function. For 3D detection some time the video cam or web cam are used but for that person must have to come closer to the system and this cam takes several images and matches the point-to-point node with the base image. Problem with the face detection is that if a person has a twin brother or sister this would be possible that they have same node point and system may get fooled. Figure 4:3D face detection
3. IRIS Biometric Iris is a new technology which uses the pupil for identification. Similar to the finger prints every person has their unique pupil and eye design. Figure 5 : Sclera Outer Boundary Most of the mobile now using iris scanner technology which is more authenticate and fast. Also it works at night i.e., at low light. As shown in figure 5 and 6 measuring length and dimension of sclera and pupil the pattern is saved. Also colour and contrast are also processed by using pigmentation of image. Size of dilator muscle is also calculated and further used for verification and identification [text4]. This technology know as polar mapping, as shown in figure 6 mapping and measurement is performed so that exact matching with accuracy is obtain. This technique is unique in their way and better than finger print. This system does not require any physical contact as finger print does. Only you have to stand near scanner and in second it will identify you.
Figure 6:Iris polar mapping technology Figure 7:Sharabat Gula after 18 years Few years ago iris helps to find lost girl in Afghan. In 1994 a person who work for national geography had taken a picture of afghan girl named Sharbat Gula in Pakistan. That image got famous due to her green eyes. After that photographer tried to find that girl for 17 years but he failed. After that explore team got a women having similar face then her photo is taken and they transfer it to FBI for iris matching and surprisingly face get matched! From this we get an idea how effective is iris scanning. Various tools which are available for iris scanning and are quite popular in forensic science are LG iris access 3000 and OKI Iris pass.
4. Speech Detection Now world came to new era of artificial intelligence in which the voice can now access the computer also computer have voice as well. Speech can be access key to technology which microphone is used to read voice. Siri and Google assistant of apple and Google work on voice. Many of the mobile have facilities to unlock the system using voice. But this is not secure because a person can do another person voice copy and may unlock the system. 5. Veins recognition Another unique and popular technique mostly used as a feature recognitionn is the veins recognition. This technique is not much popular but the algorithm is toughed one. Similar to finger print our palm veins have their different position. Each person has different physic and behaviour depending on them veins are present. Hence recognition system will detect the pattern of human palm so that comparison is done. This technique have accuracy and reliable but it take time to detect the pattern and area of palm is wide hence area required is more. Methodology for IRIS Recognition System Figure 8:Veins recognition using Palm Scanning
IRIS is now in reality, before it seems like imaginary things all over the world. Iris is very accurate in the form of biometric. It has ability to fast scanning than finger print scanner. It rapidly matches the colour ring of tissue around the pupil and these are individual unique in their properties. This scanning happens in two parts which first is identification of unique traits in the iris as well as it does analysis subsequently. To understand properly we will take example of Samsung mobiles. In Samsung note 7 infrared as well as normal cameras is used where its scanning and identification rate is at least 240 features of iris which is more than finger print. Iris scanning has two step of processing in which first is formation and registration and another is comparing or matching. A recent paper had surveying on the ocular biometric notes that infrared sensor reorganisation is more profitable than normal visual range spectrum cameras basically in low light condition this must be in consideration in the place of opting in virtual reality. Some of paper have also studied on iris recognition under degrade under visual light because of focusing problems. Also some of parties have built software algorithm but use of infrared is not ever in realistic. According to Rattani s research vasculature is effective data source for eyes scan rather than iris scanning, this system uses vessel which carries our blood in the white of the eye which is exact opposite of iris also which is movement dependent capture is done. Vessel of eye is also unique of every individual. This could be a replacement to the iris and best method. As part of pilot project San Bernadino Sherrif collecting nearly 189 scans per day in 2016. FBI like US patrolling agencies use state of the art techniques to scan faster and potentially which is easier than finger print scanner. The implementation matters on long distant iris scanning progressing from 2015. The communication of two fellows suggest on this idea of different scanning which can scan from 12 meters to 40 meters away while scientist of Carnegie Mellon in the united states demonstrate the iris scanning from long distantt driver via vehicles mirror. Some of countries are now able to scan the eye via surveillance camera system.
History of iris scanning is not too much old. Three years ago a handbook on iris reorganisation is published of researcher Kevin Bowyer and Mark Burge, they noted that roughly 15 years of research would list approximately 180 publications. Professor John Daugman presented a paper on algorithm can detect the eye when a person have contact lenses. Also eye tracking through VR headset has also begun according to CNET. They noted on eyes scanning and tracking through VR headset. Without spoofing topic iris scanning topic is incomplete. Bowyer and Burge have included the topic on pattern matching which spoof can be done by matching the patterns of different person s iris so that it can be identifies. This could be done using Gabor parameters which gives similar verification and identification. Some of the research experts have also voiced serious security concern regarding the iris based authentication mechanism. Security scientist, Jan Krissler claims that iris can be hacked using 4k resolution images which can be available from internet. Some of the research papers have tried to explain the sensitivity of error by doing eye duplication hacking. Eye duplication can be done easily to hack the system. Other ways to detect includes using high definition photos, contact lens and eye duplication techniques which have been researched by Javier Galbally and Marta Gomez-Barrero and this technique are used for spoofing. Overall virternity explains secure goal for iris recognition. With efficient, right and accurate team which helps for proper implementation of the technology which is also right path for selection of virternity system. To identify someone using IRIS recognition system, there are two methods and that are 1. Active 2. Passive
Figure 9: Active Iris Recognition In active, IRIS system activated 6 to 14 inches. when distance between user and camera system is near about In passive, here system activated when distance between user and camera system is near about 1 to 3 feet. Figure 10: Passive Iris Recognition
Figure 11: Iris Recognition Flow graph The Iris recognition flow graph has been explained in figure 11.It includes several stages to match the right iris pattern. Here, e, we describe each of them. Stage 1: Image Acquisition This is the first step for iris recognition, here need is capture the sequence of iris image using cameras and sensor with high quality resolution. This image shows the entiree part of eye, specially iris part and pupil part and later on to increase the quality of image, there applied some processing operation e.g. histogram equalization, filtering noise removal etc.
Figure 12: Eye Database Stage 2: Segmentation In this step, the expert first isolates the actual iris region in a digital eye image. This region has approximately two circle, one circle for the iris/sclera boundary and another, interior to the first, for the iris/pupil boundary. This step is performed with the help of Daugman sintegro differential method. This method used to localize the parameters of iris from eye image. Equation 1 shows the detection of inner and outer boundaries of the iris.,, f, (, ) 2 (1) Equation 2 shows the fast discrete implementation of Integro differential operator (IDO). ( ) ( ) ( ) ( ) = 1 ( ) 1 ( 1) (2) Here radius is incrementing very slowly. Equation 1 and 2 shows how the inner and outer boundary calculated.
Stage 3: Rubber sheet Normalization Normalization is used for mapping the segmentation iris into a fixed rectangular representation. Output is invariant with dimensional variation, angle of iris and pupil within the image. Rubber sheet normalization algorithm which is proposed by Daugman is used to iris normalization,. This algorithm is used to solve varies issue likelight illumination level, angle of camera and angular variation etc. { (, ), (, )} (, ) (3) Where r lies into a unit interval [0,1] and angle varies from [0,2 ] With, (, ) = (1 ) ( ) + ( ) (4) and (, ) = (1 ) ( ) + ( ) (5) (, ) and (, ) are parameters of pupil and iris respectively in direction. Figure 13: Rubber Sheet Normalization Step Stage 4. Feature extraction using DCT Discrete Cosine Transformation (DCT) features were extracted from the normalized iris. = (2 + 1) 2 (2 + 1) 2 (6) = 1 = 0 2 1 1 (7)
= 1 = 0 2 1 1 (8) Elementary frequency components of signal are decomposed by DCT and 2-D DCT of MXN matrix. In which the significant energy information is concentrated towards the left corner of the matrix, which is then called as Feature Vector (FV) matrix. Stage 5: Scanning methods to take partial coefficients Here, only few coefficients which are having maximum information of FV matrix are extracted. There are various scanning technics to extract the partial coefficients such as Zigzag, Raster Type-I, RasterType-II, Sawtooth Type-I, Sawtooth Type-II and Sawtooth Type-III Figure 14:ZigZag Scan Figure 15: Raster Type 1 San Figure 16: Raster Type 2 Scan
Figure 17: Sawtooth Type 1 scan Stage 6: Matching In this method, Mean square error (MSE) calculated between training and testing data. ( ) = = [ (,, ) (,, ] Where, Size of the FV matrix denoted by M and N respectively. After matching as the final step for a match or mismatch. and the value of mean square error, the results are compared Conclusion To identify someone using biometric recognition, there are many possible feature and algorithms. In this work, we discussed all the different scanning techniques for iris based recognition including the featuree extraction from the normalized iris. Here, the best results are achieved discrete wavelet transformation. Considering the heterogeneity of the database the approach also presents the dynamic computation of parameters required during the various stages. References 1. Bailey, D. (2017). Virternity: The quest for a virtual eternity. 1st ed. First published in Great Britain: VR Academic Publishing 2017. 2. Biometric security technology, IEEE Aerospace and Electronic Systems Magazine, Vol.21 nº 6, pp.15-26, ISSN: 0885-8985. June 2006 This work has been supported by FEDER and MEC, TIC-2003-08382-C05-02 3. A survey of biometric security system http://www.cse.wustl.edu/~jain/cse571-11/ftp/biomet/index.html 4. S. B. Solanke and R. R. Deshmukh, " Biometrics Iris recognition system A study of promising approaches for secured authentication," 2016 3rd International
Conference on Computing for Sustainable Global Development (INDIACom), New Delhi, 2016, pp. 811-814. 5. Zhenan Sun, Yunhong Wang, Tieniu Tan and Jiali Cui, "Improving iris recognition accuracy via cascaded classifiers," in IEEE Transactions on Systems, Man, and Cybernetics, Part C (Applications and Reviews), vol. 35, no. 3, pp. 435-441, Aug. 2005. 6. R. N. Rakvic, B. J. Ulis, R. P. Broussard, R. W. Ives and N. Steiner, "Parallelizing Iris Recognition," in IEEE Transactions on Information Forensics and Security, vol. 4, no. 4, pp. 812-823, Dec. 2009