Biometric Recognition-Computer Sciences Applications-Project Report, Study Guides, Projects, Research of Applications of Computer Sciences

Download Biometric Recognition-Computer Sciences Applications-Project Report and more Study Guides, Projects, Research Applications of Computer Sciences in PDF only on Docsity! Table of Content 1 INTRODUCTION............................................................................................................................... 1 1.1 BIOMETRICS ................................................................................................................................. 1 1.1.1 Requirements for a Biometric Identifier ................................................................................. 2 1.1.2 Biometric Technologies .......................................................................................................... 3 1.1.3 Applications of Biometric Systems .......................................................................................... 8 1.2 HAND GEOMETRY AND PALMPRINT ANATOMY............................................................................ 8 2 DATABASE ........................................................................................................................................11 3 HAND SHAPE RECOGNITION USING PCA ...............................................................................13 3.1 SNAPSHOT METHOD ....................................................................................................................13 3.1.1 Create Eigenspace .................................................................................................................13 3.1.2 Project training Images .........................................................................................................15 3.1.3 Identify test images ................................................................................................................15 3.1.4 Experimental Results .............................................................................................................16 4 EXTRACTION OF HAND GEOMETRY FEATURES USING IMAGE PROCESSING TECHNIQUES ............................................................................................................................................18 4.1 IMAGE PRE-PROCESSING ..............................................................................................................18 4.1.1 Binary Thresholding ..............................................................................................................18 4.1.2 Noise and peg removal ..........................................................................................................19 4.1.3 Edge detection .......................................................................................................................19 4.2 FEATURE EXTRACTION ...............................................................................................................20 4.2.1 Finger length .........................................................................................................................20 4.2.2 Finger width ..........................................................................................................................21 4.2.3 Hand and Palm length, width and Area ................................................................................21 4.3 EXPERIMENTAL RESULTS ............................................................................................................22 5 FUTURE DIRECTIONS ...................................................................................................................24 docsity.com ii List of Figures Figure 1-1 Hand Geometry Features [4]. ............................................................................ 9 Figure 1-2 Sample Palmprint[4]., Principal lines of palmprint[7] .................................... 10 Figure 2-1 Sample Hand image of the database ............................................................... 11 Figure 3-1 FRR and FAR vs threshold levels ................................................................... 17 Figure 3-2 FAR vs FRR to calculate Equal error rate ...................................................... 17 Figure 4-1 The hand image scanned by HP scanJet 5590 and Binarized image using Ostu’s method ................................................................................................................... 18 Figure 4-2 Image after removing noise Peg template to remove the pegs ........................ 19 Figure 4-3 Boundary of hand image ................................................................................. 20 Figure 4-4 Image cropping for fingertip identification and valley identification ............. 21 Figure 4-5 Features of Hand geometry ............................................................................ 22 Figure 4-6 Hand geometry Results FRR and FAR vs Threshold ..................................... 23 Figure 4-7 Equal Error Rate of hand geometry based identification ................................ 23 docsity.com 2 accessed only by a legitimate user and no one else. Biometric Systems identify users based on behavioral or physiological characteristics. Enterprise-wide network security infrastructures, secure electronic banking, investing and other financial transactions, retail sales, law enforcement, and health and social services are already benefiting from these technologies. A range of new applications can be found in such diverse environments as amusement parks, banks, credit unions and other financial organizations, enterprise and government networks, passport programs and driver licenses, colleges, physical access to multiple facilities (night clubs), and school lunch programs. 1.1.1 Requirements for a Biometric Identifier Any human physiological and/or behavioral characteristic can be used as a biometric identifier for person identification as long as it satisfies these requirements: a. Universality Which means that each person should have the Biometric. b. Distinctiveness This means that any two person should be ―sufficiently different‖ in terms of their biometric identifiers. c. Permanence This means that the characteristics should be invariant with time. d. Collectability Which means that the characteristics should be measured quantitatively. Other issues to be considered in a practical biometric system include: a. Performance Which refers to the achievable recognition accuracy and speed, the resources required, as well as the operational and environmental conditions that may affect the accuracy and speed b. Acceptability To what extend people are willing to accept and use the biometric system as part of their daily lives. c. Degree of intrusiveness How much co-operation is required from the user to collect the biometric sample. docsity.com 3 d. Circumvention How easy it is to fool the system by fraudulent techniques i.e. degree of vulnerability to fraud. e. Long-term system support DB management, re-enrollment, template updating, etc 1.1.2 Biometric Technologies A large number of biometric features are available for use in person identification and verification and can be categorized broadly as: a. Physiological Biometrics b. Behavioral Biometrics A brief introduction to the most common biometrics is provided below. i. DNA: Deoxyribo Nucleic Acid (DNA) is the one-dimensional ultimate unique code for one’s individuality, except for the fact that identical twins have identical DNA patterns. It is, however, currently used mostly in the context of forensic applications for person recognition. Due to the extensive testing and advanced technology required, it is not the most cost efficient Biometric science, but when a positive identification is needed it is the most reliable biometric technology. ii. Ear Recognition: The shape of the ear and the structure of the cartilaginous tissue of the pinna are distinctive. The features of an ear are not expected to be unique to an individual. The ear recognition approaches are based on matching the distance of salient points on the pinna from a landmark location on the ear. iii. Face Recognition: Face recognition systems identify an individual by analyzing the unique shape, pattern and positioning of facial features. The face is one of the most acceptable biometrics because it is one of the most common methods of recognition that humans use in their visual interactions. In addition, the method of acquiring face images is non-intrusive. Facial disguise is of concern in unattended recognition applications. It is very challenging to develop face recognition techniques that can tolerate the effects of aging, facial expressions, slight variations in the imaging docsity.com 4 environment, and variations in the pose of the face with respect to the camera. The attraction of this biometric system is that it is able to operate 'hands=free', limiting the amount of man-machine interaction. However, this system is highly unreliable and expensive. For example, it will not distinguish twins or triplets, not recognize the user after a haircut, and not recognize a person who changes from wearing and not wearing glasses. iv. Facial, hand, and hand vein infrared thermograms: The pattern of heat radiated by the human body is a characteristic of each individual body and can be captured by an infrared camera in an unobtrusive way much like a regular (visible spectrum) photograph. The technology could be used for covert recognition and could distinguish between identical twins. A thermogram-based system is non-contact and non-invasive but sensing challenges in uncontrolled environments, where heat-emanating surfaces in the vicinity of the body, such as, room heaters and vehicle exhaust pipes, may drastically affect the image acquisition phase. A related technology using near infrared imaging is used to scan the back of a clenched fist to determine hand vein structure. Infrared sensors are prohibitively expensive which is a factor inhibiting widespread use of the thermograms. v. Gait: Gait is the peculiar way one walks and is a complex spatio-temporal biometric. Gait is not supposed to be very distinctive, but is sufficiently characteristic to allow verification in some low-security applications. Gait is a behavioral biometric and may not stay invariant, especially over a large period of time, due to large fluctuations of body weight, major shift in the body weight, major injuries involving joints or brain, or due to inebriety. Acquisition of gait is similar to acquiring facial pictures and hence it may be an acceptable biometric. Because gait-based systems use video-sequence footage of a walking person to measure several different movements of each articulate joint, it is computing and input intensive. vi. Palmprint and Hand Geometry: Some features related to a human hand are relatively invariant and peculiar to an individual. Hand geometry is based on the fact that virtually every person hand is shaped differently and docsity.com 7 by a person’s health (e.g., cold), stress, emotions, and so on. Besides, some people seem to be extraordinarily skilled in mimicking others. xii. Other Technologies:Other technologies such as keystrokes, body odor, lip shape etc. are also under investigation for person identification. Table 1-1 Some of the biometrics are shown: a) ear, b) fingerprint, c) hand vein d), hand geometry, e) iris, f) facial thermogram, g) Face, h) Retina, i) Signature, j) Voice. Comparison of Biometric Technologies docsity.com 8 Various biometric identifiers described above are compared in Table 1.1. Biometirc Identifier Universality Distinctiveness Permanence Collectability Performance Acceptability Circumvention DNA H H H L H L L Ear M M H M M H M Face H L M H L H M Facial thermogram H H L H M H L Fingerprint M H H M H M M Gait M L L H L H M Hand geometry and palmprint M M M H M M M Hand vein M M M M M M L Iris H H H M H L L Retina H H M L H L L Signature L L L H L H H Voice M L L M L H H Table 1-2 Comparisons of biometric technologies. High, Medium, and Low are denoted by H, M, and L, respectively. 1.1.3 Applications of Biometric Systems Biometric applications fall into three main categories: a. Commercial –computer network login, access digital information, sensitive facilities, ATM, credit card, cellular phone, PDA, medical records, etc. b. Government –national ID, passport, driver’s license, welfare claims, border control, social security, etc. c. Forensic –terrorist identification, parenthood determination, missing children, criminal investigation, corpse identification, etc. 1.2 Hand Geometry and Palmprint Anatomy The physical dimensions of a human hand contain information that is capable to authenticate the identity of an individual. This information is popularly known as hand or palm geometry. Hand geometry, as the name suggests, refers to the geometric structure of the hand. Hand geometry provides for a good general purpose biometric, with acceptable performance characteristics and relative ease of deployment coupled to a low learning docsity.com 9 curve for users. There exists a wealth of information within the geometry of an individual hand. Since each human hand is unique, finger length, width, thickness, curvatures and relative location of these features distinguish one human being from others [4]. Figure 1-1 Hand Geometry Features [4]. Palm is the inner surface of a hand between the wrist and the fingers. There are many features exhibited in a palm. These features are: aspect ratio of palm; palm size and form of the lines on the palm; the hollowness of the palm. Comparing with the palm shape features, the relatively stable feature extracted from the hands is the print of palm. In particular, the lines on a person’s hand are unique to every individual; even our own two hands are never quite alike. For example, there are three principal lines caused by flexing hand and wrist, which are named as heart line, head line and life line, respectively [2]. The location and form of these principal lines in a palm are very important physical features for identifying an individual because they vary slowly from time to time. Due to the stability of these feature lines, they can be regarded as reliable and stable features to distinguish a person from others. docsity.com 12 All the images of PIEAS database are in JPEG format of size 1700 X 2340. The format of the file name is ―(USER_ID). (SAMPLE).jpg‖, (USER_ID) is the ID of the user (SAMPLE) is the sample number for the user (USER_ID) and jpg is the format of the image file. The disk size of each image is about 1Mb. In the dataset, 42 people are male, and the age distribution of the subjects is Below 30 years = 90% Older than 50 years=2% Between 30 and 50 years= 8% We collected the images in two weeks and to acquire a single image 10 sec. is required. docsity.com 13 3 Hand Shape Recognition using PCA Eigenspace projection examines images in a subspace. It is also known as Karhunen –Loeve (KL) and Principal Component Analysis (PCA) [9]. PCA is a useful statistical technique that has found application in fields such as face recognition, hand shape recognition and image compression. It is a common technique for finding patterns in data of high dimension. It projects images into a subspace such that the first orthogonal dimension of this subspace captures the greatest amount of variance among the images and the last dimension of this subspace captures the least amount of variance among the images. Once image are projected into subspace, similarity measure is used for matching. There are basically two methods of creating an eigenspace: Original method of eigenspace projection and Snapshot method. The original method leads to extremely large covariance matrices. Since calculation of the covariance matrix and the eigenvectors is computationally demanding hence we use the second method (snapshot method). 3.1 Snapshot Method There are three basic steps to identify images through eigenspace projection. In the first step we have to create an eigenspace using the training images. In the second step the training images are projected into the eigenspace. In the 3rd step, the test images are identified by projecting them into the eigenspace and then comparing them to the results of the previous step. 3.1.1 Create Eigenspace There are six substeps to create an eigenspace using training images. Let each image is stored in a vector of size N. 1........ (1) i i i Nx x x    a. Center data: First of all each of the training images must be centered. It can be done by subtracting the mean image from each of the training images as shown in equation 2. 1 1 , (2) p i i i i x x m Where m x P       docsity.com 14 The matlab code is for i=1:length(dirOutput) I = imread([fileFolder fileNames{i}]); s = regexp(fileNames{i},'(?<pid>\d*_)','names'); clabels(i)=str2num(s.pid(1:end-1)); I=imcrop(I,rec); I=imresize(I, 0.05, 'nearest'); I=double(rgb2gray(I)); [I,limda]=normaliz(I,vd,md); B = reshape(I,[],1); D=cat(2,D,B); End %Center calculation cen=mean(D,2); b. Create a data matrix: Data matrix is created by combining the centered images. The size of data matrix is NxP. Where P is the number of training images and each column is a single image. 1 2 | | ....| (3)pX x x x      The matlab code for creating the data matrix is. %Data matrix DM=repmat(cen,1,length(clabels)); D=D-DM; c. Create a covariance matrix: The third step is to create a covariance matrix. For this the data matrix is multiplied by its transpose. (4) T X X  The matlab code is. %Covariance Matrix C=D'*D; d. Compute eigenvalues and eigenvectors: In step four eigenvalues and eigenvectors are computed for Omega’ ' ' ' (5)V   The matlab code is docsity.com 17 -1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1 0 10 20 30 40 50 60 70 80 90 100 [50 individuals 5 images for tesing and training] X: 0.75 Y: 8.24 Palm and Hand Database Result using PCA Level E r r o r FAR FRR Figure 3-1 FRR and FAR vs threshold levels The equal error rate of the experiment is about 8.5 as shown in figure below. 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 X: 8.887 Y: 8.24 Palm and Hand Database Results using PCA FAR F R R EER Figure 3-2 FAR vs FRR to calculate Equal error rate docsity.com 18 4 Extraction of Hand Geometry Features using Image Processing Techniques In order to obtain better results than PCA we use another technique i.e. hand geometry based identification. Then we will combine these techniques to obtain more accurate results. Using Hand Geometry features to identify individuals has been developed in the past decades. The characteristics of hand geometry include the length and width of palm, length and width of figures, length of hand etc. there are mainly three steps hand geometry recognition i.e. Image pre-processing , Feature extraction and Matching. 4.1 Image pre-processing Image pre-processing is the first step for hand geometry recognition. 4.1.1 Binary Thresholding The hand images are scanned using ―hp scanjet 5590 scanner‖. The original image type is RGB, sometimes referred to as a "truecolor" image. The RGB image is converted to grayscale (256 gray levels) image. A sample grayscale image of our database is shown in figure 4.1.Then the grayscale images are binarized to obtain the binary images. In our proposed hand geometry based identification and verification system we use global image threshold using Otsu's method [10]. A sample binary image is shown in figure 4.2. The input image, shown in figure4.1 is in jpeg format. Figure 4-1 The hand image scanned by HP scanJet 5590 and Binarized image using Ostu’s method docsity.com 19 4.1.2 Noise and peg removal After obtaining the binary images it is necessary to remove the noise form the images. There are different techniques to remove the noise from images. In our proposed system, we used median filter to remove the noise. Median filtering is a nonlinear operation often used in image processing to reduce "salt and pepper" noise. To remove the pegs from the images we use a peg removal template such as shown in figure below. Figure 4-2 Image after removing noise Peg template to remove the pegs 4.1.3 Edge detection After noise removing the next step of image preprocessing is edge detection. In our proposed system, Sobel method is used for edge detection [10]. The Sobel method finds edges using the Sobel approximation to the derivative. It returns edges at those points where the gradient of the input image is maximum. docsity.com 22 starting from the index finger right most valley point the image is traversed along y axis until we found bottom most point at which the value obtained is 1. w2 w3 w1 w4 L2 and w6 L4 and w8 L1 and w5 L3 and w7 Palm wdith Palm lengthHand length Figure 4-5 Features of Hand geometry 4.3 Experimental Results The system has been tested on 500 images from 50 individuals i.e. 10 images from each individual. Out of ten images from each individual, first five images are used for training and the remaining five images are used for testing. The false acceptance and false rejection rates are shown in figure 4.7 below. The second row of the table shows the false acceptance rate according to the threshold level. Similarly the third row of the table shows the false rejection rate according to the threshold level. The optimal threshold value is obtained by plotting the FAR and FRR against threshold levels from minus 1 to plus 1. The optimum threshold value that is calculated is about 0.6 as shown in figure 4.7. The results of the experiment are shown in figure below. At 0.6 threshold value the error rate is about 9.1. docsity.com 23 -1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1 0 10 20 30 40 50 60 70 80 90 100 50 Person 5 Images for trainnig 5 images for testing X: 0.6 Y: 9.04 Threshold E r r o r Palm and Hand Database Result using Hand Geometry FAR FRR Figure 4-6 Hand geometry Results FRR and FAR vs Threshold 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 X: 9.211 Y: 9.04 FAR F R R Palm and Hand Database Result using hand geometry Equal error rate Figure 4-7 Equal Error Rate of hand geometry based identification docsity.com 24 5 Future Directions Our future direction is to improve the accuracy of Hand geometry based identification system. For this we will consider some more features of the hand for example increase the no. of finger widths and aspect ratios. We will also perform wavelet analysis for palm texture recognition. In our future work we will also update the PIEAS database for time dependency analysis. docsity.com