SignatureVerification using Neural NetworkAbstract— A novel system for protecting fingerprint and palmvein privacy by combining two different fingerprintand palm veins into a new identity.
In the enrollment, two fingerprint and palmveins are captured from two different fingers. We extract the minutiaepositions from one fingerprint and palm vein, the orientation from the other fingerprintand palm vein, and the reference points from both fingerprint and palm veins.Based on this extracted information and our proposed coding strategies, acombined minutiae template is generated and stored in a database.
In theauthentication, the system requires two query fingerprint and palm veins fromthe same two fingers which are used in the enrollment. A two-stage fingerprintand palm vein matching process is proposed for matching the two query fingerprintand palm veins against a combined minutiae template. By storing the combinedminutiae template, the complete minutiae feature of a single fingerprint andpalm vein will not be compromised when the database is stolen. Furthermore,because of the similarity in topology, it is difficult for the attacker todistinguish a combined minutiae template from the original minutiae templates.With the help of an existing fingerprint and palm vein reconstruction approach,we are able to convert the combined minutiae template into a real-look alikecombined fingerprint and palm vein. Thus, a new virtual identity is created forthe two different fingerprint and palm veins, which can be matched usingminutiae-based fingerprint and palm vein matching algorithms.
The experimentalresults show that our system can achieve a very low error rate with FRR 0.4% atFAR 0.1%. Compared with the state-of-the-art technique, our work has the advantagein creating a better new virtual identity when the two different fingerprintand palm veins are randomly chosen. Keywords— Fingerpint ,Palm vein Rrecoginization,Matlab/Simulink. I. INTRODUCTION Nowadaysa very large number of identification systems exist that are based on differenttypes of biometrics.
This includes iris, fingerprint, retina, voice, face, palmand vascular pattern recognition. This last one is quite a new emergingtechnology in the biometric field, and has been gradually adopted over theworld. The idea of using the hand vascular pattern as a biometric was firstconsidered in the early 1990s but it wasn’t until 1997 that a commercialproduct was developed. In 2000 it finally became popular when an applicationwas created for personal identification based on the vein pattern on the backof the hand 1. Since its introduction, hand vein pattern technology hasexpanded to fingers and palm based systems and was adopted in 2007 by theInternational Standard Organization (ISO) where the storage and transmission ofvascular biometric images was standardized.
The demand for secureidentification systems has increased exponentially over the last ten years.These systems are required to be very reliable but also easy to use since theirapplication is no longer restricted to high-security facilities. The advantagesof hand vein pattern recognition are due to the fact that veins lie underneaththe skin, which makes them easily accessible for the system but also hard toalter. In this perspective, the accessibility of vein pattern compared withother biometrics and its ease of use have made it a very interestingalternative for applications where a high level of security is required. It isalso a good alternative to biometric systems that require physical contactinorder to identify the individual, especially in environments such as hospitalswhere hygiene has high priority.
An identification system should be fast,simple and secure and due to its desirable advantages, vein pattern technologyis being considered into various authentication solutions for use in publicplaces (access control, time and attendance, security, hospitals).The marketfor hand vascular pattern technology is growing rapidly and today it is an areaof ongoing research that draws a lot of attention.II.
LITERATURE SURVEYThe 1 Yingbo Zhou and Ajay Kumar haveproposed a method for Human Identification Using Palm-Vein Images. This paperpresents two new approaches to improve the performance of palm-vein-basedidentification systems and they are Holistic approaches using subspace learningand Line/curve matching using vessel extraction. This approach performs verywell even with the minimum number of enrollment images (one sample fortraining). 2Xuekui Yan , Wenxiong Kang , Feiqi Deng , Qiuxia Wu have proposed a method forPalm vein recognition based on multi-sampling and feature-level fusion. Toaddress the unsatisfactory recognition performance of a single-sample approachin single biometric systems, multi- algorithm approaches have been proposed toensure that richer feature information can be extracted for better recognitionperformance. In this method we used a bidirectional matching algorithm insteadof unidirectional matching, is adopted for efficient mismatching removal. 3 Jen-Chun Lee has proposed a novelbiometric system based on palm vein image. He consider the palm vein as a pieceof texture and apply texture-based feature extraction techniques to palm veinauthentication in his work.
A 2-D Gabor filter provides the optimizedresolution in both the spatial and frequency domains, thus it is a basis forextracting local features in the palm vein recognition. He proposed aninnovative and robust directional coding technique to encode the palm veinfeatures in bit string representation. The bit string representation, calledVein Code, offers speedy template matching and enables more effective templatestorage and retrieval. The similarity of two Vein Codes is measured bynormalized hamming distance.
High accuracy has been obtained by the proposedmethod and the speed of the method is rapid enough for real-time palm veinrecognition. 4Kuang-Shyr Wua , Jen-Chun Leeb , Tsung-Ming Loc, Ko-Chin Changd , Chien-PingChang have proposed a secure palm vein recognition system. In this paper, adirectional filter bank involving different orientations is designed to extractthe vein pattern and the minimum directional code (MDC) is employed to encodethe linebased vein features in binary code.
A total of 5120 palm vein images from256 persons are used to verify the validity of the proposed palm veinrecognition approach. High accuracies (>99%) and low equal error rate(0.54%) obtained by the proposed method show that proposed approach is feasibleand effective for palm vein recognition. 5Mansi Manocha and Parminder Kaur have proposed a method for Palm VeinRecognition for Human Identification Using NN ie. Neural network. The proposedalgorithm is an alternative to currently employed palm-vein identificationapproaches that do not take advantage from the cross-level image measurements.Further improvement in the performance from the proposed approaches usingfeature discretization and image quality measurements is expected and suggestedfor the further work on the largescale palm image databases with the help ofNeural Networks.
6Gitanjali Sikka , Er. Vikas Wasson have proposed a method for Palm VeinRecognition with Fuzzy-Neuro Technique. In this paper feature is extractedusing FuzzyNeuro is used to enhance the response time and accuracy of system .The fuzzy-neuro technique is based on the combination of the fuzzy-logic (fuzzysets) and pattern recognition based feed forward neural network. This algorithmhas been made able to tackle the light variations, noise and other specificimage characteristics in the palmvein recognition systems. 7Jing-Wein Wang , Tzu-Hsiung Chen have proposed a Building Palm Vein CapturingSystem for Extraction . In this paper The performance of the accurateextraction ratio is 93.
35% but the major drawback of this system is extractionsdue to bad quality of the palm vein pattern images system may lead to the fatalerrors of the process. II. Flow Chart III. PROPOSEDSYSTEM Image segmentation isthe authorized process of dividing an image into multiple parts.
This istypically used to identify objects or other relevant information in digitalimages. Image segmentation is the process of partitioning a digital image intomultiple segments (sets of pixels, also known as super pixels). The goal ofsegmentation is to simplify or change the representation of an image intosomething that is more meaningful and easier to analyze.
Image segmentation istypically used to locate objects and boundaries (lines, curves, etc.) inimages. More precisely, image segmentation is the process of assigning a labelto every pixel in an image such that pixels with the same label share certainvisual characteristics. The result of image segmentation is a set of segmentsthat collectively cover the entire image or a set o contours extracted from theimage (see edge detection). Each of the pixels in a region is similar withrespect to some characteristic or computed property, such as color, intensity,or texture. The original image is captured with the black unwanted background.
Including the background reduced the accuracy of the original image, becausethe position of palm usually varies across different palm vein images, it isnecessary to image segmentation in region of interest (ROI) before featureextraction and matching with database. When a palm is irradiating by theuniform light source Modules:? Image Pre-processing? Finger Vein Verification? Pattern Extraction? Sobel Edge Detector? Pattern Recognition ImagePre-processing: Pre-processingis a common name for operations with images at the lowest level of abstractionboth input and output are intensity images. The aim of pre-processing is animprovement of the image data that suppresses unwanted distortions or enhancessome image features important for further processing. Finger VeinVerification: Finger veinrecognition is a method of biometric authentication that usespattern-recognition techniques based on images of human finger veinpatterns beneath the skin’s surface.
Finger vein recognition is one of many formsof biometrics used to identify individuals and verify their identity. PatternExtraction: Thesemethods, called partition of attribute values or grouping of attribute values,can be applied to decision tables with symbolic value attributes. If datatables contain symbolic and numeric attributes, some of the proposed methodscan be used jointly with discretization methods. Moreover, these methods areapplicable for incomplete data. The optimization problems for grouping ofattribute values are either NP-complete or NP-hard. Hence we propose someheuristics returning approximate solutions for such problems Sobel EdgeDetector:The Sobeloperator performs a 2-D spatial gradient measurement on an image and soemphasizes regions of high spatial frequency that correspond toedges.
Typically it is used to find the approximate absolute gradient magnitudeat each point in an input grayscale image. PatternRecognition:Patternrecognition is a branch of machine learning that focuses on therecognition of patterns and regularities in data, although it is in somecases considered to be nearly synonymous with machine learning.Patternrecognition systems are in many cases trained from labeled “training”data (supervised learning), but when no labeled data are available otheralgorithms can be used to discover previously unknown patterns (unsupervisedlearning). CONCLUSION:Anovel system for fingerprint privacy protection by combining two fingerprintsinto a new identity. In the enrollment, the system captures two fingerprintsfrom two different fingers.
A combined minutiae template containing only apartial minutiae feature of each of the two fingerprints will be generated andstored in a database. To make the combined minutiae template look real as anoriginal minutiae template, three different coding strategies are introducedduring the combined minutiae template generation process. In the authenticationprocess, two query fingerprints from the same two fingers are required. Atwo-stage fingerprint matching process is proposed for matching the two queryfingerprints against the enrolled template.
Our combined minutiae template hasa similar topology to an original minutiae template. Therefore, we are able tocombine two different fingerprints into a new virtual identity byreconstructing a real-look alike combined fingerprint from the combinedminutiae template.