A Novel Approach to Fractal Dimension based Fingerprint Recognition System

Size: px

Start display at page:

Download "A Novel Approach to Fractal Dimension based Fingerprint Recognition System"

Curtis Wiggins
5 years ago
Views:

1 Volue: 03 Issue: 04 Apr p-issn: A Novel Approach to Fractal Diension based Fingerprint Recognition Syste Chiteranjan Sahu 1, Vinay Jain 2 1M.E. Scholar, Dept. of Electronics and Telecounication, SSGI, SSTC, Bhilai 2Associate Professor, Dept. of Electronics and Telecounication, SSGI, SSTC, Bhilai *** Abstract Fingerprint recognition syste ay be defined as A odus operandi to identify the individual fro his or her fingerprint identity for security or privacy to identify the criinals for evidence in court laws. A fingerprint recognition schee is a coputer appliance for involuntarily identify a person by digital iage. Bioetric syste takes two distinctiveness characteristics are behavioral and physiological bioetric inforation. The behavioral bioetric paraeters are signature, gait, speech as well as keystroke; each and every one of these paraeters is change with age and atosphere. But in physiological character for exaple face, fingerprint, pal print and iris cannot transfor throughout the era of a person. Fingerprints are extensively used in daily life owing to its viability, uniqueness, stability, precision, consistency, and adequacy. The iutability and exclusivity are distinctive characteristics of fingerprints that ake it dependable feature for identify the people in life span. A fingerprint looks siilar to a pattern found on a fingertip. A good attribute quality of fingerprints contains 25 to 80 finer points depending on the sensor resolution and finger position on the sensor. The counterfeit inutiae are false ridge break for the reason that of insufficient aount of ink use into it or cross-connections by reason of over inking. In this paper, a 2. BIOMENTRICS novel approach to finger print based recognition syste has been introduced. New features has been calculated for fingerprint recognition syste and the results obtained hasve been copared with existing ethod. Key Words: Finger print recognition, Fractal diension. 1. INTRODUCTION Nowadays, iportant applications are based on fingerprint recognition priarily is autoatic fingerprint identification syste (AFIS). In AFIS syste, sall part of discriinative features is utilized. The other discriinative inforation available on fingerprint iages into atching stage can strongly ephasize the individual fingerprints and iprove the perforance for fingerprint systes on large scale databases[1]. The vital step in fingerprint recognition procedure, which affect the syste accuracy is atching aong teplate and query fingerprint. The atching algoriths classify into three types: inutiae-based approach, correlation based ethod and feature-based technique. Conversely, the score of these algoriths is not high especially in case that the fingerprint is taken in siilar finger because they are rotate or intersection is too sall. Fingerprint detection is one of the popular and reliable personal bioetric identification ethods. It is well known that iage enhanceent is extreely iperative step to ensure the extraction of reliable features. Bioetric detection fro a print prepared by an ipression of the ridges in the skin of a finger; often use as proof in criinal investigations. A bioetric syste is in essence patternrecognition schees that recognize a person based on a feature vector that feature vector is usually store in a database after being extracted. Match the finger prints, one that is already in the database of the sensor and second the fingerprint that we enrolled in the sensor currently. Get the atching score and decide the result on the atching score basis, whether the fingerprint is atched or not. To avoid the coplexity with the persistent ethods and for copensate to price factor of fingerprint detection, a new iage processing technique to identify the fingerprint is projected here. This ethod requires a fingerprint iage to exaine the fingerprint representation in order to get iages and a PC to process those iages. With the ai of this technique is straightforward, oney-spinning and provides accurate results. In the precedent, several iage processing technique have been developed to spot the iage of a person of diverse features using nuerous score atching techniques. Alost all theory, digital technique that authenticates the iages of fingerprint was used. Few iitate the conventional police routine of atching finger points others are using directly saple-atching devices; and still others are a bit ore unique. Soe verification approach can recognize when a live finger is accessible; a few can t. Fingerprint recognition, the ost 2016, IRJET ISO 9001:2008 Certified Journal Page 67

Volue: 03 Issue: 04 Apr-2016 www.irjet.net p-issn: 2395-0072 iperative step which affects the syste accurateness in atching between teplate and query fingerprint.

2 Volue: 03 Issue: 04 Apr p-issn: iperative step which affects the syste accurateness in atching between teplate and query fingerprint. A inutia atching is well-known and broadly used technique for fingerprint atching, its strict equivalence with the way that forensic experts evaluate fingerprints and its approval as a proof for identify the people in the courts of law in alost all country. Solitary of the ost coon application of device pattern recognition is autoatic fingerprint detection, because of this; there is a popular delusion that fingerprint recognition is a totally solved proble. On the converse, fingerprint detection is at rest a coplex and very difficult recognition task. A great assortent of fingerprint devices is accessible than any of the other bioetric. As the worth of these devices and processing costs drop, using fingerprints for user recognition is gaining acceptance regardless of the criinal disgrace. Fingerprint verification is a superior choice for in-hoe systes, wherever we can give user adequate clarification and training, and where the syste or a device operates in a prohibited environent. It is not aazing that the workspace access application area see to be based alost absolutely on fingerprints, the otive is that the relatively low-priced, sall agnitude, and ease of integration of fingerprint confiration devices. Geoetry of hand involves analyze and easure the nature of the hand. These bioetric create a good equilibriu between perforance distinctiveness and are relatively easy to use. It ay be appropriate in that way where there are ore users or where users access the syste once in a while and are perhaps less disciplined in their approach to the syste. Accuracy can be high if preferred and bendable perforance tuning and configuration can provide accoodation for a wide range of applications. Organization is use hand geoetry reader in various scenarios like include tie and attendance recording, where they have proved treendously popular. Ease of integration into other syste and process, tied with accessibility, and ake hand geoetry as an obvious first step for nuerous bioetric project. Figure 1.1 represents features of fingerprint iages. Fig -1: Finger print details (a) Fingerprint features (b) Fingerprint patterns 3. FINGERPRINT RECOGNITION SYSTEM A bioetric is the ost secure and convenient authentication tool for identification. It cannot be borrowed, stolen, forgotten, and forging one is practically ipossible. The Fig 1 describes the process involved in bioetric syste for security. In this syste, the fingerprints iages are processed through bioetric devices then enrolled the iage and finally store it into database and sae process applied for verifying the iage. The table 1.1 reveals the coparison between different bioetric syste through its ease of use, accuracy, cost, acceptance, error incidence, security level, long-ter stability[2]. 2016, IRJET ISO 9001:2008 Certified Journal Page 68

3 Volue: 03 Issue: 04 Apr p-issn: Required security level Long-ter stability 4. BOX-COUNTING METHOD Fig -2: Block diagra of basic fingerprint recognition syste, (1) capture the selected bioetric; (2) process the bioetric and extract and enroll the bioetric database; (3) store the teplate database in a local repository, a central repository, or a portable token such as a sart card; (4) live-scan the bioetric; (5) process the bioetric then extract the bioetric teplate; (6) atch the scanned bioetric against stored databases; (7) provide a atching score to business applications or other applications; (8) record a secure audit trail with respect to syste use. Table -1: Coparison between different bioetrics *The large nuber of factors involved akes a siple cost coparison ipractical. Characteristi c Fingerprint s Hand Geoetr y Ease of Use Low Error incidence Dryness, dirt, age Hand injury, age Retina Iris Face Signature Voice Glasses Poor Lightin g Lighting, age, glasses, hair Changing signature s Noise, colds, weathe r The box-counting ethod is based on the nuber of boxes of size essential to fill the whole area of an iage[3]. Changing the size of way that the nuber of boxes also changes, in other conditions, the lesser the size of the bigger the nuber of boxes, at the sae tie as the larger the size of, the fewer the nuber of boxes. The boxcounting techniques define the fractal diension of an object by the expression: For coputer appliance the data is usually discretized. The pixel covering ethod is proposed to estiate the FD of fractal binarized iages whose points is represented by 1, while the background is represented by 0. The iage is Y is divided into square with width. represent the iniu nuber of sets with radius less than or equal that covers Y. Hence, a group of data is obtained and the FD is estiated by changing the value of it is the slope of the line derived fro these data using the least squares linear regression. It is not easy to calculate the box diension of a locally bounded subset, where value of (1) Therefore, we can estiate the using box counting diension. Hence, we proposed new estiation ethod to calculate. Very Accuracy 5. TRAINING Cost * * * * * * * User acceptance The fingerprint iages are taken for creation of database the average fractal values can be store in the database of the iages. Database used for training is DB1 FVC2002. Many fingerprint iages are low quality. Size of each iage is 388x374 pixels and its resolution is 96 dpi. FVC2002 DB1 has 100 fingerprint iages. The iages are store in database. In training set, the iage is read then 2016, IRJET ISO 9001:2008 Certified Journal Page 69

Volue: 03 Issue: 04 Apr-2016 www.irjet.net p-issn: 2395-0072 preprocessed by the ethods used in preprocessing (enhanceent, binarization and thinning) this will describe above.

4 Volue: 03 Issue: 04 Apr p-issn: preprocessed by the ethods used in preprocessing (enhanceent, binarization and thinning) this will describe above. After preprocessing the feature of the iage can be extracted and calculate their FD values and also calculate the ean FD values. In training phase, we store the person s inforation (like nae, address etc.) to the database for identify the persons. The database iage is shown in fig 4.2 and training phase is shown in fig 3. FDROI FDROI FDROI FDROI FDROI FDROI FDRROI FDROI FDROI LCROI LCROI LCROI LCROI Fig -3: Training phase 6. RESULTS AND DISCUSSION In this section, the proposed ethod of finger print recognition has been evaluated using fractal diension. In these ethod total twenty four features of every finger in database has been extracted. On the basis of their FD values a ean value of the features has been calculated. First four features i.e. ridge terinations (RT), bifurcations count (BC), inutia count (MC) and inutia count after false inutia reoval (MCFMR), are based on the texture of the fingerprint after preprocessing. The iages represented in table 1 represent features value for fingerprint recognition Table -2: Feature value of Proposed Method Features Iage1 Iage2 Iage3 Iage4 Iage5 RT BC MC MCFMR Table -3: Features and FD value of Proposed Method Features Iage1 Iage2 Iage3 Iage4 Iage5 FDROI LCROI LCROI LCROI LCROI LCROI LCROI LCROI Feature Mean 7. CONCLUSIONS Fingerprint recognition systes work well in atching systes that reveals the atching or non atching ratio between teplate and input iage. Instead of this, the fingerprint recognition systes also work well in different orientation fields. Fingerprint recognition algoriths fail under those conditions where huans need to identify other people while do not analyze how the object causes neurons in the fingerprint iage and then how the signals are interpreted in the brain using neural network approach. This kind of detection ethod can be used to identify criinals fro their specific features. FDROI , IRJET ISO 9001:2008 Certified Journal Page 70

5 Volue: 03 Issue: 04 Apr p-issn: REFERENCES [1] K. Ito, A. Morita, T. Aoki, T. Higuchi, H. Nakajia, and K. Kobayashi, A fingerprint recognition algorith using phase-based iage atching for low-quality fingerprints, in IEEE International Conference on Iage Processing 2005, 2005, vol. 2, pp. II 33. [2] A. T. Gowthai and H. R. Maatha, Fingerprint Recognition Using Zone Based Linear Binary Patterns, Procedia Coput. Sci., vol. 58, pp , [3] M. Ge, Q. Lin, and W. Lu, Realizing the Box-counting Method for Calculating Fractal Diension of Urban For Based on Reote Sensing Iage, in 2006 IEEE International Syposiu on Geoscience and Reote Sensing, 2006, pp , IRJET ISO 9001:2008 Certified Journal Page 71

TensorFlow and Keras-based Convolutional Neural Network in CAT Image Recognition Ang LI 1,*, Yi-xiang LI 2 and Xue-hui LI 3

2017 2nd International Conference on Coputational Modeling, Siulation and Applied Matheatics (CMSAM 2017) ISBN: 978-1-60595-499-8 TensorFlow and Keras-based Convolutional Neural Network in CAT Iage Recognition