Design of Biometric Fingerprint Image Enhancement Algorithm by Using Iterative Fast Fourier Transform

Among all the minutia based fingerprint identification system, the performance depends on the quality of input fingerprint images. In this paper, we have designed and implemented an algorithm of fingerprint image enhancement by using Iterative Fast Fourier Transform (IFFT). We have designed an approach for removing the false minutia generated during the fingerprint processing and a method to reduce the false minutia to increase the efficacy of identification system. We have used fingerprint Verification Competition 2006 (FVC 2006) as a database for implementation of proposed algorithm. Experimental results show that the results of our enhancement algorithm are better than existing algorithm of fast Fourier transform.


I. INTRODUCTION
Biometrics provides automated method to identify a person based on physiological or behavioural characteristics. Among the features measured are face, fingerprints, hand geometry, handwriting, iris, retina, vein, and voice. Biometric technologies are becoming the foundation of an extensive array of highly secure identification and personal verification solutions. As the level of security breaches and transaction fraud increases, the need for highly secure identification and personal verification technologies is becoming apparent. Each approach uses some aspect of an individual's physical or behavioral attributes as a means of authenticating the individual's identity. According to [4] the most pervasive biometric in use today is fingerprint, a physical biometric. [6] Had given the concept that fingerprints have a unique pattern of ridges and furrows. Fingerprint readers use this uniqueness to generate a patternrarely do they actually use the full print for identificationbased on areas where line print lines merge, fork or looplike the round "whorl" you can find in the finger prints. This pattern is stored in a databaseeither in a remote computer or in the device itself. When a person scans a print, this device compares the pattern generated by the print with one in the database to make a positive identification. [13] Discussed that there are various phases of fingerprint identification are in existence as image loading, image enhancement, normalization, thinning, minutia marking and minutia extraction. Fingerprint image quality is a vital issue to measure the performance of fingerprint identification system. So quality assessment of fingerprint data leads to identify the fingerprints in a better way. The main purpose of such procedure is to enhance the image by improving the clarity of ridge structure or increasing the consistence of the ridge orientation. In noisy regions, it is difficult to define a common orientation of the ridges. The process of enhancing the image before the feature extraction is also called pre-processing. According to [9] for this purpose generally two methods are used. First is normalization, this is a method to improve the image quality by eliminating noise and correcting the deformations of the image intensity caused by non-uniform finger pressure during the image capture. The idea of normalization consists in changing the intensity of each pixel. The normalization preserves the clarity and contrast of the ridges; however it is not able to connect broken ridges or improve the separation of the parallel ridges. Second is transformation. In particular, for detection of high or low frequencies. As the ridges have structure of repeated and parallel lines, it is possible to determine the frequency and the ridge orientation using transformation. According to [4]. Ridge Thinning is to eliminate the redundant pixels of ridges till the ridges are just one pixel wide. After the fingerprint ridge thinning, marking minutia points is relatively easy. In general, for each 3x3 window, if the central pixel is 1 and has exactly 3 one-value neighbors, then the central pixel is a ridge branch. If the central pixel is 1 and has only 1 one-value neighbor, then the central pixel is a ridge ending.

II. PREVIOUS WORK
This section summarizes the literature on fingerprint, palm and iris identification, elaborated with various normalization and enhancement techniques used for quality improvement of identification system. (Roger Clarke, 1994) identified need of identification and various factors required for human identification. According to this paper various identification measures as names, codes, knowledge-based, token-based and biometrics had taken into account. (Hong et al., 1996) emphasizes that Gabor filter has optimal joint resolution in both spatial and frequency domains. (Kansaei et al. 1997)

IV. FALSE MINUTIA REMOVAL
If we match minutia based on matching points it may generate some extra minutia (false minutia) and erroneous matching points which can decrease the false rejection rate. So, it is necessary to remove false minutia as well as false matching points. Generally, fingerprint geometric variations are limited as rotations and translations. With the consideration of this fact set the k th minutia as origin and align its direction to zero (along x) and then accommodate all other minutia points in the fingerprint to the new origin The difference between minutia as origin and all other minutia origin direction is reflected as the orientation of each minutia and it is adjusted with the origin minutia. Minutia position value toward bottom right are positive while angle value are anti-clockwise from bottom to the top of the x axis on the right within [0,pi]and are clockwise from bottom to top of the x axis on the left within [0,-pi]. Based on this observation, we select a value of majority orientation, length and keep the matching pairs that have the majority orientation and length. This reduces the number of false minutia and matching points. As earlier discussed in the previous section proposed technique can improve the quality of poor fingerprint images especially for light/dark fingerprint impressions by using the statistics of the global texture unit information. Sometimes for such situation, just using the texture filtering cannot give the satisfactory results. This may cause the erroneous false minutia in the minutia extraction process. As discussed above if we apply the orientation property of fingerprint the problem can be alleviated largely.

V. MINUTIA MATCH
Given two set of minutia of two fingerprint images, the minutia match algorithm determines whether the two minutia sets are from the same finger or not. An alignment-based match algorithm is used in our work. It includes two consecutive stages. First is alignment stage and second is match stage. In alignment stage two fingerprint images to be matched, choose any one minutia from each image, calculate the similarity of the two ridges associated with the two referenced minutia points. If the similarity is larger than a threshold (we have considered 90% in this paper), transform each set of minutia to a new coordination system whose origin is at the referenced point and whose x-axis is coincident with the direction of the referenced point. In Match stage, approach is to elastically match minutia is achieved by placing a bounding box around each template minutia. If the minutia to be matched is within the rectangle box and the direction discrepancy between them is very small, then the two minutia are regarded as a matched minutia pair. Each minutia in the template image either has no matched minutia or has only one corresponding minutia.

VI. EXPERIMETATION RESULTS
We have used four datasets DB1, DB2, DB3 and DB4 from the fingerprint database i.e. Fingerprint Verification Competition 2004 (FVC 2004). Each of these four disjoint datasets is collected by using different sensors. This database consists of 1800 fingerprint images those have been used in this research. In FVC2004, data collection is done with introducing difficulties as wet/dry impressions, rotated fingerprint, light/dark fingerprint impression etc. The performance of the proposed fingerprint enhancement algorithm has been evaluated on FVC2004 DB1, DB2, DB3, DB4 public domain fingerprint databases. Results of proposed enhancement algorithm are shown in figures 1.1 and 1.2. For an image database, each sample is matched against the remaining samples of the same finger to compute the False Rejection Rate. If the matching g against h is performed, the symmetric one (i.e., h against g) is not executed to avoid correlation. All the scores for such matches are composed into a series of Correct Score. Also the first sample of each finger in the database is matched against the first sample of the remaining fingers to compute the False Acceptance Rate. If the matching g against h is performed, the symmetric one (i.e., h against g) is not executed to avoid correlation. All the scores from such matches are composed into a series of Incorrect Score. Our algorithm tests all the images without any modification or fine tuning for the database. Proposed algorithm can better differentiate imposturous minutia pairs from genuine minutia pairs in a certain confidence level.

CONCLUSION
This paper describes a fingerprint enhancement method based on iterative fast fourier transformation. Image enhancement is critical to the quality of any fingerprint identification system. In this paper, we have shown the comparative study of fast fourier enhancement algorithm and iterative fast fourier enhancement algorithm. It has been shown with the help of a performance graph. Experiments on a public domain fingerprint database (i.e. FVC 2004) demonstrates that the use of minutia descriptors leads to an order of magnitude reduction in the false accept rate without significantly affecting the genuine accept rate. Based on the observation of good quality rolled images, the ridge and valleys intervals of each image are considered in order to select the Region of Interest (ROI) for effective enhancement. Sometimes, during fingerprint enhancement false minutia may be generated. In this paper, we have designed a method to remove the false minutia in order to maintain the efficacy of enhancement mechanism.