The Research On Fingerprint Identification System

AFIS(Automated Fingerprint Identification System) is more and more emphasized recently,although many contributions have been achieved in finger- print recognition,by now there are many problems in the identification process as follows:There is no uniform fingerprint recognition technology standard about system design and implement,thus technology integration and evaluation become very hard.And in the algorithm, it is repeated and inaccurate to extract fingerprint singular points.Minutiae points extraction is inefficient. Fingerprint adjusting and matching methods are difficult and complex.This paper studies some essential points of fingerprint recognition and introduces some improved algorithms.Firstly,this paper designs the logic modules and function modules.Secondly,it establishes a automatic,efficient and opened auto fingerprint identification system which supplies some interfaces for multiple pattern recognition.Then this paper discusses the problems of fingerprint singular and minutiae points extraction,images adjusting and matching methods.In the singular points extraction algorithm,a new algorithm is presented based on the arithmetic from Poincare.This algorithm shows less computational complexity,more precision and more efficiency.A new minutiae extraction method based on a ridge following algorithm in which the minutiae is directly detected from gray-scale images is introduced.In the matching algorithm,this paper presents a improved images adjusting method and matching algorithm that is more robust to nonlinear deformation between fingerprint images.Experiments show that the speed and precision of the fingerprint matching are improved effectively.There are 6 chapters in the paper.Chapter 1 is introduction.Chapter 2 is the description of AFIS opened frame work.Chapter 3 is the fingerprint automatic gathering and the extraction of singular points.Chapter 4 is the research on the fingerprint minutiae extraction from gray-scale images.Chapter 5 studies the fingerprint adjusting and matching algorithm.The last is the summarization and prospects.The main contents are as follows:(1)Design AFIS frame work and interface definition for establishing a automatic,efficient and opened system.This paper firstly analyses the fingerprint identification theory and technology,then designs the logic modules and function modules,finally establishes fingerprint recognition system.(2)Realize the fingerprint automatic gathering based on the PC and the fingerprint gathering instrument.(3)Research on singular point extraction method speedly and accuratelyThe fingerprint classification method is usually based on the number,type and position of the singular points,so it is important to detect core points and delta points credibly.In addition,some automatic fingerprint matching methods are based on singular points,Therefore automatic singular points extraction is an extremely critical step in the whole system.Because of the low quality of fingerprint images,it is still difficult to position core points and delta points accurately.In this paper,a robust method for the detection of fingerprint singular points is presented based on the arithmetic from Poincare.The traditional methods detect singular points through the whole image repeatedly with some changeable windows,then compute the average as the final result.The improved method extracts the points synchronously on all the sampled images with different windows,then regards the maximum curvature points as the final core points.This algorithm compared with traditional methods shows less computational complexity,more precision and more robustness.Our algorithm provides a feasible foundation of theories and technologies for the research of fingerprint singular point extraction method.(4)Research on fingerprint minutiae extraction method based on gray-scale imagesIn this paper,an initiated approach to automatic minutiae detection is proposed.According to this scheme,the minutiae is directly detected from gray-scale image.This technique is based on a ridge following algorithm that follows the image ridgelines until a termination or a bifurcation occurs.The results achieved have been compared with those obtained through some known approaches which require binarization and thinning.The comparison shows the superiority of our technique in terms of speed and accuracy.The low computational complexity makes this method particularly suitable for applications where efficiency is a primary issue.(5)Fingerprint matching methodFingerprint matching is one of the most important problems in AFIS.It is the last step of all the identification process.Most automatic fingerprint verification systems are based on minutiae matching.Presently there are two popular minutia points:ending points and bifurcation points.In this paper core points and delta points are considered in the matching algorithm.We adjust the fingerprint images based on singular point.If there are the right points which can be used as datum mark,we can directly calculate the rotation and parallel translation parameters so as to adjust the right position for the following matching.In this paper,we proposed a minutia matching algorithm based on polar coordinates.In addition,we used changeable sized bounding box to make our algorithm more robust to nonlinear deformation between fingerprint images.Experiments show that the speed and precision of the fingerprint matching are improved effectively.If there is no singular point,we adopt the algorithm based on the curve fitting to confirm the datum mark,adjust the fingerprint images and do the matching.Experiments done on a set of fingerprint images show that our algorithm is fast and has high accuracy and provides a feasible foundation of theories and technologies for the research of auto fingerprint identification system deeply.