2. Summative account of the papers studied3
3.1. Minutiae recognition system based on genetic algorithms3 3.2. Fingerprint matching algorithms for integrated circuit cards4 3.3. Algorithm based on tree comparisons using ratio of relational distances.5 3. Comparative account of the papers studied6
A SURVEY ON FINGERPRINT VERIFICATION ALGORITHMS
Ms. Tasleem Fathima J
Fingerprints are the most widely used and trusted biometrics that uniquely identify an individual and are used for various purposes like authentication, law enforcement etc. Some of the techniques used to match finger prints are: filter based, minutiae or Galton feature’s based, correlation based, pattern matching. The most commonly used technique is minutiae based which involves identifying the various features of a fingerprint like ridge endings, ridge bifurcation, island, core, delta, crossover etc. The accurate locations and directions of these minutiae points identified are stored in the database for testing the methods proposed by the papers. There are around 20-60 minutiae points in a fingerprint. The number of minutiae points considered highly affects the efficiency of the system. The algorithms proposed by the papers under study strive to improve the efficiency of the computation involved. Each of the algorithms is application specific and has their own advantages and disadvantages. 1. INTRODUCTION
Minutiae points are the points which give the location of distinct features of a scanned fingerprint image. However, fingerprint matching by minutiae identification may not be reliable in all cases. It depends on the scanner used, the clarity of the image obtained and the interoperability of the scanning devices. Depending on the requirement and the resources available (ex. Storage space and speed requirement) the minutiae characteristics, their locations or co-ordinates directions and types are used in fingerprints recognition and verification. Using additional information like the number of ridges between two minutiae points considered can help improve the accuracy. Furthermore, not all minutiae points are considered during verification because the intermediate steps involved like binarization and skeletonization may introduce spurious points or remove certain non-prominent minutiae points. Using all the minutiae points also decreases the efficiency and accuracy of the system. During matching a minimum set of minutiae points (which varies depending on the application) considered in the input image should match with the minutiae points of the base image if they are from the same individual.
2. SUMMATIVE ACCOUNT OF THE PAPERS STUDIED
The above described steps are used by most of the fingerprint matching techniques but steps like skeletonization and binarization are optional. 3.1 Minutiae recognition system based on genetic algorithms This paper deals with fingerprint minutiae recognition and verification system. The fingerprint verification system involves extraction of minutiae points and then comparing the obtained minutiae set along with that stored already. There are various steps involved in minutiae extraction. First, the image has to be obtained. This is then followed by block direction procedure which is used to obtain the directional image using Coetzee algorithm. It is based on the intensity of pixels. Then, this image is segmented and binarized. Segmentation involves grouping of homogeneous pixels and assigning them the same label based on their gray levels. The gray image is converted to 0s and 1s which is binarization using appropriate low pass filters. The image is skeletonized such that the objects in the image are of one pixel wide. Two algorithms namely Zhong and Suen algorithm and Marthon algorithms can be used....