Thumb Impression Based Security System

Abstract

Our project is about “Thumb Impression Based Security System”. This is basically a software/stimulation project which covers the major areas like biometrics and digital image processing where as MATLAB is the core software to be used. The motivation factor behind this project is the current security issues in Pakistan where a reliable Biometric system is needed. There are a number of options available such as Speech, Iris, DNA, Palm etc but Thumb is the most reliable and easy to acquire. Automatic fingerprint identification system (AFIS) also known as biometric system is commonly being used for the identification of an individual due to distinctive features of the finger prints of human beings. Automatic Fingerprint Identification Systems (AFIS) are widely used for personal identification due to uniqueness of fingerprints. Fingerprint matching results and their accuracy depends on presence of valid minutiae. Different approaches to disaffect have being adapted in this verification system out of which one technique is minutiae based fingerprint matching technique. However the accuracy of this system is mostly depends on the availability of valid minutiae therefore in presence of false minutiae the techniques may not be proved accurate or trustworthy. In this project, we design and implement a complete automated fingerprint identification system which is invariant to rotation and scaling. A new method for false minutiae removal and true minutiae validation will be implemented to increase the accuracy of implemented system. We propose a windowing post-processing method that takes into account the neighborhood of each minutia within defined window and check for minutia validation and invalidation. In support of our method we also outline a complete preprocessing system based on new segmentation technique to help extract region of interest accurately from the given images of fingerprint. In order to check the validity of implemented system and comparison with already published algorithms, we use publicly available fingerprint databases FVC 2002 and FVC 2004.