The impressive amount of recent technological advancements in the area of information systems have brought along, besides the multitude of positive aspects, some negative aspects too. The most obvious one is represented by the fact that the technological innovations are prone to various categories of threats. Making sure that information stays safe, unaltered and secret is an integral part of providing technology that behaves in the manner it is supposed to. Along with researching techniques of effectively securing the communication, other aspects that also deserve to receive attention are authentication, authorization and accounting. One security aspect that has been the most intensely researched in the past from the three, is authentication. From the various methods used in verifying the identity of users, one more recent one is biometrics that can significantly heighten the safety and security of a system. However, authenticating the user into an information system may not be entirely safe all the time. The aim of this paper is to present an overview of the different plausible methods of combining cryptography related concepts and biometrics techniques. This additional attachment of cryptography has proven to make the process of gaining access to an information system much more secure. The problem tackled in this paper will be presented from a two way perspective: on one hand, it will be discussed how biometrics can make use of cryptography-specific solutions in order to enhance its powers and, on the other hand, it will be presented how cryptography aspects can use the specific biometric data of a user to generate encryption keys that are much harder to decipher or to obtain. After this methods are presented, the theoretical basis for measuring performance of a biometric system will be presented and a survey on current performance results on fuzzy vault techniques will be enumerated and described.

Enhancing security by combining biometrics and cryptography

In this digital world, the problem of recognizing a person is resolved by various biometric applications. Along with various unique physical biometric patterns like face, iris, fingerprints, etc., some of the behavioural biometric patterns like signatures, voice, gait, etc. are also used for recognizing an individual. Among all these, handwritten signatures play a major role in our daily life. In this paper, we have discussed the recent work done on online and offline modes which are the two ways of digitizing the handwritten signatures. We have done a comparison of these works based on various features and techniques used. The raw features such as the image of signature and position coordinates, time, pressure, etc. are considered in offline and online signature verification, respectively. For matching the signature pattern, many Artificial-Intelligence-based techniques and machine learning algorithms like support vector machine, hidden Markov model, neural network, fuzzy logic, deep learning, etc. have been used in developing different handwritten signature verification systems. We have also highlighted some signature verification systems which considered handwritten signatures of different languages like Arabic, Chinese, Japanese, Bangla, Hindi, English, etc. The performances of such signature verification systems lead us to do more work using other Indic-scripts-based signatures.

A Survey of Digitized Handwritten Signature Verification System

Biometric features posses the significant advantage of being difficult to lose, forget or duplicate. Hence, a FIS-based method is used for signature verification. FIS is well suited for this task due to the similarity between an individual signatures with subtle differences between each signature sample. Signature samples are collected using a tablet PC. The individuals draw their signatures usinga pressure sensitive pen on the tablet. Eight dynamic features are extracted from the signature data. These eight features are then fuzzified for training of a FIS. The system is then used to determine whether the signature is genuine or forged. A False Acceptance Rate (FAR) of 10.67% and a False Rejection Rate (FRR) of 8.0% demonstrate the promise of this system.

Ng Zhi Lun

Papers

Handwritten signature verification: Online verification using a fuzzy inference system