How to estimate artifacts on an iris image in polar domain is an important question for any iris recognition system which pursues high recognition rate as its goal. In literature, there are many different existing algorithm that estimate iris occlusion in either Cartesian or polar coordinate. In this paper, our goal is not to propose another new method to compete with existing method. Rather, our goal is to propose a new algorithm which can take any iris mask estimated by existing algorithm, and refine it into a much more accurate mask. In this way, our proposed method could co-work with any other existing algorithm and improve iris recognition performance. Experimental results show our proposed method can improve iris recognition rate by a great lead compared to the performance of the system using the unrefined iris masks.
BioHashing generates a BioCode from a user's biometric features by projecting them onto user-specific random vectors and then discritizes the projection coefficients into zero or one. Since biometric features are distorted by the non-invertible transforms and template matching is performed in a transformed state, it has been claimed that BioHashing is oneway and a BioCode is as secure as a hashed password. However, we disprove it by showing that even without a genuine's private random vectors, a preimage of a BioCode is easily calculated from a lost BioCode and an attacker can gain an illegal access to a system.
Biometrics is a deeply studied and highly developed technology. While biometric systems have been used primarily in limited applications requiring high security tasks like criminal identification and police work, more recently they have been receiving increasing demand for person recognition applications. In spite of all developments in biometrics, there is no study investigating relationships between biometric features in the literature. This study presents a novel intelligent approach analysing the existence of any relationship among fingerprints and face parts. Proposed approach is based on artificial neural networks. Developed system generates the stationary face parts of a person including eyebrows, eyes and nose from only one fingerprint image of the same person without knowing any information about his or her face with the errors among 1.4 % and 4.8 %. The satisfactory results have indicated that there are close realitionships among fingerprints and faces. Improving of the proposed system is still sustained for the purpose of analysing and modelling of this relationship for the future developments in biometrics and security applications.
The dramatic growth in practical applications for iris biometrics has been accompanied by relevant developments in the underlying algorithms and techniques. Among others, one active research area concerns about the development of iris recognition systems less constrained to users, either increasing the imaging distances, simplifying the acquisition protocols or the required lighting conditions. In this paper we address the possibility of perform reliable recognition using visible wavelength images captured under high heterogeneous lighting conditions, with subjects at-a-distance (between 4 and 8 meters) and on-the-move. The feasibility of this extremely ambitious type of recognition is analyzed, its major obstacles and challenges discussed and some directions for forthcoming work pointed.
This paper presents writer authentication using features of handwritten single characters, PIN words and signatures. The kinematics and dynamics of handwriting movement were recorded with a novel ballpoint pen equipped with a diversity of sensors mounted inside the pen. The time series provided by five different sensor channels including refill and finger grip pressures, acceleration and tilt data was analyzed by using a DTW algorithm. To speed up computation a "Reduced Dynamic Time Warping RDTW" technique was applied which is based on the sum of sensor channels and a two steps down-sampling of the multi-dimensional time series. Preliminary results have shown that processing time and memory size could significantly be reduced without deterioration of authentication performance using single characters, signatures and PIN words. Excellent accuracy in recognition was achieved which is mainly due to RDTW technique and the high quality of data sampled by a novel pen device.
