Showing papers by "Vincenzo Piuri published in 2009"

Agent-based image iris segmentation and multiple views boundary refining

Abstract: The paper presents two different methods to deal with the problem of iris segmentation: an agent-based method capable to localize the center of the pupil and a method to process the iris boundaries by a multiple views approach. In the first method, an agent corresponds to the coordinates of a specific point of analysis in the input image. A population of agents is deployed in the input image, then, each agent collects local information concerning the intensity patterns visible in its region of interest. By iterations, an agent changes its position accordingly to the local properties, moving towards the estimation of the pupil center. If no available information is present in its region of interest, the agent will move itself along a random walk. After few iterations, the population tends to spread and then concentrate in the inner portion of the pupil. Once the center of the pupil has been located, the inner and outer iris boundaries are refined by an approach based on multiple views analysis. This method starts with a set of points that can be considered as an approximation of the pupil center. For each point, a detailed estimation of the iris boundaries is computed, and the final description of the iris boundaries is obtained by merging all the obtained descriptions. The two methods were tested using CASIA v.3 and UBIRIS v.2 images. Experiments show that the proposed approaches are feasible, also in eye images taken in noisy or non-ideal conditions, achieving a total error segmentation accuracy up to 97%.

A Multi-biometric Verification System for the Privacy Protection of Iris Templates

Abstract: Biometric systems have been recently developed and used for authentication or identification in several scenarios, ranging from institutional purposes (border control) to commercial applications (point of sale). Two main issues are raised when such systems are applied: reliability and privacy for users. Multi-biometric systems, i.e. systems involving more than a biometric trait, increase the security of the system, but threaten users’ privacy, which are compelled to release an increased amount of sensible information. In this paper, we propose a multi-biometric system, which allows the extraction of secure identifiers and ensures that the stored information does not compromise the privacy of users’ biometrics. Furthermore, we show the practicality of our approach, by describing an effective construction, based on the combination of two iris templates and we present the resulting experimental data.

Neural-based iterative approach for iris detection in iris recognition systems

Abstract: The detection of the iris boundaries is considered in the literature as one of the most critical steps in the identification task of the iris recognition systems. In this paper we present an iterative approach to the detection of the iris center and boundaries by using neural networks. The proposed algorithm starts by an initial random point in the input image, then it processes a set of local image properties in a circular region of interest searching for the peculiar transition patterns of the iris boundaries. A trained neural network processes the parameters associated to the extracted boundaries and it estimates the offsets in the vertical and horizontal axis with respect to the estimated center. The coordinates of the starting point are then updated with the processed offsets. The steps are then iterated for a fixed number of epochs, producing an iterative refinements of the coordinates of the pupils center and its boundaries. Experiments showed that the method is feasible and it can be exploited even in non-ideal operative condition of iris recognition biometric systems.

An ontology-based approach to human telepresence

Abstract: Detecting human presence automatically is a challenging task since several environmental parameters may affect the quality and the continuity of detection. Although many techniques have been developed so far in the literature to solve this problem, they generally rely on well-defined operational context. Hence, they are sensitive to uncontrolled variables and unpredicted events. In this work an ontology-based approach to human telepresence detection is presented. Contrarily to classic sensor-driven techniques, a top-down methodology is applied. Starting from a formal description of the problem ontology, a set of high-response rate and low-response rate sensors is employed in a computational model. As a consequence of this model, a multi-sensor equipped device has been experimentally setup to conduct measurements on real scenarios. Experiments have been devised to estimate the robustness of the detection. In particular, some preliminary evaluations related to using a minimal set of chemical sensors are reported.

Data fusion for user presence identification

Abstract: Aim of this work is to present a new approach to the problem of user presence monitoring in working environments. Particularly, this work is focused on the evaluation of the presence or absence of a user in front of a terminal. This question is of paramount importance in applications requiring the user's presence e.g. video surveillance systems, control centrals, etc. The authors propose a technique of data fusion using signals from various low cost sensors.

A low-cost neural-based approach for wood types classification

Abstract: In many applications such as the furniture and the wood panel production, the classification of wood kinds can provide relevant information concerning the aspect, the properties and the preparation procedures of the products. Usually, the wood kind classification is made by trained operators, but this solution suffers of important drawbacks: it is time consuming and it has low repeatability/accuracy since the classification is related to the operator experience and fatigue. In the literature, some attempts to solve this applicative problem by automatic systems are present, but, unfortunately, these solutions present complex measures and setups. In this paper, we present a novel approach for wood kinds classification based on a neural network system which exploits the emitted spectrum of the wood samples filtered with a bank of low-cost optical filters coupled with a set of photo detectors. The structure of the proposed system can be directly implemented in an embedded low-cost system. The results of the system simulations are very satisfactory and they demonstrate that this approach is feasible and very promising.