Gil Santos | Universidade da Beira Interior (original) (raw)

Papers by Gil Santos

Every human being is entitled, by his very nature, to a set of physiological and behavioral featu... more Every human being is entitled, by his very nature, to a set of physiological and behavioral features that characterize him. The study of such features led to the development of a considerable amount of systems and applications, referred as biometric systems. The use of biometric systems has been significantly growing over the last years, particularly in the field of security: authentication, access control, criminal identification, etc. Being a highly demanding sector, it is then natural that greater focus is placed on the biometric traits that are able to deliver high discrimination between subjects whilst being less prone to forgery. However, such constraints represent a significant impact on both system’s usability and flexibility, requiring from the user a significant amount of cooperation. In this context, the iris is a primordial trait. Existing biometric recognition systems based on the iris follow the pioneer approach proposed by John Daugman, that proved itself as an excell...

Journal of Signal Processing Systems, 2015

In this paper, we propose a re-weighted elastic net (REN) model for biometric recognition. The ne... more In this paper, we propose a re-weighted elastic net (REN) model for biometric recognition. The new model is applied to data separated into geometric and color spatial components. The geometric information is extracted using a fast cartoon-texture decomposition model based on a dual formulation of the total variation norm allowing us to carry information about the overall geometry of images. Color components are defined using linear and nonlinear color spaces, namely the redgreen-blue (RGB), chromaticity-brightness (CB) and hue-saturation-value (HSV). Next, according to a Bayesian fusion-scheme, sparse representations for classification purposes are obtained. The scheme is numerically solved using a gradient projection (GP) algorithm. In the empirical validation of the proposed model, we have chosen the periocular region, which is an emerging trait known for its robustness against low quality data. Our results were obtained in the publicly available UBIRIS.v2 data set and show consistent improvements in recognition effectiveness when compared to related state-of-the-art techniques.

The dramatic growth in practical applications for iris biometrics has been accompanied by relevan... more The dramatic growth in practical applications for iris biometrics has been accompanied by relevant developments in the underlying algorithms and techniques. Along with the research focused on near-infrared images captured with subject cooperation, efforts are being made to minimize the trade-off between the quality of the captured data and the recognition accuracy on less constrained environments, where images are obtained at the visible wavelength, at increased distances, over simplified acquisition protocols and adverse lightning conditions. At a first stage, interpolation effects on normalization process are addressed, pointing the outcomes in the overall recognition error rates. Secondly, a couple of post-processing steps to the Daugman's approach are performed, attempting to increase its performance in the particular unconstrained environments this thesis assumes. Analysis on both frequency and spatial domains and finally pattern recognition methods are applied in such efforts. This thesis embodies the study on how subject recognition can be achieved, without his cooperation, making use of iris data captured at-a-distance, on-the-move and at visible wavelength conditions. Widely used methods designed for constrained scenarios are analyzed.

Proceedings - 2010 IEEE International Symposium on Multimedia, ISM 2010, 2010

The human iris supports contactless data acquisition and can be imaged covertly. These factors gi... more The human iris supports contactless data acquisition and can be imaged covertly. These factors give raise to the possibility of performing biometric recognition procedure without subjects' knowledge and in uncontrolled data acquisition scenarios. The feasibility of this type of recognition has been receiving increasing attention, as is of particular interest in visual surveillance, computer forensics, threat assessment, and other security areas. In this paper we stress the role played by the spectrum of the visible light used in the acquisition process and assess the discriminating iris patterns that are likely to be acquired according to three factors: type of illuminant, it's luminance, and levels of iris pigmentation. Our goal is to perceive and quantify the conditions that appear to enable the biometric recognition process with enough confidence.

IET Biometrics, 2015

Substantial efforts have been put into bridging the gap between biometrics and visual surveillanc... more Substantial efforts have been put into bridging the gap between biometrics and visual surveillance, in order to develop automata able to recognise human beings 'in the wild'. This study focuses on biometric recognition in extremely degraded data, and its main contributions are threefold: (1) announce the availability of an annotated dataset that contains high quality mugshots of 101 subjects, and large sets of probes degraded extremely by 10 different noise factors; (2) report the results of a mimicked watchlist identification scheme: an online survey was conducted, where participants were asked to perform positive and negative identification of probes against the enrolled identities. Along with their answers, volunteers had to provide the major reasons that sustained their responses, which enabled the authors to perceive the kind of features that are most frequently associated with successful/failed human identification processes. As main conclusions, the authors observed that humans rely greatly on shape information and holistic features. Otherwise, colour and texture-based features are almost disregarded by humans; (3) finally, the authors give evidence that the positive human identification on such extremely degraded data might be unreliable, whereas negative identification might constitute an interesting alternative for such cases.

2010 3rd International Congress on Image and Signal Processing, 2010

The growth in practical applications for iris biometrics has been accompanied by relevant develop... more The growth in practical applications for iris biometrics has been accompanied by relevant developments in the underlying algorithms and techniques. Efforts are being made to minimize the tradeoff between the recognition error rates and data quality, acquired in the visible wavelength, in less controlled environments, over simplified acquisition protocols and varying lighting conditions. This paper presents an approach that can be regarded as an extension to the widely known Daugman's method. Its basis is the analysis of the distribution of the concordant bits when matching iriscodes on both the spatial and frequency domains. Our experiments show that this method is able to improve the recognition performance over images captured in less constrained acquisition setups and protocols. Such conclusion was drawn upon trials conducted for multiple datasets.

2011 International Joint Conference on Biometrics (IJCB), 2011

Corner detection has been motivating several research works and is particularly important in diff... more Corner detection has been motivating several research works and is particularly important in different computer vision tasks, acting as basis for further image understanding stages. Particularly, the detection of eye-corners in facial images is relevant for domains such as biometric systems and assisted-driving systems. Having empirically evaluated the state-of-the-art of eye-corner detection proposals, we observed that they only achieve satisfactory results when dealing with good quality data. Hence, in this paper we describe an eyecorner detection method with particular focus on robustness, i.e., the suitability to deal with degraded data, toward the applicability in real-world conditions. Our experiments show that the proposed method outperforms others either in noisefree and degraded data (blurred, rotated and with significant variations in scale), which is regarded as the main achievement.

Computer Vision and Image Understanding, 2012

Despite the substantial research into the development of covert iris recognition technologies, no... more Despite the substantial research into the development of covert iris recognition technologies, no machine to date has been able to reliably perform recognition of human beings in real-world data. This limitation is especially evident in the application of such technology to large-scale identification scenarios, which demand extremely low error rates to avoid frequent false alarms. Most previously published works have used intensity data and performed multi-scale analysis to achieve recognition, obtaining encouraging performance values that are nevertheless far from desirable. This paper presents two key innovations. (1) A recognition scheme is proposed based on techniques that are substantially different from those traditionally used, starting with the dynamic partition of the noise-free iris into disjoint regions from which MPEG-7 color and shape descriptors are extracted. (2) The minimal levels of linear correlation between the outputs produced by the proposed strategy and other state-of-the-art techniques suggest that the fusion of both recognition techniques significantly improve performance, which is regarded as a positive step towards the development of extremely ambitious types of biometric recognition.

2013 IEEE Symposium on Computational Intelligence in Biometrics and Identity Management (CIBIM), 2013

The periocular region has recently emerged as a promising trait for unconstrained biometric recog... more The periocular region has recently emerged as a promising trait for unconstrained biometric recognition, specially on cases where neither the iris and a full facial picture can be obtained. Previous studies concluded that the regions in the vicinity of the human eye-the periocular region-have surprisingly high discriminating ability between individuals, are relatively permanent and easily acquired at large distances. Hence, growing attention has been paid to periocular recognition methods, on the performance levels they are able to achieve, and on the correlation of the responses given by other. This work overviews the most relevant research works in the scope of periocular recognition: summarizes the developed methods, and enumerates the current issues, providing a comparative overview. For contextualization, a brief overview of the biometric field is also given.

Advances in Visual Computing, 2015

In recent years, periocular recognition has become a popular alternative to face and iris recogni... more In recent years, periocular recognition has become a popular alternative to face and iris recognition in less ideal acquisition scenarios. An interesting example of such scenarios is the usage of mobile devices for recognition purposes. With the growing popularity and easy access to such devices, the development of robust biometric recognition algorithms to work under such conditions finds strong motivation. In the present work we assess the performance of extended versions of two state-ofthe-art periocular recognition algorithms on the publicly available CSIP database, a recent dataset composed of images acquired under highly unconstrained and multi-sensor mobile scenarios. The achieved results show each algorithm is better fit to tackle different scenarios and applications of the biometric recognition problem.

IEEE International Joint Conference on Biometrics, 2014

Using the periocular region for biometric recognition is an interesting possibility: this area of... more Using the periocular region for biometric recognition is an interesting possibility: this area of the human body is highly discriminative among subjects and relatively stable in appearance. In this paper, the main idea is that improved solutions for defining the periocular region-of-interest and better pose / gaze estimates can be obtained by segmenting (labelling) all the components in the periocular vicinity. Accordingly, we describe an integrated algorithm for labelling the periocular region, that uses a unique model to discriminate between seven components in a single-shot: iris, sclera, eyelashes, eyebrows, hair, skin and glasses. Our solution fuses texture / shape descriptors and geometrical constraints to feed a two-layered graphical model (Markov Random Field), which energy minimization provides a robust solution against uncontrolled lighting conditions and variations in subjects pose and gaze.

2013 IEEE Symposium on Computational Intelligence in Biometrics and Identity Management (CIBIM), 2013

Facial expressions result from movements of muscular action units, in response to internal emotio... more Facial expressions result from movements of muscular action units, in response to internal emotion states or perceptions, and it has been shown that they decrease the performance of face-based biometric recognition techniques. This paper focuses in the recognition of facial expressions and has the following purposes: 1) confirm the suitability of using dense image descriptors widely known in biometrics research (e.g., local binary patterns and histograms of oriented gradients) to recognize facial expressions; 2) compare the effectiveness attained when using different regions of the face to recognize expressions; 3) compare the effectiveness attained when the identity of subjects is known / unknown, before attempting to recognize their facial expressions.

Face Recognition in Adverse Conditions

There are several scenarios where a full facial picture cannot be obtained nor the iris properly ... more There are several scenarios where a full facial picture cannot be obtained nor the iris properly imaged. For such cases, a good possibility might be to use the ocular region for recognition, which is a relatively new idea and is regarded as a good trade-off between using the whole face or the iris alone. The area in the vicinity of the eyes is designated as periocular and is particularly useful on less constrained conditions, when image acquisition is unreliable, or to avoid iris pattern spoofing. This chapter provides a comprehensive summary of the most relevant research conducted in the scope of ocular (periocular) recognition methods. The authors compare the main features of the publicly available data sets and summarize the techniques most frequently used in the recognition algorithms in this chapter. In addition, they present the state-of-the-art results in terms of recognition accuracy and discuss the current issues on this topic, together with some directions for further work.

Pattern Analysis and Applications, 2015

Using information near the human eye to perform biometric recognition has been gaining popularity... more Using information near the human eye to perform biometric recognition has been gaining popularity. Previous works in this area, designated periocular recognition, show remarkably low error rates and particularly high robustness when data are acquired under less controlled conditions. In this field, one factor that remains to be studied is the effect of facial expressions on recognition performance, as expressions change the textural/shape information inside the periocular region. We have collected a multisession dataset whose single variation is the subjects' facial expressions and analyzed the corresponding variations in performance, using the state-of-the-art periocular recognition strategy. The effectiveness attained by different strategies to handle the effects of facial expressions was compared: (1) single-sample enrollment; (2) multisample enrollment, and (3) multisample enrollment with facial expression recognition, with results also validated in the well-known Cohn-Kanade AU-Coded Expression dataset. Finally, the role of each type of facial expression in the biometrics menagerie effect is discussed.

Lecture Notes in Computer Science, 2015

Efforts in biometrics are being held into extending robust recognition techniques to in the wild ... more Efforts in biometrics are being held into extending robust recognition techniques to in the wild scenarios. Nonetheless, and despite being a very attractive goal, human identification in the surveillance context remains an open problem. In this paper, we introduce a novel biometric system-Quis-Campi-that effectively bridges the gap between surveillance and biometric recognition while having a minimum amount of operational restrictions. We propose a fully automated surveillance system for human recognition purposes, attained by combining human detection and tracking, further enhanced by a PTZ camera that delivers data with enough quality to perform biometric recognition. Along with the system concept, implementation details for both hardware and software modules are provided, as well as preliminary results over a real scenario.

Pattern Recognition Letters, 2012

International Conference on Computational Intelligence and Security, 2009

Over the last years the usage of mobile devices has substantially grown, along with their capabil... more Over the last years the usage of mobile devices has substantially grown, along with their capabilities and applications. Extending biometric technologies to such gadgets is quite desirable, as it would rep- resent the ability to perform biometric recognition virtually anytime, anywhere, and by everyone. This paper focus on biometric recognition on mobile environments using the iris and periocular information as main traits, and its main contributions are three-fold: 1) announce the availability of an iris and periocular dataset containing images acquired with 10 different mobile setups, along with the corresponding iris segmentation data. Such dataset allows to evaluate both iris segmentation and recognition methods, as well as periocular recognition techniques; 2) report the outcomes of device-specific calibration techniques that compensate for the different color perception inherent to each setup; 3) propose the application of well-known iris and periocular recognition strategies, based on classical encoding and matching techniques, giving evidence on how they can be fused to overcome the issues associated with mobile environments.

Substantial efforts have been put into bridging the gap between biometrics and visual surveillanc... more Substantial efforts have been put into bridging the gap between biometrics and visual surveillance, in order to develop automata able to recognize human beings in the wild. This paper focuses on biometric recognition in extremely degraded data, and its main contributions are three-fold: 1) announce the availability of an annotated dataset that contains high quality mugshots of 101 subjects, and large sets of probes degraded extremely by 10 different noise factors; 2) report the results of a mimicked watchlist identification scheme: an online survey was conducted, where participants were asked to perform positive and negative identification of probes against the enrolled identities. Along with their answers, volunteers had to provide the major reasons that sustained their responses, which enabled us to perceive the kind of features that are most frequently associated with successful / failed human identification processes. As main conclusions, we observed that humans rely greatly on shape information and holistic features. Otherwise, colour and texture-based features are almost disregarded by humans; 3) finally, we give evidence that the positive human identification on such extremely degraded data might be unreliable, whereas negative identification might constitute an interesting alternative for such cases.

Journal of Signal Processing Systems, 2015

Proceedings - 2010 IEEE International Symposium on Multimedia, ISM 2010, 2010

IET Biometrics, 2015

2010 3rd International Congress on Image and Signal Processing, 2010

2011 International Joint Conference on Biometrics (IJCB), 2011

Computer Vision and Image Understanding, 2012

2013 IEEE Symposium on Computational Intelligence in Biometrics and Identity Management (CIBIM), 2013

Advances in Visual Computing, 2015

IEEE International Joint Conference on Biometrics, 2014

2013 IEEE Symposium on Computational Intelligence in Biometrics and Identity Management (CIBIM), 2013

Face Recognition in Adverse Conditions

Pattern Analysis and Applications, 2015

Lecture Notes in Computer Science, 2015

Pattern Recognition Letters, 2012

International Conference on Computational Intelligence and Security, 2009

Substantial efforts have been put into bridging the gap between biometrics and visual surveillanc... more Substantial efforts have been put into bridging the gap between biometrics and visual surveillance, in order to develop automata able to recognize human beings in the wild. This paper focuses on biometric recognition in extremely degraded data, and its main contributions are three-fold: 1) announce the availability of an annotated dataset that contains high quality mugshots of 101 subjects, and large sets of probes degraded extremely by 10 different noise factors; 2) report the results of a mimicked watchlist identification scheme: an online survey was conducted, where participants were asked to perform positive and negative identification of probes against the enrolled identities. Along with their answers, volunteers had to provide the major reasons that sustained their responses, which enabled us to perceive the kind of features that are most frequently associated with successful / failed human identification processes. As main conclusions, we observed that humans rely greatly on shape information and holistic features. Otherwise, colour and texture-based features are almost disregarded by humans; 3) finally, we give evidence that the positive human identification on such extremely degraded data might be unreliable, whereas negative identification might constitute an interesting alternative for such cases.