Search

Due to the proliferation of sports-related multimedia content on the WWW, effective visual search and retrieval present interesting research challenges. These are caused by poor image quality, a wide range of possible camera points of view, pose variations on the part of athletes engaged in playing a sport, deformations of text appearing on sports person’s clothing and uniforms in motion, occlusions caused by other objects, etc. To address these challenges, this paper presents a new method for detecting text on human bodies in sports images. Unlike most existing methods, which attempt to exploit locations of a player’s torso, face, and skin, we propose an end-to-end episodic learning approach that employs inductive learning criteria for detecting clothing regions in an image, which are, in turn, then used for text detection. Our method integrates a Residual Network (ResNet) and Pyramidal Pooling Module (PPM) for generating a spatial attention map. The Progressive Scalable Expansion Algorithm (PSE) is adapted for text detection from these regions. Experimental results on our own dataset as well as several benchmarks (like RBNR and MMM which contain images of runners in marathons, and Re-ID which is a person re-identification dataset) demonstrate that the proposed method outperforms existing methods in terms of precision and F1-score. We also present results for sports images chosen from natural scene text detection datasets such as CTW1500 and MS-COCO to show the proposed method is effective and reliable across a range of inputs.

The growth of biometrics‐based authentication in various services raises the need to protect biometric data at the storage level. Specifically, biometric templates need to be protected after features are extracted to avoid the leakage of biometric data and subsequent linkability issues. An approach based on discrete hashing is presented with the assistance of semantic labels to generate discriminative and privacy preserving protected templates in this work. The proposed approach can easily be adopted for a closed‐enrolment set in which enrolment images are known a priori whereas the challenge of learning templates for a single subject remains open. To extend this approach for individual subject, the concept of auxiliary pseudouser enrolment data is introduced, through which a protected template can be generated at the user level. Through the use of a moderately sized multimodal biometric database of 94 subjects, the effectiveness of the proposed approach is illustrated to achieve a robust and secure template protection with irreversibility, unlinkability and renewability. With the set of experiments, the performance of the template protection approach is established and benchmarked against the popular bloom‐filter technique. The proposed approach results in a high genuine match rate (≈100% at a false accept rate of 0.01%) and low equal error rate (EER ≈ 0%) and outperforms traditional approaches while satisfying other requirements of biometric template protection when the closed enrolment set is known. With auxiliary pseudousers, the performance of the proposed approach for user‐level protected template creation results in an EER of 2.5%, indicating very low performance degradation compared with the known enrolment dataset. Along with the set of experimental validation of the proposed approach, a security analysis of the proposed approach is presented to demonstrate the unlinkability of the biometric templates using a state‐of‐art unlinkability metric.

Detecting multiple license plate numbers in crowded street scenes is challenging and requires the attention of researchers. In contrast to existing methods that focus on images that are not crowded with vehicles, in this work we aim at situations that are common and complex, for example, in city environments where numerous vehicles of different types like cars, trucks, motorbike etc. may present in a single image. In such cases, one can expect large variations in license plates in terms of quality, backgrounds, and various forms of occlusion. To address these challenges, we explore Adaptive Progressive Scale Expansion based Graph Attention Network (APSEGAT). This approach extracts local information which represents the license plates irrespective of vehicle types and numbers because it works at the pixel level in a progressive way, and identifies the dominant information in the image. This may include other parts of vehicles, drivers and pedestrians, and various other background objects. To overcome this problem, we integrate concepts of graph attention networks with progressive scale expansion networks. For evaluating the proposed method, we use our own dataset, named as AMLPR, which contains images captured in different crowded street scenes in different time span, and the benchmark dataset namely, UFPR-ALPR, which provides images of a single vehicle, and another benchmark dataset called, UCSD, which contains images of cars with different orientations. Experimental results on these datasets show that the method outperforms existing methods and is effective in detecting license plate numbers in crowded street scenes.

Methods developed for normal 2D text detection do not work well for text that is rendered using decorative, 3D effects, etc. This paper proposes a new method for classification of 2D and 3D natural scene text images so that an appropriate recognition method can be chosen accordingly based on the classification results for better performance. The proposed method explores local gradient differences for obtaining candidate pixels, which represent a stroke. To study the spatial distribution of candidate pixels, we propose a measure, called COLD, which is denser for pixels toward the center of strokes and scattered for non-stroke pixels. This observation leads us to introduce mass features for extracting the regular spatial pattern of COLD, which indicates a 2D text image. The extracted features are fed into a Neural Network (NN) for classification. The proposed method is tested on (i) a new dataset introduced in this work (ii) a second dataset assembled from standard natural scene datasets (iii) Non-Text Image datasets which does not contain text, rather it contains objects. Experimental results of the proposed method on images with text and non-text show that the proposed method is independent of text. The proposed approach improves text detection and recognition performance significantly after classification.

Identifying Tattoo is an integral part of forensic investigation and crime identification. Tattoo text detection is challenging because of its freestyle handwriting over the skin region with a variety of decorations. This paper introduces Deformable Convolution and Inception based Neural Network (DCINN) for detecting tattoo text. Before tattoo text detection, the proposed approach detects skin regions in the tattoo images based on color models. This results in skin regions containing Tattoo text, which reduces the background complexity of the tattoo text detection problem. For detecting tattoo text in the skin regions, we explore a DCINN, which generates binary maps from the final feature maps using differential binarization technique. Finally, polygonal bounding boxes are generated from the binary map for any orientation of text. Experiments on our Tattoo-Text dataset and two standard datasets of natural scene text images, namely, Total-Text, CTW1500 show that the proposed method is effective in detecting Tattoo text as well as natural scene text in the images. Furthermore, the proposed method outperforms the existing text detection methods in several criteria.

A new trend of smart city development opens up many challenges. One such issue is that automatic vehicle driving and detection for toll fee payment in night or limited light environments. This paper presents a new work for enhancing license plates captured in limited or low light conditions such that license plate detection methods can be expanded to detect images at night. Due to the popularity of Convolutional Neural Network (CNN) in solving complex issues, we explore U-Net-CNN for enhancing contrast of license plate pixels. Since the difference between pixels that represent license plates and pixels that represent background is too due to low light effect, the special property of U-Net that extracts context and symmetric of license plate pixels to separate them from background pixels irrespective of content. This process results in image enhancement. To validate the enhancement results, we use text detection methods and based on text detection results we validate the proposed system. Experimental results on our newly constructed dataset which includes images captured in night/low light/limited light conditions and the benchmark dataset, namely, UCSD, which includes very poor quality and high quality images captured in day, show that the proposed method outperforms the existing methods. In addition, the results on text detection by different methods show that the proposed enhancement is effective and robust for license plate detection.

Ocular images have been used to supplement and complement the face-based biometrics. Ocular images are further investigated for identifying the gender of a person such that the soft label can be used to boost biometric performance of the system. Although there are number of works in visible spectrum and Near-Infrared spectrum for gender classification using ocular images, there are limited works in spectral imaging which explore ocular images for gender classification. Considering the advantages of spectral imaging, we explore the problem of gender identification using ocular images obtained using spectral imaging. To this end, we have employed a recent database of 104 unique ocular instances across 2 different sessions and 5 different attempts in each session with a spectral imaging camera capable of capturing 8 different images corresponding to different bands. Further, we present a new framework of using fused feature descriptors in kernalized space to fully leverage the number of spectral images for robust gender classification. With the set of experiments, we obtain an average classification accuracy of \(81\%\) with the proposed approach of using fused GIST features along with the weighted kernel representation of features in collaborative space.

Given the need to protect the biometric data increasing due to recent enforcement of GDPR, we address the problem of protected template creation for iris recognition in this work. To this end, we introduce a new Morton Filter Based Template Protection for iris codes. The novel approach relies on creating a multi-bucket based protected template derived from single iris codes. By exploiting the low-rank iris codes from state-of-art iris feature extraction schemes to derive non-noisy iris bits, we design the new framework for creating protected templates by employing Morton Filters. The proposed template protection scheme is further evaluated on constrained and unconstrained iris recognition systems by employing two publicly available databases (IITD Iris and CASIA v4 Distance Dataset). Unlike previous works, to the best of our knowledge, this is the first work to demonstrate the applicability of template protection scheme in an unconstrained setting. The proposed approach achieves a low Equal Error Rate (EER ≈ 0%) on both constrained and unconstrained databases, exemplifying the scalability and adaptability. The approach being antagonistic to various configurations (block size and bit lengths), achieves a high degree of unlinkability deeming it practical for iris recognition systems.

Multi-patch based and hoslistic image deep sparse features for improved iris recognition using collaborative subspace.Explores color channel along with patch based approach for better feature representation.Extensive analysis & results presented for both MICHE-I & MICHE-II databases.High verification accuracy (MICHE-I & II) with single sample iris enrolment. The challenge of recognizing iris in visible spectrum images captured using smartphone stems from heavily degraded data (due to reflection, partial closure of eyes, pupil dilation due to light) where the iris texture is either not visible or visible to very low extent. In order to perform reliable verification, the set of extracted features should be robust and unique to obtain high similarity scores between different samples of same subject while obtaining high dissimilarity score between samples of different subjects. In this work, we propose multi-patch deep features using deep sparse filters to obtain robust features for reliable iris recognition. Further, we also propose to represent them in a collaborative subspace to perform classification via maximized likelihood, even under single sample enrolment. Through the set of extensive experiments on MICHE-I iris dataset, we demonstrate the robustness of newly proposed scheme which achieves high verification rate (GMR > 95%) with low Equal Error Rate (EER < 2%). Further, the robustness of proposed feature representation is reiterated by employing simple distance measures which has outperformed the state-of-art techniques. Additionally, the scheme is tested on the MICHE-II challenge evaluation dataset where the results are promising with GMR=100% on limited sub-corpus of iPhone data.

The emergence of Full Field-Optical Coherence Tomography (FF-OCT) for fingerprint imaging has shown it's ability in addressing and solving the drawbacks of traditional fingerprinting solutions such as spoofing attacks, low accuracy for abraded fingerprint. With the availability of multiple internal fingerprints (from subsurface captured at different depths), it is also essential to consider the aspects of ideal biometrics where the privacy of the fingerprint data is preserved. In this work, we propose a new framework for fingerprint template protection, highly customized to FF-OCT by exploring the interplay between subsurface. As a first of it's kind work attempting template protection for FF-OCT fingerprints, we explore deeply learnt features to derive first level of template for subsurface fingerprint image. We further propose to intertwine subsurface level templates to provide better and robust templates. With the set of extensive experiments on a FF-OCT fingerprint database of 200 unique fingerprints with a total of 2400 images, we demonstrate reliable biometric performance resulting in EER of 5.69% for unprotected template at first layer (subsurface) of fingerprint in FF-OCT, an EER of 5.86% for the protected templates at same layer and EER of 5.08% with the final protected templates with proposed intertwining of subsurface fingerprint. Further, through the security analysis, we also validate the strength of the proposed approach with near ideal unlinkability.

Fingerprint-based biometric systems have shown reliability in terms of accuracy in both biometric and forensic scenarios. Although fingerprint systems are easy to use, they are susceptible to presentation attacks that can be carried out by employing lifted or latent fingerprints. This work presents a systematic study of the fingerprint presentation attack detection (PAD aka., spoofing detection) using textural features. To this end, this chapter reports an evaluation of both hand-crafted features and naturally learned features via deep learning techniques for fingerprint presentation attack detection. The evaluation is presented on publicly available fake fingerprint database that consists of both bona fide (i.e., real) and presentation attack fingerprint samples captured by capacitive, optical and thermal sensors. The results indicate the need for further approaches that can detect attacks across data from different sensors.

Smartphone-based biometrics authentication has been increasingly used for many popular everyday applications such as e-banking and secure access control to personal services. The use of biometric data on smartphones introduces the need for capturing and storage of biometric data such as face images. Unlike the traditional passwords used for many services, biometric data once compromised cannot be replaced. Therefore, the biometric data not only should not be stored as a raw image but also needs to be protected such that the original image cannot be reconstructed even if the biometric data is available. The transforming of raw biometric data such as face image should not decrease the comparison performance limiting the use of biometric services. It can therefore be deduced that the feature representation and the template protection scheme should be robust to have reliable smartphone biometrics. This chapter presents two variants of a new approach of template protection by enforcing the structure preserving feature representation via manifolds, followed by the hashing on the manifold feature representation. The first variant is based on the Stochastic Neighbourhood Embedding and the second variant is based on the Laplacian Eigenmap. The cancelability feature for template protection using the proposed approach is induced through inherent hashing approach relying on manifold structure. We demonstrate the applicability of the proposed approach for smartphone biometrics using a moderately sized face biometric data set with 94 subjects captured in 15 different and independent sessions in a closed-set scenario. The presented approach indicates the applicability with a low Equal Error Rate, \(EER = 0.65\%\) and a Genuine Match Rate, \(GMR = 92.10\%\) at False Match Rate (FMR) of \(0.01\%\) for the first variant and the second variant provides \(EER = 0.82\%\) and \(GMR = 89.45\%\) at FMR of \(0.01\%\). We compare the presented approach against the unprotected template performance and the popularly used Bloom filter template.

The need to protect biometric data within iris systems has resulted in a number of template protection schemes. A primary issue with current template protection schemes for iris recognition is the unavoidable biometric error rates, i.e., for any given False Non-Match Rate (FNMR) there is a high False Match Rate (FMR), especially at lower values of FNMR. In this work, we primarily focus on addressing this problem using a new approach with Cuckoo Filtering simultaneously using both stable bits and discriminative bits to derive a stronger template protection scheme. The proposed template protection scheme performs in a robust manner for various configurations as compared to earlier template protection schemes that need empirical fine-tuning. With the set of experiments on a publicly available iris dataset, we benchmark our results against the state-of-art template protection scheme based on Bloom-Filters. Specifically, we demonstrate the gain in performance and robustness of proposed approach at lower FNMR and invariance of performance to configurations of template protection scheme. With a specific configuration of proposed approach, we achieve Genuine Match Rate (GMR) = 100% at FMR = 0:01% and EER = 0% in the best case and GMR = 98:44% at FMR = 0:01% and EER = 0:33% in the worst case on IITD Iris database.

Periocular characteristics is gaining prominence in biometric systems and surveillance systems that operate either in NIR spectrum or visible spectrum. While the ocular information can be well utilized, there exists a challenge to compare images from different spectra such as Near-Infra-Red (NIR) versus Visible spectrum (VIS). In addition, the ocular biometric templates from both NIR and VIS domain need to be protected after the extraction of features to avoid the leakage or linkability of biometric data. In this work, we explore a new approach based on anchored kernel hashing to obtain a cancelable biometric template that is both discriminative for recognition purposes while preserving privacy. The key benefit is that the proposed approach not only works for both NIR and the Visible spectrum, it can also be used with good accuracy for cross-spectral protected template comparison. Through the set of experiments using a cross-spectral periocular database, we demonstrate the performance with \(EER=1.39\%\) and \(EER=1.61\%\) for NIR and VIS protected templates respectively. We further present a set of cross-spectral template comparison by comparing the protected templates from one spectrum to another spectra to demonstrate the applicability of the proposed approach.

Despite the challenge in recognizing disguise images on Face Recognition Systems (FRS), it has not received significant attention. Considering the fact that most of the FRS operates in visible spectrum which is not optimal solution for disguise face recognition, we explore spectral imaging and thereby the intrinsic features of spectral imaging for disguise face recognition. The major contribution of this work lies in presenting new approach that extract the histogram features independently for the spectral band images using Histogram Oriented Gradient (HOG) method and learning the unique discriminative features in affine feature space for disguise face recognition problem in robust manner. We present an extensive experimental validation results in terms of verification rate to present the significance of our proposed method. Further, we employ state-of-the-art disguise face recognition methods on individual spectral bands and fused spectral bands. The Proposed method demonstrates best performance in recognizing disguise face images compared to the state-of-the-art methods.

Fingervein based biometric verification systems are widely used in various access control applications due to their accuracy and reliability. However, almost all existing fingervein systems scan only one finger at a time and this may not be a reliable solution for more secured applications demanding higher accuracy in verification and reliability. In this work, we propose a new multi-fingervein capture system that can simultaneously capture three different fingers in a single capture instance. The developed fingervein sensor is validated for operational purposes through images captured from 20 unique identities(subjects) with 600 fingervein images. The suitability of captured images for biometric applications is vetted through the use of four different state-of-the-art verification algorithms. The results are benchmarked and compared using individual fingervein data and in addition the employability of score level fusion is presented to improve the performance. The reported results in this article support the anticipated increase in performance of the multi-fingervein verification.

A light field sensor can provide useful information in terms of multiple depth (or focus) images, holding additional information that is quite useful for biometric applications. In this paper, we examine the applicability of a light field camera for biometric applications by considering two prominently used biometric characteristics: 1) face and 2) iris. To this extent, we employed a Lytro light field camera to construct two new and relatively large scale databases, for both face and iris biometrics. We then explore the additional information available from different depth images, which are rendered by light field camera, in two different manners: 1) by selecting the best focus image from the set of depth images and 2) combining all the depth images using super-resolution schemes to exploit the supplementary information available within the set elements. Extensive evaluations are carried out on our newly constructed database, demonstrating the significance of using additional information rendered by a light field camera to improve the overall performance of the biometric system.

The gaining popularity of the visible spectrum iris recognition has sparked the interest in adopting it for various access control applications. Along with the popularity of visible spectrum iris recognition comes the threat of identity spoofing, presentation, or direct attack. This paper presents a novel scheme for detecting video presentation attacks in visible spectrum iris recognition system by magnifying the phase information in the eye region of the subject. The proposed scheme employs modified Eulerian video magnification (EVM) to enhance the subtle phase information in eye region and novel decision module to classify it as artefact(spoof attack) or normal presentation. The proposed decision module is based on estimating the change of phase information obtained from EVM, specially tailored to detect presentation attacks on video-based iris recognition systems in visible spectrum. The proposed scheme is extensively evaluated on the newly constructed database consisting of 62 unique iris video acquired using two smartphones—iPhone 5S and Nokia Lumia 1020. We also construct the artefact database with 62 iris acquired by replaying normal presentation iris video on iPad with retina display. Extensive evaluation of proposed presentation attack detection (PAD) scheme on the newly constructed database has shown an outstanding performance of average classification error rate = 0% supporting the robustness of the proposed PAD scheme.

Abstarct In this paper we propose a novel approach for face anti-spoofing called extended local ternary co-relation pattern (ELTCP). The ELTCP encodes co-relation of ternary edges based on the centre pixel gray values with its immediate directional neighbour and its next immediate directional neighbour. The proposed method is robust against presentation attacks by extracting the spatial information in all the directions. Three Experiments were performed by using all the four texture descriptors (LBP, LTP, LGS and ELTCP) and the results are compared. The proposed face anti-spoofing method performs better than LBP, LTP and LGS.

Presentation attacks against Face Recognition System (FRS) have incrementally posed challenges to create new detection methods. Among the various presentation attacks, disguise attacks allow concealing the identity of the attacker thereby increasing the vulnerability of the FRS. In this paper, we present a new approach for attack detection in multi-spectral systems, where face disguise attacks are carried out. The approach is based on using spectral signatures obtained from a spectral camera operating in eight narrow spectral bands across the Visible (VIS) and Near Infra-Red (NIR) (530nm to 1000nm) spectrum and learning deeply coupled auto-encoders. The robustness of the proposed approach is validated using a newly collected spectral face database of subjects conducting both bona fide (i.e. real) presentations and disguise attack presentations. The database is designed to capture 2 different kinds of attacks from 54 subjects, amounting to a total number of 6480 samples. Extensive experiments carried on the multi-spectral face database indicate the robust performance of proposed scheme when benchmarked with three different state-of-the-art methods.

Face recognition systems are widely known for their vulnerability against presentation attacks or spoofing attacks. The exponential deployment of face recognition systems has been further challenged even by the simple and low-cost face artefacts generated using conventional printers. In this paper, we present a novel scheme to detect face presentation attacks posed by high-quality print attacks which are relatively difficult to detect. The proposed scheme leverages the phase information extracted from the spatial-frequency representation of the given image. We also present a new face presentation attack database collected using the iPhone 6S. The new database is comprised of 100 subjects collected in two different sessions that have resulted in a total of 31228 samples (or images). Extensive experiments are carried out on the newly constructed database and the obtained results show the improved performance of the proposed scheme when compared aaainst six different state-of-the-art methods.

Multi-spectral imaging has been explored to handle a set of deficiencies found in traditional imaging that capture the images only in visible spectrum (VIS) or Near-Infra Red (NIR) spectrum. The promising performance obtained in the experimental works indicates the use-case in real-life biometric systems. As biometric systems should also consider protecting biometric templates, it is required to have an efficient template protection scheme for multi-spectral biometric systems to avoid the leakage of biometric data and subsequent linkability issues. In this work, we propose a new template protection scheme for multi-spectral biometric systems through the use of biometric information across different spectra to provide protected templates. Through the proposed approach of kernalized hashing, we can reach a fully unlinkable template protection scheme that works across all spectra in a multi-spectral system with a comparable performance to unprotected system. Further, we propose a template level fusion across all the spectral bands to improve the performance of the multi-spectral biometric system with integrated template protection. Through the use of a relatively large sized multispectral face biometric database of 168 subjects captured in 9 narrow spectral bands in VIS and NIR bands (530nm to l000nm), we illustrate the effectiveness of the proposed approach in achieving a robust and secure template protection while accounting for irreversibility, unlinkability and renewability. Through the experiments we establish the performance of the proposed template protection approach and demonstrate a high Genuine Match Rate $(GMR\approx 100\%$ at False Accept Rate of $FMR=0.01\%$ and low Equal Error Rate $EER=\approx 0\%$ , while satisfying other requirements of biometric template protection. Further, we present a security analysis of the proposed approach to demonstrate the unlinkability of the biometric templates.

The increasing use of smartphone biometrics proportionally increases the need to protect the biomemc data. Unlike the traditional passwords, biometric data once comprised cannot be replaced. The smartphones thus need to protect the biometric templates and store the protected templates instead of the raw biomemc images such as face image. The challenge in outaining optimal biometric templates without compromising the performance also relies on preserving the structural neighbourhood of the biometric characteristics. We therefore address these two problems by employing cancelaule biometric templates via hashing. The approach further is based on neighbourhood structure preserving manifold which we refer as Manifold-structure Preserving Biometric Template (MaP-BiT). The proposed approach also demonstrates the properties of an ideal biometric template protection scheme and obeys irreversibility, unlinkability and renewability. In order to validate the applicauility of the proposed approach, we employ a face and periocular database captured using the smart-phone that consists of 94 subjects in 15 different and independent sessions. The proposed approach results in a low Equal Error Rate $(\approx 0.80$%) with a Genuine Match Rate ($\approx 97.5$% at False Match Rate of 0.01%). The results indicate the promising nature of the proposed approach as compared to the traditional unprotected templates while outperforming the bloom-filter based template protection.

Face recognition has become a ubiquitous way of establishing identity in many applications. Gender transformation therapy induces changes to face on both for structural and textural features. A challenge for face recognition system is, therefore, to reliably identify the subjects after they undergo gender change while the enrolment images correspond to pre-change. In this work, we propose a new framework based on augmenting and fine-tuning deep Residual Network-50 (ResNet-50). We employ YouTube database with 37 subjects whose images are self-captured to evaluate the performance of state-of-the-schemes. Obtained results demonstrate the superiority of the proposed scheme over twelve different state-of-the-art schemes with an improved Rank — 1 recognition rate.

Gender classification based on the facial characteristic, has been widely studied in the literature across visible and near infrared spectrum. In this paper, we explore the applicability of extended multi-spectral imaging for the gender classification by quantifying the photometric property of the captured image. We proposed a novel scheme based on the Spectral Angle Mapper (SAM) that can effectively capture the spectral information across the multi-spectral bands that is further classified using the linear Support Vector Machine (SVM). Extensive set of experiments are carried out using a newly constructed multi-spectral face database with 78300 samples stemming from 145 subjects in six different scenarios. The obtained results show the best average classification accuracy of 93.51%, signifying the applicability of the proposed approach on the extended multi-spectral face data for robust gender classification.

Spectral face recognition is gaining importance as the information from different bands can lead to robust face representation that are presentation (a.k.a, spoofing) attack resistant. However, the key challenge here is to process the high dimensional spatio-spectral data to extract and represent the reliable data while discarding redundant data information for processing. In this work, we present a new approach to represent the spectral images in a holistic manner by employing the well-known Projection Metric Learning (PML) in Grassmann manifold such that the redundant information from the spectral data is discarded while retaining significant information. The approach is adopted to learn discriminative information from an high dimensional spectral dataset. Further, we propose an extension using collaborative representation of learnt projection metrics for improving the classification accuracy of spectral data. With the extensive set of experiments conducted on a relatively large scale extended-spectral face image database (6048 images) of 168 subjects, we demonstrate the applicability of the proposed framework. The obtained results indicates highest accuracy by achieving a Rank-1 recognition rate of 98.21% in classifying the spectral face images.

Gender prediction based on facial features has received significant attention in computer vision and biometric community. Most of the gender classification studies mainly focused there attention on approaches that operate in the visible spectrum. In this paper we present gender classification using extended multi-spectral face data captured in nine narrow spectral bands across the visible near infrared spectrum (530nm to 1000nm). Further, we present the proposed method, that learns for this image set the discriminative spectral band features in the affine space and then classifies the features with a Support Vector Machine (SVM) in a robust manner. The extensive experimental results are presented on the reasonable sample size of 78300 spectral band images using our proposed method. The obtained results shows 90.49±3.56% average classification accuracy, indicating the applicability of our proposed method for gender classification.

This work presents the 2nd Cross-Spectrum Iris/Periocular Recognition Competition (Cross-Eyed2017). The main goal of the competition is to promote and evaluate advances in cross-spectrum iris and periocular recognition. This second edition registered an increase in the participation numbers ranging from academia to industry: five teams submitted twelve methods for the periocular task and five for the iris task. The benchmark dataset is an enlarged version of the dual-spectrum database containing both iris and periocular images synchronously captured from a distance and within a realistic indoor environment. The evaluation was performed on an undisclosed test-set. Methodology, tested algorithms, and obtained results are reported in this paper identifying the remaining challenges in path forward. © 2017 IEEE SIDUS-AIR FastPass PROTECT HFACE SWAN

Good biometric performance of iris recognition motivates it to be used for many large scale security and access control applications. Recent works have identified visible spectrum iris recognition as a viable option with considerable performance. Key advantages of visible spectrum iris recognition include the possibility of iris imaging in on-the-move and at-a-distance scenarios as compared to fixed range imaging in near-infra-red light. The unconstrained iris imaging captures the images with largely varying radius of iris and pupil. In this work, we propose a new segmentation scheme and adapt it to smartphone based visible iris images for approximating the radius of the iris to achieve robust segmentation. The proposed technique has shown the improved segmentation accuracy up to 85% with standard OSIRIS v4.1. This work also proposes a new feature extraction method based on deepsparsefiltering to obtain robust features for unconstrained iris images. To evaluate the proposed segmentation scheme and feature extraction scheme, we employ a publicly available database and also compose a new iris image database. The newly composed iris image database (VSSIRIS) is acquired using two different smartphones - iPhone 5S and Nokia Lumia 1020 under mixed illumination with unconstrained conditions in visible spectrum. The biometric performance is benchmarked based on the equal error rate (EER) obtained from various state-of-art schemes and proposed feature extraction scheme. An impressive EER of 1.62% is obtained on our VSSIRIS database and an average gain of around 2% in EER is obtained on the public database as compared to the well-known state-of-art schemes.

Vulnerability of iris recognition systems remains a challenge due to diverse presentation attacks that fail to assure the reliability when adopting these systems in real-life scenarios. In this paper, we present an in-depth analysis of presentation attacks on iris recognition systems especially focusing on the photo print attacks and the electronic display (or screen) attack. To this extent, we introduce a new relatively large scale visible spectrum iris artefact database comprised of 3300 iris normal and artefact samples that are captured by simulating five different attacks on iris recognition system. We also propose a novel presentation attack detection (PAD) scheme based on multiscale binarized statistical image features and linear support vector machines. Extensive experiments are carried out on four different publicly available iris artefact databases that have revealed the outstanding performance of the proposed PAD scheme when benchmarked with various well-established state-of-the-art schemes.

In the past, many efforts have been directed to develop biometrie template protection schemes to guard bio-metric reference data, i.e. templates. One fundamental premise in the design of such schemes is that the average entropy of the templates should be maximised in order to improve the level of protection. In parallel, several works have addressed the difficult problem of measuring the average entropy of biometric characteristics. However, the impact of the correlation present in different regions of a single biometric characteristic (e.g., left and right part of the face) or within two instances of a single subject (e.g., left and right palmprints) on the joint entropy of a multi-biometric template has been overlooked os far. In this paper, we address this issue and propose a way to measure such correlation from an information theoretical perspective. We then apply the proposed measure to a particular case study based on periocular biometrics, using the MobBIO database. The results show that up to 70% of the information comprised in both periocular regions of a given subject is correlated. Finally, we analyse the implications of such average mutual information loss on biometric template protection schemes.

Face biometrics is widely used in various applications including border control and facilitating the verification of travellers' identity claim with respect to his electronic passport (ePass). As in most countries, passports are issued to a citizen based on the submitted photo which allows the applicant to provide a morphed face photo to conceal his identity during the application process. In this work, we propose a novel approach leveraging the transferable features from a pre-trained Deep Convolutional Neural Networks (D-CNN) to detect both digital and print-scanned morphed face image. Thus, the proposed approach is based on the feature level fusion of the first fully connected layers of two D-CNN (VGG19 and AlexNet) that are specifically fine-tuned using the morphed face image database. The proposed method is extensively evaluated on the newly constructed database with both digital and print-scanned morphed face images corresponding to bona fide and morphed data reflecting a real-life scenario. The obtained results consistently demonstrate improved detection performance of the proposed scheme over previously proposed methods on both the digital and the print-scanned morphed face image database.

Fingerprint recognition has been extensively used in numerous civilian applications ranging from border control to everyday identity verification. The threats to current systems emerge from two facts that can be attributed to potential loss in accuracy due to damaged external fingerprints and attacks on the sensors by creation of an artefacts (e.g. silicone finger) simply by lifting the latent fingerprints. In the growing need for attack resistant biometric fingerprint recognition that can be operated without supervision, a new generation of sensors has been investigated, which can capture the subsurface fingerprint pattern. In this work, we explore a subsurface fingerprint imaging technique by employing a custom-built in-house Full-Field Optical Coherent Tomography (FF-OCT) sensor for capturing the subsurface fingerprint. Further, we evaluate a newly constructed database of 200 unique fingerprint samples collected in 2 different sessions with 6 layers of fingerprint images corresponding to 6 subsurface fingerprints. We also propose a framework based on quality metrics to fuse the subsurface fingerprint images to achieve a robust verification accuracy, which has resulted in Equal Error Rate (EER) of 0%. We also provide an extensive set of experiments to gauge the reliability of subsurface fingerprint recognition and deduce a set of important conclusions for the path forward in FFOCT subsurface fingerprint imaging.

Multispectral face recognition systems are widely used in various access control applications. The vulnerability of multispectral face recognition sensors towards low-cost Presentation Attack Instrument (PAI) such as printed photos used in attacks has emerged as a serious security threat. In this paper, we present a novel framework to detect presentation attacks against an extended multispectral face sensor. The proposed framework stems from the idea of exploring the complementary information available from different bands of an extended multispectral face sensor. To this extent, two different frameworks are proposed where the first framework is based on image fusion and the second builds on the Presentation Attack Detection (PAD) score level fusion. Extensive experiments are carried out on the extended multispectral face sensor database comprising of 50 subjects with two different presentation attacks generated using the printed photo artefacts. The obtained results indicate the superior performance of the PAD score level fusion on detecting both known and unknown attacks.

With the availability of sensor technology across the broad electromagnetic spectrum, multi-spectral imaging is increasingly used in biometric systems. Especially for face recognition, multi-spectral imaging has gained a lot of attention due to it's invariant property against variation caused by unknown illumination. However, obtaining best performance using multi-spectral imaging is still a challenge due to presence of a modality gap between the spectral imaging data and redundant band information. In this paper, we propose a fused band representation with a set of selected bands represented in Quaternion space for spectral band images to efficiently maintain the inter band relationship in spatial domain. The selection is based on measuring the information content in bands using entropy and fusion is carried out in Quaternion space for three best bands. The features from newly obtained image is collaboratively represented to achieve robust performance. The proposed approach is experimentally validated on the extended multi-spectral face database of 168 subjects, whose spectral band images are captured in 9 narrow spectral bands in visible and near infrared range (530nm to 1000nm). The quantitative performance analysis, obtained using the proposed method indicates 96.13% recognition rate at Rank-1, outperforming other state-of-the-art methods.

Presentation attack on the face recognition systems is well studied in the biometrics community resulting in various techniques for detecting the attacks. A low-cost presentation attack (e.g. print attacks) on face recognition systems has been demonstrated for systems operating in visible, multispectral (visible and near infrared spectrum) and extended multispectral (more than two spectral bands spanning from visible to near infrared space, commonly in 500nm-1000nm). In this paper, we propose a novel method to detect the presentation attacks on the extended multispectral face recognition systems. The proposed method is based on characterising the reflectance properties of the captured image through the spectral signature. The spectral signature is further classified using the linear Support Vector Machine (SVM) to obtain the decision on presented sample as an artefact or bona-fide. Since the reflectance property of the human skin and the artefact material differ, the proposed method can efficiently detect the presentation attacks on the extended multispectral system. Extensive experiments are carried out on a publicly available extended multispectral database (EMSPAD) comprised of 50 subjects with two different Presentation Attack Instrument (PAI) generated using two different printers. The comparison analysis is presented by comparing the performance of the proposed scheme with the contemporary schemes based on the image fusion and PAD score level fusion. Based on the obtained results, the proposed method has indicated the best performance in detecting both known and unknown attacks.

Multi-spectral imaging has recently acquired significant attention in biometrics based authentication due to it’s potential ability to capture spatio-spectral images across the electromagnetic spectrum. Especially, in the case of facial biometrics, multi-spectral imaging has shown significant promising results under unknown/varying illumination environment. However, the challenge arises when surveillance cameras provide the visible images while the enrollment are spectral band images. In order to address the backward/cross compatibility of probing visible images from regular surveillance cameras against the high quality spectral band images in enrollment, development of robust algorithms are required. In this paper, we present a new approach of selecting optimal band based on highest correlation coefficients of individual feature vectors from bands in comparison with feature vectors from visible images of respective individual classes for robust recognition performance. The proposed approach of band selection is validated on a newly collected face database of 168 subjects whose face images are collected in 9 different spectral bands and correspondingly their visible images from a regular camera operating in visible spectrum. The extensive set of experiments conducted on the new database with selected single band and multiple spectral bands in enrollment data versus the visible probe image has indicated the significance of the band selection. The new approach of spectral to visible matching with the proposed band selection method shows significant Rank- 1 recognition rate of 94.04% supporting the applicability of proposed method

Morphed face images are artificially generated images, which blend the facial images of two or more different data subjects into one. The resulting morphed image resembles the constituent faces, both in visual and feature representation. If a morphed image is enroled as a probe in a biometric system, the data subjects contributing to the morphed image will be verified against the enroled probe. As a result of this infiltration, which is referred to as morphed face attack, the unambiguous assignment of data subjects is not warranted, i.e. the unique link between subject and probe is annulled. In this work, we investigate the vulnerability of biometric systems to such morphed face attacks by evaluating the techniques proposed to detect morphed face images. We create two new databases by printing and scanning digitally morphed images using two different types of scanners, a flatbed scanner and a line scanner. Further, the newly created databases are employed to study the vulnerability of state-of-the-art face recognition systems with a comprehensive evaluation.

Contact lens detection in the eye is a significant task to improve the reliability of iris recognition systems. A contact lens overlays the iris region and prevents the iris sensor from capturing the normal iris region. In this paper, we present a novel scheme for detection to detecting a contact lens using Deep Convolutional Neural Network (CNN). The proposed CNN architecture ContlensNet is structured to have fifteen layers and configured for the three-class detection problem with the following classes: images with textured (or colored) contact lens, soft (or transparent) contact lens, and no contact lens. The proposed ContlensNet is trained using numerous iris image patches and the problem of overfitting the network is addressed by using the dropout regularization method. Extensive experiments are carried out on two publicly available large-scale databases, namely: IIIT-Delhi Contact lens iris database (IIITD) and Notre Dame cosmetic contact lens database 2013 (ND) that are comprised of contact lens iris samples captured using four different sensors. The obtained results have demonstrated the improved performance of the proposed scheme with an average performance improvement of more than 10% in Correct Classification Rate (CCR%) when compared with eight different state-of-the-art contact lens detection systems.

Presentation attacks (a.k.a, direct attacks or spoofing attacks) against face recognition systems have emerged as a serious security threat. To mitigate these attacks on conventional face recognition systems, several Presentation Attack Detection (PAD) algorithms have been developed, which address various Presentation Attack Instruments (PAI) including 3D face masks, 2D photo, wrap photo and electronic display, that can be used for the attack. In this paper, we demonstrate and evaluate the vulnerability of an extended Multispectral face recognition system. The extended Multispectral system captures the face image across various spectral bands, thus, we propose to study each of these spectral bands for the vulnerability towards presentation attacks. We have employed a commercial Multispectral camera - SpectraCam™ that can capture seven different spectral bands to collect both bona-fide (a.k.a, live or normal or real) samples as well as artefact (or spoof) face samples. Extensive experiments are carried out on the newly compiled database to provide insights on the vulnerability of the extended Multispectral face system towards PAI generated using a printer. We have created the face artefacts using two different printers, which include laser and inkjet printers. Further, we have also evaluated the state-of-the-art PAD algorithms that are widely employed in conventional face PAD systems. Our study reveals the vulnerability of extended Multispectral face recognition system with respect to the print attack. The results obtained using state-of-the-art PAD algorithms further indicate the challenge to detect the presentation attacks in extended Multispectral face recognition systems.

Biometric authentication based on face recognition acquired enormous attention due to its non-intrusive nature of image capture. Recently, with the advancement in sensor technology, face recognition based on Multi-spectral imaging has gained lot of popularity due to its potential of capturing discrete spatio-spectral images across the electromagnetic spectrum. Our contribution here is to study empirically, the extensive comparative performance analysis of 22 photometric illumination normalization techniques for robust Multi-spectral face recognition. To evaluate this study, we developed a Multi-spectral imaging sensor that can capture Multi-spectral facial images across nine different spectral band in the wavelength range from 530 nm to 1000 nm. With the developed sensor we captured Multi-spectral facial database for 231 individuals, which will be made available in the public domain for the researcher community. Further, quantitative experimental performance analysis in the form of identification rate at rank 1, was conducted on 22 photometric normalization techniques using four state-of-the-art face recognition algorithms. The performance analysis indicates outstanding results with utmost all of the photometric normalization techniques for six spectral bands such as 650 nm, 710 nm, 770 nm, 830 nm, 890 nm, 950 nm.

Multispectral palmprint is considered as an effective biometric modality to accurately recognize a subject with high confidence. This paper presents a novel multispectral palmprint recognition system consisting of three functional blocks namely: (1) novel technique to extract Region of Interest (ROI) from the hand images acquired using a contact less sensor (2) novel image fusion scheme based on dependency measure (3) new scheme for feature extraction and classification. The proposed ROI extraction scheme is based on locating the valley regions between fingers irrespective of the hand pose. We then propose a novel image fusion scheme that combines information from different spectral bands using a Wavelet transform from various sub-bands. We then perform the statistical dependency analysis between these sub-bands to perform fusion either by selection or by weighted fusion. To effectively process the information from the fused image, we perform feature extraction using Log-Gabor transform whose feature dimension is reduced using Kernel Discriminant Analysis (KDA) before performing the classification by employing a Sparse Representation Classifier (SRC). Extensive experiments are carried out on a CASIA multispectral palmprint database that shows the strong superiority of our proposed fusion scheme when benchmarked with contemporary state-of-the-art image fusion schemes. HighlightsNovel adaptive weighted image fusion scheme to combine different multi-spectral palmprints.Dynamic switching between selection and weighted fusion to achieve high accuracy.Novel Scheme for palmprint ROI extraction based on hand shape.Extensive qualitative and quantitative analysis of the proposed (ROI & fusion) scheme.Benchmarking with well known state-of-the-art schemes that include 5 different contemporary multi-spectral image fusion schemes and 2 well known state-of-the-art palmprint recognition schemes.

Face recognition has attained a greater importance in bio-metric authentication due to its non-intrusive property of identifying individuals at varying stand-off distance. Face recognition based on multi-spectral imaging has recently gained prime importance due to its ability to capture spatial and spectral information across the spectrum. Our first contribution in this paper is to use extended multi-spectral face recognition in two different age groups. The second contribution is to show empirically the performance of face recognition for two age groups. Thus, in this paper, we developed a multi-spectral imaging sensor to capture facial database for two different age groups (≤ 15years and ≥ 20years) at nine different spectral bands covering 530nm to 1000nm range. We then collected a new facial images corresponding to two different age groups comprises of 168 individuals. Extensive experimental evaluation is performed independently on two different age group databases using four different state-of-the-art face recognition algorithms. We evaluate the verification and identification rate across individual spectral bands and fused spectral band for two age groups. The obtained evaluation results shows higher recognition rate for age groups ≥ 20years than ≤ 15years, which indicates the variation in face recognition across the different age groups.

Multi-spectral face recognition has been an active area of research over the past few decades. However, the vulnerability of multi-spectral face recognition systems is a growing concern that argues the need for Presentation Attack Detection (PAD) (or countermeasure or anti-spoofing) schemes to successfully detect targeted attacks. In this work, we present a novel feature descriptor L α MT i F that can effectively capture time-frequency features from the maximum response obtained on the high pass band image, which is obtained from the scale-space decomposition of the presented image. The proposed feature descriptor can effectively capture the micro-texture patterns that can be effectively used describe the variation from the presented image. We then propose a new framework using the proposed L α MT i F features that process the input multi-spectral face image independently. These extracted features are then classified using a linear Support Vector Machine (SVM) to obtain the binary decision. Finally, we carry out a decision fusion using the And rule to obtain the final decision. Extensive experiments are carried out on publicly available multi-spectral face datasets that have indicated the efficacy of the proposed scheme.

Biometric recognition of an individual can be done based on physiological characteristics such as a face, palmprint, vein, etc. and behavioural characteristics which include gait, keystroke, mouse dynamics, etc. Among the various biometric modalities, the wrist vein recognition has been known for high accuracy, stability and resistance to spoofing. The success of wrist vein biometrics strongly correlates with the quality of the wrist vein image captured using the sensor. In this paper, we present a new wrist vein sensor that can capture a high-quality wrist vein image using Near infrared (NIR) lighting. The NIR lighting employed in this work can emit light in a spectrum of 940nm that in turn is used to illuminate the wrist area of the hand. The presented sensor employs a single camera with a simple physical structure that will further improve the quality of the light to properly illuminate the wrist hand region to capture good quality wrist vein images. Extensive experiments are carried out on our newly collected database comprised of 50 subjects resulting in 100 unique wrist veins. We also present a comprehensive evaluation of nine different state-of-the-art techniques that demonstrated the outstanding performance of the Log-Gabor and Sparse Representation Classifier (SRC) with an EER of 1.63% and the GMR of 94.01 at FMR = 10−1%.

Periocular characteristics are being used as a supplementary feature for the biometric systems employing iris characteristics to mitigate effects of the noisy iris on authentication performance. In the same lines, ocular characteristics are also used to enhance the performance of face based systems under the impact of pose, expression and illumination. However, the iris and face systems are operated in Near-Infra-Red (NIR) and visible spectrum respectively. In order have both the systems work for ocular images and be compatible to each other, the biometric system needs to be robust enough to handle the biometric data emerging from different spectrum. In this work, we employ an ocular image database collected using the visible and NIR cameras. We propose a new framework employing a bank of Binarized Statistical Image filters along with χ2 distance metric along with simple fusion to handle the cross-spectrum data. Set of experiments conducted on the cross-spectrum periocular database indicate the robustness of the system with the achieved GMR of 96.04% at the FMR of 0.01%. The obtained performance indicates the applicability of proposed framework for realistic cross-spectrum biometric authentication scenario.

Multi-spectral face recognition has acquired significant attention over a last few decades due to its potential of capturing spatial and spectral information across the electromagnetic spectrum. In this paper, we present a new imaging scheme that can obtain the multi-spectral face image at nine different spectra covering 530nm–1000nm. We prepared a new database comprising of 230 subjects using our new low-cost multi-spectral face imaging device. Extensive experiments are presented for evaluating the performance of the four different state-of-the-art face recognition algorithms on both individual bands and the fused spectral face image. Obtained results show the improved face recognition performance of Log-Gabor features with Collaborative Representation (CRC) as the classifier.

Widespread deployment of Automatic Border Control (ABC) along with the electronic Machine Readable Travel Documents (eMRTD) for person verification has enabled a prominent use case of face biometrics in border control applications. Many countries issue eMRTD passports on the basis of a printed biometric face photo submitted by the applicant. Some countries offer web-portals for passport renewal, where citizens can upload their face photo. These applications allow the possibility of the photo being altered to beautify the appearance of the data subject or being morphed to conceal the applicant identity. Specifically, if an eMRTD passport is issued with a morphed facial image, two or more data subjects, likely the known applicant and one or more unknown companion(s), can use such passport to pass a border control. In this work we propose a novel scheme to detect morphed face images based on facial micro-textures extracted using statistically independent filters that are trained on natural images. Given a face image, the proposed method will obtain a micro-texture variation using Binarized Statistical Image Features (BSIF), and the decision is made using a linear Support Vector Machine (SVM). This is first work carried out towards detecting the morphed face images. Extensive experiments are carried out on a large-scale database of 450 morphed face images created using 110 unique subjects with different ethnicity, age, and gender that indicates the superior performance.

Smartphone based periocular recognition has received substantial attention from the biometric research community. In this work, we propose a new scheme for the smartphone based periocular recognition. The proposed scheme is based on the texture features extracted from the periocular images using Maximum Response (MR) filters. These texture features are then classified using a deep neural network based on deeply coupled autoencoders. Extensive experiments are carried out on the large-scale VISOB database with 550 subjects captured using three different smartphones. The obtained results demonstrate the average performance of the proposed scheme with GMR of over 92% at FMR = 10−3.

The challenges of presentation attacks (spoofing attacks) at sensor level is increasing for biometric systems due to the evolving method of artefact presentation. The sophisticated attacks now employ high quality printed artefacts and electronic screens to present the biometric samples which make it difficult to separate the real presentations and artefact presentations. In this work, we propose a new scheme to detect the artefacts in both NIR and visible spectrum biometric sensors for ocular biometric systems using a new set of feature descriptor. The scheme employs adaptive and quantized texture patters obtained from local microfeatures and global spatial features for different color channels in an image. Further, the texture descriptors are used to learn a spectrally regressed discriminant classifier to classify the normal ocular images against the artefact ocular images. The proposed scheme is used to perform extensive experiments on 5 publicly available ocular datasets including 2 datasets acquired in NIR domain and two datasets acquired using smartphones along with a dataset acquired using high quality camera. The experiments conducted on all the datasets have consistently indicated the robust performance against attacks by showing a classification error of 0%.