Automatic speaker verification (ASV) technology now reports a reasonable level of accuracy in its applications in voice-based biometric systems. However, it requires adequate amount of speech data for enrolment and verification; otherwise, the performance becomes considerably degraded. For this reason, the trade-off between the convenience and security is difficult to maintain in practical scenarios. The utterance duration remains a critical issue while deploying a voice biometric system in real-world applications. A large amount of research work has been carried out to address the limited data issue within the scope of SV. The advancements and research activities in mitigating the challenges due to short utterance have seen a significant rise in recent times. In this study, the authors present an extensive survey of SV with short utterances considering the studies from recent past and include latest research offering various solutions and analyses. The review also summarises the major findings of the studies of duration variability problem in ASV systems. Finally, they discuss a number of possible future directions promoting further research in this field.

The performance of fingerprint comparison algorithms depends on the reliability and accuracy of the features extracted from the fingerprints. The accuracy of the feature extraction algorithms is assumed to depend on the quality of the fingerprint images. Especially, low-quality images can be challenging for feature extraction algorithms. Image enhancement may allow to extract features more accurately. There is a lack of extensive and quantitative evaluation of image enhancement methods. This study investigates the impact of seven typical image enhancement methods on biometric sample quality and on biometric performance. The interrelation of image quality and biometric performance is investigated on 14 datasets. Biometric quality measures are estimated based on image quality metrics NFIQ1 and NFIQ2.0. Biometric performance is tested using MINDTCT and FingerJetFX for feature extraction and BOZORTH3 for biometric comparison. This work shows that the biometric performance can be improved by image enhancement. The significance of improvements depends on both the quality of the datasets and the feature extraction. Thus, there is no single best improvement algorithm. A correlation of changes in scores and image qualities can only be found on the level of entire datasets. No significant correlation can be found for single biometric comparisons.

Age estimation at a distance has potential applications including visual surveillance and monitoring in public places. Far from the camera, image resolution is significantly degraded. In fact, age estimation using classical methods such as face is not reliable. Given that gait is very sensitive to ageing, gait analysis is the suitable solution for age estimation at a great distance from the camera. Medical and biomechanical studies prove that older adults adapt their walking toward a safer and more stable gait and an established balance. Indeed, in this study the authors propose a gait-based descriptor for age classification using a silhouette projection model. The proposed model encapsulates both spatiotemporal longitudinal (SLP) and transverse (STP) projections of the silhouette during a gait cycle. The proposed model aims to represent the arms' swing, the head's pitch, the hunched posture and the stride's length, which are among the most outstanding ageing characteristics that appear on the elderly's gait. Although age classification using gait is a very challenging task, SLP and STP curves analysis shows a considerable discrimination between young and elderly people. Also, experiments conducted on the OU-ISIR database prove that their proposed descriptor outperforms existing ones by reaching an important recognition rate.

Person re-identification is the process of finding people across different cameras. In this process, focus often lies in developing strong feature descriptors or a robust metric learning algorithm. While the two aspects are the most important steps in order to secure a high performance, a less explored aspect is late fusion of complementary features. For this purpose, this study proposes a late fusing scheme that, based on an experimental analysis, combines three systems that focus on extracting features and provide supervised learning on different abstraction levels. To analyse the behaviour of the proposed system, both rank aggregation and score-level fusion are applied. The authors’ proposed fusion scheme increases results on both small and large datasets. Experimental results on VIPeR show accuracies 5.43% higher than related systems, while results on PRID450S and CUHK01 increase state-of-the-art results by 10.94 and 14.84%, respectively. Furthermore, a cross-dataset test shows an increased rank-1 accuracy of 28.26% when training on CUHK02 and testing on VIPeR. Finally, an analysis of the late fusion shows aggregation to be better when individual results are unequally distributed within top-10 while score-level fusion provides better results when two individual results lie within top-5 while the last lies outside top-10.

This work is inspired by the observation of surprising daily fluctuations in the number of valid iris code bits used to match irises in the NEXUS program operated by the Canadian Border Security Agency. These fluctuations have an impact on iris comparison scores but cannot be simply explained by pupil dilation, which does not have a clear pattern that would generalise to a population. To check if fluctuations in the number of valid iris code bits may be explained by eyelid aperture observed in a controlled, laboratory environment, the eyelid aperture was measured for 18 subjects participating in an acquisition every 2 h during the day. Simultaneously, the pupil dilation was measured to check the existence of a daily pattern for a population and for single subjects. There are two interesting outcomes of this work. First, there are statistically significant changes during the day in both pupil dilation and eyelid opening observed for individual subjects. Second, these changes do not generalise well into a common pattern for the group. Consequently, the diurnal fluctuations in the number of bits compared and the comparison score observed in the NEXUS program cannot be explained by changes in pupil dilation nor by eyelid aperture.

This paper proposes an Incremental Fuzzy-Rough Nearest Neighbour (IncFRNN) technique for biometric authentication modelling using feature extracted visual evoked. Only small training set is needed for model initialisation. The embedded heuristic update method adjusts the knowledge granules incrementally to maintain all representative electroencephalogram (EEG) signal patterns and eliminate those rarely used. It reshapes the personalized knowledge granules through insertion and deletion of a test object, based on similarity measures. A predefined window size can be used to reduce the overall processing time. This proposed algorithm was verified with test data from 37 healthy subjects. Signal pre-processing steps on segmentation, filtering and artefact rejection were carried out to improve the data quality before model building. The experimental paradigm was designed in three different conditions to evaluate the authentication performance of the IncFRNN technique against the benchmarked incremental K-Nearest Neighbour (KNN) technique. The performance was measured in terms of accuracy, area under the Receiver Operating Characteristic (ROC) curve (AUC) and Cohen's Kappa coefficient. The proposed IncFRNN technique is proven to be statistically better than the KNN technique in the controlled window size environment. Future work will focus on the use of dynamic data features to improve the robustness of the proposed model.

Computational intelligence is employed to solve factual and complicated global problems, though neural networks (NNs) and evolutionary computing have also affected these issues. Biometric traits are applicable for detecting crime in security systems because they offer attractive features such as stability and uniqueness. Although various methods have been proposed for this objective, feature shortcomings such as computational complexity, long run times, and high memory consumption remain. The current study proposes a novel human iris recognition approach based on a multi-layer perceptron NN and particle swarm optimisation (PSO) algorithms to train the network in order to increase generalisation performance. A combination of these algorithms was used as a classifier. A pre-processing step was performed on the iris images to improve the results and two-dimensional gabor kernel feature extraction was applied. The data was normalised, trained, and tested using the proposed method. A PSO algorithm was applied to train the NN for data classification. The experimental results show that the proposed method performs better than many other well-known techniques. The benchmark Chinese Academy of Science and Institute of Automation (CASIA)-iris V3 and Center for Machine Learning and Intelligent Systems at the University of California, Irvine (UCI) machine learning repository datasets were used for testing and comparison.

For Entry-Exit technologies, such as US VISIT and Smart Borders (e-borders), a watchlist normally contains high-quality biometric traits and is checked only against visitors. The situation can change drastically if low-quality images are added into the watchlist. Motivated by this fact, we introduce a systematic approach to assessing the risk of travellers using a biometric-enabled watchlist where some latency of the biometric traits is allowed. The main results presented herein include: (1) a taxonomical view of the watchlist technology, and (2) a novel risk assessment technique. For modelling the watchlist landscape, we propose a risk categorisation using the Doddington metric. We evaluate via experimental study on large-scale facial and fingerprint databases, the risks of impersonation and mis-identification in various screening scenarios. Other contributions include a study of approaches to designing a biometric-enabled watchlist for e-borders: a) risk control and b) improving performance of the e-border via integrating the interview supporting machines.