Raymond Veldhuis

Biometric search consists of comparing a biometric probe with biometric references in a database to find identities that match the probe. In addition to being resource-intensive, biometric search can lead to privacy violations when biometric data is exposed. To protect the sensitive biometric data, state-of-the-art search solutions are based on fully homomorphic encryption (FHE). However, they either reduce search accuracy to gain efficiency or lack efficiency and fail to meet the 10s NIST-FRVT requirement for a one-in-a-million search. In this paper, we present a fast and accurate biometric search solution under encryption for biometric data represented as vectors, a commonly used compact representation for compressing high-dimensional data. Our search solution is exhaustive and does not require any pre-selection (e.g., indexing) or dimensionality reduction techniques, which can degrade search accuracy. Also, it is parallelizable and supports dynamic databases. Our efficiency gain stems from our chunking and vertical organization of the encrypted reference database. This enables us to optimally leverage the single-instruction multiple-data (SIMD) property of FHE schemes. For a database of one million encrypted references represented by 512-dimensional vectors and a security level of 128 bits, our solution runs a fully encrypted exhaustive search in 9.23s measured on a 64-bit computer Intel Xeon Platinum 8358 with 64 CPUs and 64GB RAM. Hence, our solution is the first to satisfy the 10s NIST-FRVT limit under encryption, outperforming the state-of-the-art solutions by more than two orders of magnitude (at least 246 times faster) while achieving an accuracy improvement of at least 4.6% FNIR for a fixed FPIR at 0.1%.

Deep learning-based face recognition systems produce templates that encode sensitive information next to identity, such as gender and ethnicity. This poses legal and ethical problems as the collection of biometric data should be minimized and only specific to a designated task. We propose two privacy constraints to hide the gender attribute that can be added to a recognition loss. The first constraint relies on the minimization of the angle between gender-centroid embeddings. The second constraint relies on the minimization of the angle between gender specific embeddings and their opposing gender-centroid weight vectors. Both constraints enforce the overlapping of the gender specific distributions of the embeddings. Furthermore, they have a direct interpretation in the embedding space and do not require a large number of trainable parameters as two fully connected layers are sufficient to achieve satisfactory results. We also provide extensive evaluation results across several datasets and face recognition networks, and we compare our method to three state-of-the-art methods. Our method is capable of maintaining high verification performances while significantly improving privacy in a cross-database setting, without increasing the computational load for template comparison. We also show that different training data can result in varying levels of effectiveness of privacy-enhancing methods that implement data minimization.

The recognition performance of biometric systems strongly depends on the quality of the compared biometric samples. Motivated by the goal of establishing a common understanding of face image quality and enabling system interoperability, the committee draft of ISO/IEC 29794-5 introduces expression neutrality as one of many component quality elements affecting recognition performance. In this study, we train classifiers to assess facial expression neutrality using seven datasets. We conduct extensive performance benchmarking to evaluate their classification and face recognition utility prediction abilities. Our experiments reveal significant differences in how each classifier distinguishes "neutral" from "non-neutral" expressions. While Random Forests and AdaBoost classifiers are most suitable for distinguishing neutral from non-neutral facial expressions with high accuracy, they underperform compared to Support Vector Machines in predicting face recognition utility.

Face recognition from side‐view positions poses a considerable challenge in automatic face recognition tasks. Pose variation up to the side‐view is an issue of difference in appearance and visibility since only one eye is visible at the side‐view poses. Traditionally overlooked, recent advancements in deep learning have brought side‐view poses to the forefront of research attention. This survey comprehensively investigates methods addressing pose variations up to side‐view and categorizes research efforts into feature‐based, image‐based, and set‐based pose handling. Unlike existing surveys addressing pose variations, our emphasis is specifically on extreme poses. We report numerous promising innovations in each category and contemplate the utilization and challenges associated with side‐view. Furthermore, we introduce current datasets and benchmarks, conduct performance evaluations across diverse methods, and explore their unique constraints. Notably, while feature‐based methods currently stand as the state‐of‐the‐art, our observations suggest that cross‐dataset evaluations, attempted by only a few researchers, produce worse results. Consequently, the challenge of matching arbitrary poses in uncontrolled settings persists.

This paper studies the impact of illumination direction and bundle width on finger vascular pattern imaging and recognition performance. A qualitative theoretical model is presented to explain the projection of finger blood vessels on the skin. A series of experiments were conducted using a scanner of our design with illumination from the top, a single‐direction side (left or right), and narrow or wide beams. A new dataset was collected for the experiments, containing 4,428 NIR images of finger vein patterns captured under well‐controlled conditions to minimize position and rotation angle differences between different sessions. Top illumination performs well because of more homogenous, which enhances a larger number of visible veins. Narrower bundles of light do not affect which veins are visible, but they reduce the overexposure at finger boundaries and increase the quality of vascular pattern images. The narrow beam achieves the best performance with 0% of FNMR@FMR0.01%, and the wide beam consistently results in a higher false nonmatch rate. The comparison of left‐ and right‐side illumination has the highest error rates because only the veins in the middle of the finger are visible in both images. Different directional illumination may be interoperable since they produce the same vascular pattern and principally are the projected shadows on the finger surface. Score and image fusion for right‐ and left‐side result in recognition performance similar to that obtained with top illumination, indicating the vein patterns are independent of illumination direction. All results of these experiments support the proposed model.