Santosh Tirunagari

Detection of Face Spoofing Using Visual Dynamics

Rendering a face recognition system robust is vital in order to safeguard it against spoof attacks carried out using printed pictures of a victim (also known as print attack) or a replayed video of the person (replay attack). A key property in distinguishing a live, valid access from printed media or replayed videos is by exploiting the information dynamics of the video content, such as blinking eyes, moving lips, and facial dynamics. We advance the state of the art in facial antispoofing by applying a recently developed algorithm called dynamic mode decomposition (DMD) as a general purpose, entirely data-driven approach to capture the above liveness cues. We propose a classification pipeline consisting of DMD, local binary patterns (LBPs), and support vector machines (SVMs) with a histogram intersection kernel. A unique property of DMD is its ability to conveniently represent the temporal information of the entire video as a single image with the same dimensions as those images contained in the video. The pipeline of DMD + LBP + SVM proves to be efficient, convenient to use, and effective. In fact only the spatial configuration for LBP needs to be tuned. The effectiveness of the methodology was demonstrated using three publicly available databases: (1) print-attack; (2) replay-attack; and (3) CASIA-FASD, attaining comparable results with the state of the art, following the respective published experimental protocols.

The 1st Competition on Counter Measures to Finger Vein Spoofing Attacks

The vulnerability of finger vein recognition to spoofing attacks has emerged as a crucial security problem in the recent years mainly due to the high security applications where biometric technology is used. Recent works shown that finger vein biometrics is vulnerable to spoofing attacks, pointing out the importance to investigate counter-measures against this type of fraudulent actions. The goal of the 1st Competition on Counter Measures to Finger Vein Spoofing Attacks is to challenge researchers to create counter-measures that can detect printed attacks effectively. The submitted approaches are evaluated on the Spoofing-Attack Finger Vein Database and the achieved results are presented in this paper.

Windowed DMD as a microtexture descriptor for finger vein counter-spoofing in biometrics

Recent studies have shown that it is possible to attack a finger vein (FV) based biometric system using printed materials. In this study, we propose a novel method to detect spoofing of static finger vein images using Windowed Dynamic mode decomposition (W-DMD). This is an atemporal variant of the recently proposed Dynamic Mode Decomposition for image sequences. The proposed method achieves better results when compared to established methods such as local binary patterns (LBP), discrete wavelet transforms (DWT), histogram of gradients (HoG), and filter methods such as range-filters, standard deviation filters (STD) and entropy filters, when using SVM with a minimum intersection kernel. The overall pipeline which consists ofW-DMD and SVM, proves to be efficient, and convenient to use, given the absence of additional parameter tuning requirements. The effectiveness of our methodology is demonstrated using FV-Spoofing-Attack database which is publicly available. Our test results show that W-DMD can successfully detect printed finger vein images because they contain micro-level artefacts that not only differ in quality but also in light reflection properties compared to valid/live finger vein images.

Patient level analytics using self-organising maps: A case study on Type-1 Diabetes self-care survey responses

Survey questionnaires are often heterogeneous because they contain both quantitative (numeric) and qualitative (text) responses, as well as missing values. While traditional, model-based methods are commonly used by clinicians, we deploy Self Organizing Maps (SOM) as a means to visualise the data. In a survey study aiming at understanding the self-care behaviour of 611 patients with Type-1 Diabetes, we show that SOM can be used to (1) identify co-morbidities; (2) to link self-care factors that are dependent on each other; and (3) to visualise individual patient profiles; In evaluation with clinicians and experts in Type-1 Diabetes, the knowledge and insights extracted using SOM correspond well to clinical expectation. Furthermore, the output of SOM in the form of a U-matrix is found to offer an interesting alternative means of visualising patient profiles instead of a usual tabular form.

Data forensic techniques using Benford's law and Zipf's law for keystroke dynamics

The selection and application of biometrics traits for authentication and identification have recently attracted a significant amount of research interest. In this paper we investigate the use of keystroke data to distinguish between humans using keystroke biometric systems and non-humans for auditing application. Recently, Benfords Law and Zipfs Law, which are both discrete Power law probability distributions, have been effectively used to detect fraud and discriminate between genuine data and fake/tampered data. As such, our motivation is to apply the Benfords Law and Zipfs Law on keystroke data and to determine whether they follow these laws and discriminate between humans using keystroke biometric systems from non-humans. From the results, we observe that, the latency values of the keystroke data from humans actually follow the Benfords law and Zipfs law, but not the duration values. This implies that, latency values from humans would follow the two laws, whereas the latency values from non-humans would deviate from the Benfords law and Zipfs law. Even though, the duration values from humans deviates from the Benfords law, they do follow a pattern that we can develop an accurate model for the duration values. We perform experiments using the benchmark data set developed by Killourhy and Maxion, CMU [1] and obtain divergences of 0.0008, 0.029 and 0.05 for the keyup-keydown (latency), keydown-keydown, and duration of the keystroke data, respectively. Moreover, P-values of 0.7770, 0.6230 and 0.0160 are obtained for the keyup-keydown (latency), keydown-keydown, and duration of the keystroke data, respectively. We observe that the latency (which is the time elapsed between release of the first key and pressing down of the next key) is one of the most important features used by administrators for auditing purposes to detect anomalies during their employees logging into their company system.

Can DMD obtain a Scene Background in color

A background model describes a scene without any foreground objects and has a number of applications, ranging from video surveillance to computational photography. Recent studies have introduced the method of Dynamic Mode Decomposition (DMD) for robustly separating video frames into a background model and foreground components. While the method introduced operates by converting color images to grayscale, we in this study propose a technique to obtain the background model in the color domain. The effectiveness of our technique is demonstrated using a publicly available Scene Background Initialisation (SBI) dataset. Our results both qualitatively and quantitatively show that DMD can successfully obtain a colored background model.

Local binary patterns as a feature descriptor in alignment-free visualisation of metagenomic data

Shotgun sequencing has facilitated the analysis of complex microbial communities. However, clustering and visualising these communities without prior taxonomic information is a major challenge. Feature descriptor methods can be utilised to extract these taxonomic relations from the data. Here, we present a novel approach consisting of local binary patterns (LBP) coupled with randomised singular value decomposition (RSVD) and Barnes-Hut t-stochastic neighbor embedding (BH-tSNE) to highlight the underlying taxonomic structure of the metagenomic data. The effectiveness of our approach is demonstrated using several simulated and a real metagenomic datasets.

Automatic detection of acute kidney injury episodes from primary care data

Acute kidney injury (AKI) is characterised by a rapid deterioration in kidney function, and can be identified by examining the rate of change in a patients estimated glomerular filtration rate (eGFR) signal. Due to the potentially irreversible nature of the damage AKI episodes cause to renal function, their detection plays a significant role in predicting a kidneys effectiveness. Although algorithms for the detection of AKI are available for patients under constant monitoring, e.g. inpatients, their applicability to primary care settings is less clear as the eGFR signal often contains large lapses in time between measurements. However, waiting for hospital admittance before AKI is undesirable, as detecting AKI early can help to mitigate the degradation of kidney function and the associated increase in morbidity and mortality. Traditionally, a clinician in a primary care setting would manually identify AKI episodes from direct observation of eGFR signals. While this approach may work for individual patients, the time consuming nature of it precludes quick large-scale monitoring. We therefore present two alternative automated approaches for detecting AKI: as the outlier points when using Gaussian process regression and using a novel technique we entitle Surrey AKI detection algorithm (SAKIDA). Using SAKIDA, we can identify the number of AKI episodes a patient experiences with an accuracy of 70%, when evaluated against the performance of human experts.

Movement correction in DCE-MRI through windowed and reconstruction dynamic mode decomposition

Images of the kidneys using dynamic contrast-enhanced magnetic resonance renography (DCE-MRR) contains unwanted complex organ motion due to respiration. This gives rise to motion artefacts that hinder the clinical assessment of kidney function. However, due to the rapid change in contrast agent within the DCE-MR image sequence, commonly used intensity-based image registration techniques are likely to fail. While semi-automated approaches involving human experts are a possible alternative, they pose significant drawbacks including inter-observer variability, and the bottleneck introduced through manual inspection of the multiplicity of images produced during a DCE-MRR study. To address this issue, we present a novel automated, registration-free movement correction approach based on windowed and reconstruction variants of dynamic mode decomposition (WR-DMD). Our proposed method is validated on ten different healthy volunteers’ kidney DCE-MRI data sets. The results, using block-matching-block evaluation on the image sequence produced by WR-DMD, show the elimination of