Logo des Repositoriums

P296 - BIOSIG 2019 - Proceedings of the 18th International Conference of the Biometrics Special Interest Group

https://dl.gi.de/handle/20.500.12116/34220
Auflistung nach:

Auflistung P296 - BIOSIG 2019 - Proceedings of the 18th International Conference of the Biometrics Special Interest Group nach Erscheinungsdatum

Zu einem Punkt im Index springen:
1 - 10 von 23
  • Konferenzbeitrag
    Style Your Face Morph and Improve Your Face Morphing Attack Detector
    (BIOSIG 2019 - Proceedings of the 18th International Conference of the Biometrics Special Interest Group, 2019) Seibold, Clemens; Hilsmann, Anna; Eisert, Peter
    A morphed face image is a synthetically created image that looks so similar to the faces of two subjects that both can use it for verification against a biometric verification system. It can be easily created by aligning and blending face images of the two subjects. In this paper, we propose a style transfer based method that improves the quality of morphed face images. It counters the image degeneration during the creation of morphed face images caused by blending. We analyze different state of the art face morphing attack detection systems regarding their performance against our improved morphed face images and other methods that improve the image quality. All detection systems perform significantly worse, when first confronted with our improved morphed face images. Most of them can be enhanced by adding our quality improved morphs to the training data, which further improves the robustness against other means of quality improvement.
  • Konferenzbeitrag
    Gait verification using deep learning with a pairwise loss
    (BIOSIG 2019 - Proceedings of the 18th International Conference of the Biometrics Special Interest Group, 2019) Yalavarthi, Vijaya Krishna; Grabocka, Josif; Mandalapu, Hareesh; Schmidt-Thieme, Lars
    A unique walking pattern to every individual makes gait a promising biometric. Gait is becoming an increasingly important biometric because it can be captured non-intrusively through accelerometers positioned at various locations on the human body. The advent of wearable sensors technology helps in collecting the gait data seamlessly at a low cost. Thus gait biometrics using accelerometers play significant role in security-related applications like identity verification and recognition. In this work, we deal with the problem of identity verification using gait. As the data received through the sensors is indexed in time order, we consider identity verification through gait data as the time series binary classification problem. We present deep learning model with a pairwise loss function for the classification.We conducted experiments using two datasets: publicly available ZJU dataset of more than 150 subjects and our self collected dataset with 15 subjects. With our model, we obtained an Equal Error Rate of 0.05% over ZJU dataset and 0.5% over our dataset which shows that our model is superior to the state-of-the-art baselines.
  • Konferenzbeitrag
    Fast and Accurate Continuous User Authentication by Fusion of Instance-based, Free-text Keystroke Dynamics
    (BIOSIG 2019 - Proceedings of the 18th International Conference of the Biometrics Special Interest Group, 2019) Ayotte, Blaine; Banavar, Mahesh K.; Hou, Daqing; Schuckers, Stephanie
    Keystroke dynamics study the way in which users input text via their keyboards, which is unique to each individual, and can form a component of a behavioral biometric system to improve existing account security. Keystroke dynamics systems on free-text data use n-graphs that measure the timing between consecutive keystrokes to distinguish between users. Many algorithms require 500, 1,000, or more keystrokes to achieve EERs of below 10%. In this paper, we propose an instancebased graph comparison algorithm to reduce the number of keystrokes required to authenticate users. Commonly used features such as monographs and digraphs are investigated. Feature importance is determined and used to construct a fused classifier. Detection error tradeoff (DET) curves are produced with different numbers of keystrokes. The fused classifier outperforms the state-of-the-art with EERs of 7.9%, 5.7%, 3.4%, and 2.7% for test samples of 50, 100, 200, and 500 keystrokes.
  • Konferenzbeitrag
    Multi-resolution Local Descriptor for 3D Ear Recognition
    (BIOSIG 2019 - Proceedings of the 18th International Conference of the Biometrics Special Interest Group, 2019) Ganapathi, Iyyakutti Iyappan; Ali, Syed Sadaf; Prakash, Surya
    Several approaches have shown promising results in human ear recognition. However, factors such as the pose, illumination, and scaling have an enormous impact on recognition performance. 3D models are insensitive to these factors and are found to be better at enhancing recognition performance with strong geometric information. Low cost 3D data acquisition has also boosted the research community in recent times to explore more about 3D object recognition. We present a local multi-resolution descriptor in this paper to recognize human ears in 3D. For each key-point in 3D ear, a local reference frame (LRF) is constructed. Using multi-radii, we find neighbors at each key-point and the neighbors obtained from each radius are projected to create a depth image using the LRF. Further, a descriptor is computed by employing neural network based auto-encoders and the local statistics of the depth images. The descriptor is used to locate the potential correspondence matching points in the probe and gallery images for a coarse arrangement, followed by a fine alignment to compute the registration error. The obtained registration error is used as the matching score. The proposed technique is assessed on UND-J2 dataset to demonstrate its effectiveness.
  • Konferenzbeitrag
    Adversarial learning for a robust iris presentation attack detection method against unseen attack presentations
    (BIOSIG 2019 - Proceedings of the 18th International Conference of the Biometrics Special Interest Group, 2019) Ferreira, Pedro M.; Sequeira, Ana F.; Pernes, Diogo; Rebelo, Ana; Cardoso, Jaime S.
    Despite the high performance of current presentation attack detection (PAD) methods, the robustness to unseen attacks is still an under addressed challenge. This work approaches the problem by enforcing the learning of the bona fide presentations while making the model less dependent on the presentation attack instrument species (PAIS). The proposed model comprises an encoder, mapping from input features to latent representations, and two classifiers operating on these underlying representations: (i) the task-classifier, for predicting the class labels (as bona fide or attack); and (ii) the species-classifier, for predicting the PAIS. In the learning stage, the encoder is trained to help the task-classifier while trying to fool the species-classifier. Plus, an additional training objective enforcing the similarity of the latent distributions of different species is added leading to a ‘PAIspecies’- independent model. The experimental results demonstrated that the proposed regularisation strategies equipped the neural network with increased PAD robustness. The adversarial model obtained better loss and accuracy as well as improved error rates in the detection of attack and bona fide presentations.
  • Konferenzbeitrag
    Region-Based CNNs for Pedestrian Gender Recognition in Visual Surveillance Environments
    (BIOSIG 2019 - Proceedings of the 18th International Conference of the Biometrics Special Interest Group, 2019) Yaghoubi, Ehsan; Alirezazadeh, Pendar; Assunção, Eduardo; Neves, João C.; Proença, Hugo
    Inferring soft biometric labels in totally uncontrolled outdoor environments, such as surveillance scenarios, remains a challenge due to the low resolution of data and its covariates that might seriously compromise performance (e.g., occlusions and subjects pose). In this kind of data, even state-of-the-art deep-learning frameworks (such as ResNet) working in a holistic way, attain relatively poor performance, which was the main motivation for the work described in this paper. In particular, having noticed the main effect of the subjects’ “pose” factor, in this paper we describe a method that uses the body keypoints to estimate the subjects pose and define a set of regions of interest (e.g., head, torso, and legs). This information is used to learn appropriate classification models, specialized in different poses/body parts, which contributes to solid improvements in performance. This conclusion is supported by the experiments we conducted in multiple real-world outdoor scenarios, using the data acquired from advertising panels placed in crowded urban environments.
  • Konferenzbeitrag
    Multi-algorithm Benchmark for Fingerprint Presentation Attack Detection with Laser Speckle Contrast Imaging
    (BIOSIG 2019 - Proceedings of the 18th International Conference of the Biometrics Special Interest Group, 2019) Kolberg, Jascha; Gomez-Barrero, Marta; Busch, Christoph
    The increased usage of biometric authentication systems has raised concerns regarding the security of components in a biometric system. As a consequence, preventing security issues related to presentation attacks targeting the biometric capture device are of utmost importance. To develop presentation attack detection (PAD) mechanisms, features confirming the liveness of the biometric characteristic such as the blood flow within the finger are needed. Utilising laser speckle contrast imaging (LSCI) to observe blood movement below the surface, we present an evaluation of different machine learning classifiers for fingerprint PAD. The experiments over a database comprising 35 different presentation attack instrument (PAI) species show that the detection performance varies depending on the utilised feature extraction method. A majority voting of selected classifiers and features achieves an APCER of 9% for a convenient BPCER of 0.05%.
  • Konferenzbeitrag
    Perspective Multiplication for Multi-Perspective Enrolment in Finger Vein Recognition
    (BIOSIG 2019 - Proceedings of the 18th International Conference of the Biometrics Special Interest Group, 2019) Prommegger, Bernhard; Uhl, Andreas
    Finger vein recognition deals with the identification of subjects based on their venous pattern within the fingers. It has been shown that its recognition accuracy heavily depends on a good alignment of the acquired samples. There are several approaches that try to reduce the impact of finger misplacement. However, none of this approaches is able to prevent all possible types of finger misplacements. As finger vein scanners are evolving towards contact-less acquisition, alignment problems, especially due to longitudinal finger rotation, are becoming even more important. One way to tackle this problem is capturing the vein structure from different perspectives during enrolment, but cost and complexity of capturing devices increases with the number of involved cameras. In this article, a new method to reduce the number of cameras needed for multi-perspective enrolment is presented. The reduction is achieved by introducing additional pseudo perspectives in-between two adjacent cameras. The obtained perspectives are used for additional comparisons during authentication. This way, the complexity of the enrolment devices can be reduced while keeping the recognition performance at a high level.
  • Konferenzbeitrag
    Is There Any Similarity Between a Person’s Left and Right Retina?
    (BIOSIG 2019 - Proceedings of the 18th International Conference of the Biometrics Special Interest Group, 2019) Biswas, Sangeeta; Rohdin, Johan; Mňuk, Tomáš; Drahanský, Martin
    It is often argued among biometric researchers that the left and right retinas of the same person are as different as the retinas of two different persons. In this paper we investigate to what extent this is true. We perform experiments where human volunteers are asked to judge whether a pair of the left and right retinal images displayed side-by-side belongs to the same person or two different persons. We also use two similarity measurements, structural similarity (SSIM) and cosine similarity, to do the investigation process automatically. Our experiments show that there is recognizable similarity in the left and right retina of a person. For a verification task done by human volunteers, the average accuracy was 82%. For identification tasks, automatic systems using cosine similarity were correct in up to 57%.
  • Konferenzbeitrag
    Gender and Kinship by Model-Based Ear Biometrics
    (BIOSIG 2019 - Proceedings of the 18th International Conference of the Biometrics Special Interest Group, 2019) Meng, Di; Nixon, Mark S.; Mahmoodi, Sasan
    Many studies in biometrics have shown how identity can be determined, including by images of ears. In the paper, we show how model an ear and how the gender appears to often be manifest in the ear structures, as is kinship or family relationship. We describe a new model-based approach for viewpoint correction and ear description to enable this analysis. We show that with the new technique having satisfactory basic recognition capability (recognizing individuals with performance similar to state of art), gender can achieve 67.2% and kinship 40.4% rank 1 recognition on ears from subjects with unconstrained pose.