Search results

Personal mobile devices currently have access to a significant portion of their user's private sensitive data and are increasingly used for processing mobile payments. Consequently, securing access to these mobile devices is a requirement for securing access to the sensitive data and potentially costly services. The authors propose and evaluate a first version of a pan shot face unlock method: a mobile device unlock mechanism using all information available from a 180° pan shot of the device around the user's head – utilizing biometric face information as well as sensor data of built‐in sensors of the device. The paper aims to discuss these issues.

This approach uses grayscale 2D images, on which the authors perform frontal and profile face detection. For face recognition, the authors evaluate different support vector machines and neural networks. To reproducibly evaluate this pan shot face unlock toolchain, the authors assembled the 2013 Hagenberg stereo vision pan shot face database, which the authors describe in detail in this article.

Current results indicate that the approach to face recognition is sufficient for further usage in this research. However, face detection is still error prone for the mobile use case, which consequently decreases the face recognition performance as well.

The contributions of this paper include: introducing pan shot face unlock as an approach to increase security and usability during mobile device authentication; introducing the 2013 Hagenberg stereo vision pan shot face database; evaluating this current pan shot face unlock toolchain using the newly created face database.

The purpose of this article is to improve detection of common movement. Detecting if two or multiple devices are moved together is an interesting problem for different applications. However, these devices may be aligned arbitrarily with regards to each other, and the three dimensions sampled by their respective local accelerometers can therefore not be directly compared. The typical approach is to ignore all angular components and only compare overall acceleration magnitudes – with the obvious disadvantage of discarding potentially useful information.

This paper contributes a method to analytically determine relative spatial alignment of two devices based on their acceleration time series. The method uses quaternions to compute the optimal rotation with regards to minimizing the mean squared error.

Based on real-world experimental data from smartphones and smartwatches shaken together, the paper demonstrates the effectiveness of the method with a magnitude squared coherence metric, for which an improved equal error rate (EER) of 0.16 (when using derotation) over an EER of 0.18 (when not using derotation) is shown.

After derotation, the reference system of one device can be (locally and independently) aligned with the other, and thus all three dimensions can consequently be compared for more accurate classification.

Without derotating time series, angular information cannot be used for deciding if devices have been moved together. To the best of the authors ' knowledge, this is the first analytic approach to find the optimal derotation of the coordinate systems, given only the two 3D time acceleration series of devices (supposedly) moved together. It can be used as the basis for further research on improved classification toward acceleration-based device pairing.