|Contact: Prof. Andrzej Pacut|
Objectives: Mobile devices are typically used in unsupervised environment. Simultaneously, we can observe the trend of utilising these devices in sensitive applications such as retail and mobile payment. This calls for enhanced security techniques that should be implemented to make mobile systems resistant to presentation attacks. The limited capabilities of mobile devices often require the simplification of data acquisition and processing methodologies. This contradiction leads to devices that are characterised by a lack of appropriate countermeasures against subterfuges, especially presentation of artefacts. This three year ESR research project aims to develop a set of countermeasures dedicated to mobile biometric-based systems enabling multimodal recognition. The following biometric modalities will be considered (yet are not limited to): iris, face, fingerprints and handwritten signatures. The presentation attack detection (PAD) methods, contrary to typical liveness detection existing today, should incorporate modalities simultaneously, as well as considering static and dynamic features accounting for authenticity of the biometric sample. Different types of PAD-related information should be then concatenated to obtain a final decision. The proposed methodologies will be evaluated in practice since this project assumes implementation on commercially available equipment.
Expected Results: 1) Presentation attack detection methods for single modalities suitable for mobile platforms. 2) Statistical methodology defined and applied to PAD-related information fusion, as well as performance assessment of single and multiple (fused) sources of PAD information. 3) A working prototype demonstrating capabilities of the multimodal presentation attack detection on an example mobile device.
Planned secondment(s): Two secondments would take place within this project. The first is a five month visit to NASK wherein the ESR will gain a practical knowledge on implementation of presentation attack detection techniques and methods for evaluation of biometric security. The expected results of this secondment are the set of programming tools and test plans needed for evaluation of presentation attack detection implementations. In particular, the test plans will be prepared according to the actual ISO/IEC standards devoted to PAD and biometric evaluation. The second five month secondment would be with UNIROMA wherein the ESR would leverage their skills related to fusion of biometric countermeasure information in multi-biometric mobile systems. The expected results of this secondment are the prototype methods of multi-sensor data fusion for presentation attack detection in mobile devices.
Start Date: July 2017