Biometrie-Evaluations-Zentrum (BEZ)

BEZ_keyvisual_hellblauer-Hintergrund_1920x1080_Header
BEZ (ISF)

Development of new biometric security technologies at the BEZ

3D-Kamera_Biometrie-Evaluations-Zentrum (BEZ)
Biometric modalities are becoming more and more common for security applications, for example in border control systems. In order to ensure the security of these biometric systems, continuous improvements and Presentation Attack Detection (PAD) methods are required. In the following, two approaches are presented, which are investigated at the BEZ.
BEZ_Biolab_LokiMk2_Materialbeispiele_englisch

NIR-PAD: Skin detection for multiple modalities


Biometric features are often presented on skin. Thus, the reliable detection of skin is desirable in order to detect fakes(e.g. masks) to prevent spoofing of the biometric systems. By measuring the optical properties, skin can be verified reliably due to its specific remission spectrum. In the near infra-red waveband, the used features are independent of skin type. 

In order to generate ground-truth data, the remissions spectrum of probands is measured using two spectrometers measuring the spectrum in the near infra-red and the visible waveband. New biometric systems could use the data as a reference for automated skin detection. In addition, the performance of existing systems could be evaluated with this dataset as a standard reference.

Moreover, the reflection depends on the incident and detection angle of light. With a specific designed device, the reflection is measured on varying incident angles and probe orientations using a monochrome camera and several RBGW LEDs. With this data, skin-detection systems could be
improved and the skin of synthetic characters could be refined to look more realistic.

 

BEZ-BIOLAB-Fälschungserkennung mit 3D-Kamera

3D-Face presentation attack detection


As the face is one of the most important biometric traits, 3Ddata about the presented face enables new developments in mask detection or person authentication. One promising technology in this regard is the 3D-time-of-flight technology. A 3D-ToF-camera illuminates the scene with a high-speed
modulated light source. A ToF-specific image sensor is able to measure the time of flight on pixel level and the distance is calculated from this data in a postprocessing step. With this data, a high-resolution 3D shape of the face can be generated. 

This data can be used to detect artefacts (e.g. mobile devices showing a photo) which are used to spoof the biometric system. Moreover, the alignment of other biometric sensors or guidance of the person using the biometric system is
possible.