Human Biometric Sensor Interaction Model v3.0 – Using the Kinect

The BSPA Lab is currently researching how people interact with biometric sensors over time. This study is a multi-sensor, multi-modal, multi-visit data collection involving fingerprint, hand geometry, face, dynamic signature verification, palm vein, and iris modalities.

As part of this data collection, researchers in the lab are using the Microsoft® Kinect^(TM) technology to collect data that enables them to examine how posture affects the performance of various modalities, and subsequently how it relates to the HBSI model. The Kinect^(TM) research is based on methodology developed by undergraduate students, graduate students, and faculty at Purdue University. The Kinect^(TM) sensors are used to gather X, Y, and Z coordinates of a subject’s body position relative to the sensor he/she is interacting with. Using these coordinates gathered on 20 different points of the body (head, neck, right shoulder, left shoulder, etc.), researchers are able to gather quantifiable data on the subjects position in relation to the biometric sensor to better understand how the subject’s body position affects an interaction.

The analysis and refinement of the Kinect^(TM) technology’s use in the data collection and the HBSI model will be updated on this webpage. Researchers also plan on highlighting this research in several teleconferences throughout the 2012-2013 academic year.

Craig Hebda, Rob Pingry, Weng Kwong Chan, Brent Shuler, Daniel Obot, Michael Texeira, Jay Peters, Kevin O’Connor, Jacob Hasslegren, and Stephen Elliott presented their poster on the use of Microsoft Kinect for collection of the Human Biometric Sensor Interaction data.