Multimodal Biometric Surveillance using a Kinect Sensor |
Surveillance technologies are becoming increasingly important in the fight against crime and terrorism. More recently, technologies such as biometrics are being applied to provide an automated approach to the verification of individuals – removing the need for human operators. Unfortunately, current approaches suffer from a number of issues such as expensive biometric capture devices and reliance upon uni-modal systems. This paper presents a feasibility study into the use of an off-the-shelf commercial device (Microsoft Kinect) as a multimodal biometric sensor. An experimental study is undertaken, combining two biometrics modalities (face and gait recognition) and a soft biometric (skeleton measurements). Based upon 20 participants, the study shows an Equal Error Rate (EER) of 12.3% can be achieved – outperforming the results the single uni-modal approaches were able to obtain.
Savage R, Clarke NL, Li F