Security Analysis of Impostor Attempts with Respect to Gender in Gait Biometrics

This paper presents security analysis of impostor attacks with respect to gender information in gait biometric system. In particular, we investigate how different the impostor scores generated by matching gait samples from persons of the same gender are, compared to the impostor scores generated by matching gait samples from persons of different gender. Unlike most of the previous approaches, for collecting gait we use a motion recording sensor attached to the body of the person. Hip acceleration of the person is recorded by the sensor and used for authentication. During the experiment, we have collected gait sequences from 100 persons, 70 men and 30 women. Analysis based on false acceptance rates (FAR) indicate that there might be a significant difference between the same gender and different gender impostor attempts.