We present a novel biometric solution which exploits hand gestures, tracked by the Microsoft Kinect sensor, performed in response to a circle randomly appearing in five predefined screen positions. Features of both hand and screen pointer are used for classification purposes, considering both the whole 20-path trajectory and shorter routes. In particular, we search for the “optimal” trajectory length which assures a good trade-off between precision and user effort. For identification, the approach achieves classification accuracies ranging from 0.748 to 0.942. For verification, accuracy is still satisfactory (always higher than 0.962), despite moderate specificity values.
Soft biometrics through hand gestures driven by visual stimuli
Porta Marco
2019-01-01
Abstract
We present a novel biometric solution which exploits hand gestures, tracked by the Microsoft Kinect sensor, performed in response to a circle randomly appearing in five predefined screen positions. Features of both hand and screen pointer are used for classification purposes, considering both the whole 20-path trajectory and shorter routes. In particular, we search for the “optimal” trajectory length which assures a good trade-off between precision and user effort. For identification, the approach achieves classification accuracies ranging from 0.748 to 0.942. For verification, accuracy is still satisfactory (always higher than 0.962), despite moderate specificity values.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
