Team 4. Naif Alotaibi, Rich Barilla, Francisco Betances, Aditya Chohan, Alexandra Garcia, Alexander Gazarov, Mantie Reid, and Vinnie Monaco. Biometric System Design for Handheld Devices . Outline. Introduction Related Work Background Methodology Data Collection. Introduction. - PowerPoint PPT Presentation
Biometric System Design for Handheld Devices Team 4 Naif Alotaibi, Rich Barilla, Francisco Betances, Aditya Chohan, Alexandra Garcia, Alexander Gazarov,Mantie Reid, and Vinnie MonacoOutlineIntroductionRelated WorkBackgroundMethodologyData Collection
IntroductionRelated WorkResearchers at University of Hong KongSingle-Touch, Multi-Touch, DragNeural network, error rate of 7.8%2012 ACM SIGCHI Conference on Human Factors in Computing Systems90% accuracy (single gestures)Improvement using multiple gestures (sequence)
Researchers at the University of HoustonGraphic Touch Gesture Feature (GTGF)Flick up/down, flick right/left, zoom in/out Converting touch traces to imagesLow Equal Error Rate of 2.62%Mobile device picking up motionTrajectory and angleError rate of 6.13%Accuracy declines with user movement
BackgroundTouchscreenX and Y coordinates of the touch (position) Pressure of the touchSize of the contact areaState change (Ex. Up, Down, etc) Multi-touch screens report multiple movement traces at the same time using pointers.
Motion sensors: measure acceleration and rotational forces. Position sensors: used for capturing data about the physical position of the device Useful sensors for developing our system: Accelerometer: measure the acceleration applied to the device. Gyroscope: measures the rotation around the device's axis. Orientation: measure the position relative to the earth's frame of referenceRaw data captureSystem architectureMain activity (with WebView)Biometric eventEvent bufferSQLite databaseData fileNetworkSettings activityData Collection
Results98% accuracy within single session25% with one session as test and the other as training data