BioIdent - Touchstroke based biometrics on Android platform

Data collection

• 71 subjects (56 male, 15 female)
• 8 different Android devices (resolutions: 320x480 ..1080x1205))
• personal information: gender, birth date, touch screen experience level (0,1,2,3)
• vertical strokes: text comprehension
• horizontal strokes: image gallery - choosing the favourite image
• multiple sessions

Raw Data

Feature extraction

• strokeDuration: the time needed for a stroke expressed in milliseconds;
• startX: the x coordinate of the stroke starting point;
• startY: the y coordinate of the stroke starting point;
• stopX: the x coordinate of the stroke ending point;
• stopY: the y coordinate of the stroke ending point;
• directEndToEndDistance: the length of the segment defined by the two endpoints;
• meanResultantLength: a feature characterizing the straightness of the stroke;
• upDownLeftRightFlag: orientation of the stroke; a stroke is classified horizontal (left, right) if its horizontal displacement exceeds its vertical displacement;
• directionOfEndToEndLine: the slope of the segment defined by the two endpoints;
• largestDeviationFromEndToEndLine: the maximum of the distances between points belonging to the stroke and the segment defined by the two endpoints;
• averageDirection: the average slope of the segments belonging to the stroke trajectory;
• lengthOfTrajectory: the length of the stroke;
• averageVelocity: the average velocity of the stroke;
• midStrokePressure: the pressure at the midpoint of the stroke;
• midStrokeArea: the area covered by finger at the midpoint of the stroke.

Datasets (WEKA ARFF)

• dataset1: User identity classification (71 users, ~200 strokes/user)
• dataset2: User identity classification (51 users, 100 strokes/user)
• dataset3: Gender classification (18 users, 100 strokes/user)
• dataset4: Touchscreen experience level classification (24 users, 100 strokes/user)

Android app for data collection

Publications

• Margit ANTAL, Zsolt BOKOR, László Zsolt SZABÓ (2015), Information revealed from scrolling interactions on mobile devices, Pattern Recognition Letters, 56, pp. 7-13. [Link]
• Margit ANTAL, László Zsolt SZABÓ, Zsolt BOKOR (2014), Identity information revealed from mobile touch gestures, 10th Joint Conference on Mathematics and Computer Science, Cluj-Napoca, May 21-25, 2014. Studia Informatica, vol. LIX. Special Issue 1, 5-14. [PDF] [Presentation]