iSignDB: A database for smartphone signature biometrics

Area of Science:

• Biometrics
• Human-Computer Interaction
• Data Science

Background:

• Signature verification is crucial for user authentication.
• Online signature verification captures dynamic features beyond static images.
• Existing smartphone-based datasets lack comprehensive sensor data.

Purpose of the Study:

• To introduce iSignDB, a novel biometric signature dataset collected using smartphones.
• To address the scarcity of publicly available online signature databases.
• To facilitate the development of robust smartphone-based biometric authentication systems.

Main Methods:

• Collected genuine and skilled forgery signature samples using iOS and Android smartphones (iPhone 7, Redmi Note 7).
• Utilized gyroscope, magnetometer, GPS, and accelerometer sensors to capture dynamic data.
• Recorded angular velocity, acceleration, orientation, geomagnetic field, and position data via MATLAB Mobile App.
• Ensured consistent feature sets by using smartphones with at least these four sensors.

Main Results:

• Developed the iSignDB dataset comprising genuine and skilled forgery signature samples.
• Each sample includes signature images and multi-sensor readings (acceleration, angular velocity, magnetic field, orientation, position).
• The dataset is suitable for designing authentication systems robust against spoof attacks.

Conclusions:

• iSignDB provides a valuable resource for research in smartphone-based biometric signature verification.
• The dataset can enhance the development of secure and reliable user authentication methods.
• Potential applications include behavioral analysis of users based on smartphone interaction patterns.