A Biometric Based User Authentication Scheme to Safeguard Smart Phones from Impostors

Smart phones occupy a prominent position in digital gadgets. The general population uses smart phones for communication, storing the personal information, surfing the internet and performing the bank transactions. The enhanced functionality, intuitive interfaces and the stored sensitive personal information makes smart phones vulnerable to a number of threats. Safeguarding the device and the stored information requires robust security protocols for legitimate user authentication. The traditional cryptographic methods (passwords, tokens, PINs and visual pass patterns) do not guarantee an optimum protection to safeguard the gadget and the stored personal information. In contrast, biometric based authentication provides a reliable way to authenticate the lawful users of smart phones. In recent years, ocular biometric identifier is gaining popularity in the research community. Henceforth, this work proposes a machine learning based user authentication scheme for smart phones based on the ocular biometrics. The scheme uses Felzenszwalb's HOG descriptor (FHOG) for intricate feature extraction and support vector machine (SVM) for authentication. The preliminary results on VISOB dataset display promising results.