PassBYOP: Bring Your Own Picture for Securing Graphical Passwords

Abstract

This research explores PassBYOP, a novel graphical password authentication system that leverages users' personally selected images to enhance security and usability. The system addresses the weaknesses of traditional graphical passwords by incorporating image-based selection and manipulation, increasing the complexity and resistance to attacks like shoulder surfing and screen capture. Our methodology incorporates image processing techniques and a robust verification algorithm to ensure both security and a user-friendly experience. Results demonstrate a significant improvement in security compared to existing graphical password schemes while maintaining acceptable usability.

Introduction

Graphical passwords offer a potentially more memorable and user-friendly alternative to traditional alphanumeric passwords. However, current graphical password schemes suffer from vulnerabilities such as simple patterns and susceptibility to attacks. Users often choose easily guessable patterns, reducing the effectiveness of the authentication method. This research addresses this gap by introducing PassBYOP, which allows users to select and manipulate their own images, creating highly personalized and complex graphical passwords resistant to common attacks. The increased complexity and personalization aim to improve security while maintaining usability.

Objectives

• Develop a secure and user-friendly graphical password authentication system using personalized images (PassBYOP).
• Implement robust image processing and comparison techniques to ensure accurate authentication.
• Evaluate the security and usability of PassBYOP compared to existing graphical password schemes.

Technical Details

• Android Studio with Java/Kotlin or Python (Kivy)
• Image selection and cropping functionalities
• Image segmentation and grid-based password creation
• Secure comparison algorithm using image feature vectors or hashed coordinates
• Attack resistance (shoulder surfing, screen capture)