AcSIS: Authentication System Based on Image Splicing

Abstract

Text-based passwords are widely used for the authentication of digital assets. Typically, password security and usability is a trade-off, i.e. easy-to-remember passwords have higher usability that makes them vulnerable to brute-force and dictionary attacks. Complex passwords have stronger security but poor usability. In order to strengthen the security in conjunction with the improved usability, we hereby propose a novel graphical authentication system. This system is a picture-based password scheme which comprises of the method of image splicing. Authentication data were collected from 33 different users. The usability of the method was evaluated via a comparison between the number of correct and incorrect authentication attempts and time taken. Additionally, a comparison was made between our proposed method and a complex text-based password authentication method using the authentication success rate. Authentication using image splicing proved to be resilient to brute-force attacks since the processing of images consumes a voluminous password space. The evaluation of the usability revealed that graphical passwords were easy-to-remember, resulting in a higher number of correct attempts. The proposed method produced 50% higher success rate compared to the text-based method. Findings motivate the use of the proposed method for securing digital assets.