Improvement of Shoulder-Surfing Resistant Authentication Method Using Hybrid Images

A hybrid image blends the high-frequency component of one image with the low-frequency component of a different image. The high-frequency component is easily recognized from a short distance, but difficult to recognize from a long distance. On the other hand, the low-frequency component is easy to recognize from a long distance; however, it is difficult to recognize from a short distance. IllusionPIN, an authentication method based on hybrid images, was proposed in 2017. In IllusionPIN, a user and an observer recognize different digits on the same smartphone display. For example, when the user touches the digit 3 derived from high-frequency components, the observer recognizes 7 derived from low-frequency components. Although IllusionPIN is known to resist shoulder surfing, the authentication success rates and operation times of this method have not been reported. The present evaluation of IllusionPIN revealed a major problem: overlap of the digits derived from the high- and low-frequency components degrades the distinguishing ability, thereby reducing the success authentication rate to unacceptably low levels (approximately 77%). To improve the success authentication rate of IllusionPin, we slightly shift the drawing range of one digit. Evaluation results confirmed that after this improvement, the authentication success rate exceeded 90% without disturbing the shoulder-surfing resistance.