Towards Constructing Consistent Pattern Strength Meters with User’s Visual Perception

Abstract

Pattern lock strength meters designed for securing Android devices are inconsistent in their metering, e.g., assigning higher scores to weaker patterns. In this paper, we raise this inconsistency problem by analyzing five existing pattern strength meters. We reveal that they commonly miss some important visual features and even assign erroneous weights to features. As a preliminary study toward a consistent pattern strength meter in the future, we design a rigorous user study to identify the visual features of a pattern that correspond to real-world users’ criteria to score the strength of the pattern. We conducted an online survey for 3,851 users to collect reliable labels for 625 patterns. The statistical result of the user study sheds light on a pattern strength meter that reflects the user’s visual perception with various visual features.