作者: Guixin Ye , Zhanyong Tang , Dingyi Fang , Xiaojiang Chen , Willy Wolff
DOI: 10.1145/3230740
关键词:
摘要: Pattern lock is widely used for identification and authentication on Android devices. This article presents a novel video-based side channel attack that can reconstruct locking patterns from video footage filmed using smartphone. As departure previous attacks pattern lock, this new does not require the camera to capture any content displayed screen. Instead, it employs computer vision algorithm track fingertip movement trajectory infer pattern. Using geometry information extracted tracked motions, method accurately small number of (often one) candidate be tested by an attacker. We conduct extensive experiments evaluate our approach 120 unique collected 215 independent users. Experimental results show proposed over 95% in five attempts. discovered that, contrast most people’s belief, complex do offer stronger protection under attacking scenarios. demonstrated fact we are able break all but one (with 97.5% success rate) as opposed 60% simple first attempt.We demonstrate video-side serious concern only graphical also PIN-based passwords, algorithms analysis developed easily adapted target passwords. countermeasure, propose change way constructed used. proposal successfully defeat attack. hope encourage community revisit design practical use lock.