基于智能眼镜触控行为的隐式身份认证方案  

Implicit authentication based on smart glass touch control behavior

作　　者：黄海平[1,2] 徐宁[1,2] 朱毅凯 王汝传 

机构地区：[1]南京邮电大学计算机学院,江苏南京210023 [2]江苏省无线传感网高技术研究重点实验室,江苏南京210003 [3]南京大学网络信息中心,江苏南京210023

出　　处：《南京邮电大学学报（自然科学版）》2017年第6期55-60,共6页Journal of Nanjing University of Posts and Telecommunications：Natural Science Edition

基　　金：国家自然科学基金(61373138;61672297);中国博士后基金(2015M570468;2016T90485);江苏省重点研发计划(BE2017742);江苏省六大人才高峰项目(DZXX-017)资助项目

摘　　要：随着可穿戴技术的发展,存储在智能眼镜上的各种数据给用户带来了隐私安全的挑战。文中针对智能眼镜的安全认证问题,使用Leap Motion体感控制器代替触控屏作为其输入设备,并基于此提出了一种基于触控行为的隐式身份认证方案。该方案首先通过Leap Motion设备获取用户特征数据,其次采用Bagging集成算法对特征数据进行学习训练,将BP神经网络算法和K近邻算法作为基学习器,进而在用户正常使用智能眼镜的过程中实现手势识别的隐式身份认证。文中具体以"触控中的向前滑动手势"为例,总共选取了12个特征,并提取613项数据组成了训练集。最后,通过实测的方式测得该方案在向前滑动手势下的准确度为94.117 6%。With the development of wearable technologies, private data stored on smart glasses bring security risks to the owner. Aiming at the issue of the security authentication on smart glasses, the Leap Motion sense controller is adopted instead of touch screen as its input device. An implicit authentication scheme based on touch control is proposed. Firstly, the scheme acquires user characteristic data through Leap Motion device. Secondly, Bagging integration algorithm is used to train the feature data, and the BP neural network algorithm and the K nearest neighbor algorithm are used as the algorithms of base learners. Then, the implicit identity authentication based on gesture recognition is realized in the process of normal use of smart glasses. It chooses the forward swipe gesture with selected twelve features as the example, and extracts 613 data to form a training set. Finally,the accuracy of the example regarding the forward swipe gesture is achieved by 94.117 6%.

关 键 词：智能眼镜 Leap Motion体感控制器 行为特征 隐式身份认证 BAGGING算法 

分 类 号：TP393.08[自动化与计算机技术—计算机应用技术]