一种基于DTW的动态笔迹识别算法

A Dynamic Handwriting Recognition Algorithm Based on DTW

随着移动互联网技术的快速发展,智能手机、平板电脑等移动设备也得到了广泛使用。移动设备在网上购物、在线支付、转账等方面具有方便、快捷的特性,网上电子商务交易量越来越大,导致移动平台上电子商务信息的安全隐患越发突出,信息安全技术在移动平台上的应用越来越重要。基于生物特征的身份鉴别由于克服了传统身份鉴别方式的很多缺陷而得到了广泛的应用。手写笔迹是生物特征识别的重要领域之一,因具有采集设备简单、易采集、安全等优点,被广大用户所接受。针对在移动平台上识别手写笔迹的问题,提出一种基于DTW的动态笔迹识别算法。算法结合了FastDTW算法中限定路径弯折斜率,保证了识别效率;并通过放松时间弯曲时对起始端点的对齐,提高了识别的准确率。实验结果表明,算法的错误拒绝率(FRR)和错误接纳率(FAR)分别为4.5%和1.5%,识别效率和识别准确率较于经典DTW算法和FastDTW算法有所提高。

With the rapid development of mobile internet technology,smart phones,tablet PCs and other mobile devices have also been widely used.Mobile devices in the online shopping,online payment,transfer and other aspects of a convenient and fast features,online e-commerce trading volume is growing,leading to mobile platform e-commerce information security risks become more prominent,information security technology on the mobile platform Application is becoming more and more important.Biometrics-based identity authentication has been widely used to overcome many of the shortcomings of traditional authentication methods.Handwriting handwriting is one of the important areas of biometric identification,because of the collection device is simple,easy to collect,security and other advantages,by the majority of users accepted.A dynamic handwriting recognition algorithm based on DTW is proposed to identify handwriting handwriting on mobile platform.The algorithm combines the slope of the path in the Fast DTW algorithm to ensure the recognition efficiency,and improves the recognition accuracy by aligning the starting end points when the relaxation time is bent.The experimental results show that the error rejection rate(FRR)and false acceptance rate(FAR)of the algorithm are 4.5%and 1.5%respectively,and the recognition efficiency and recognition accuracy are improved compared with the classical DTW algorithm and FastDTW algorithm.