基于改进DTW算法的手写签名识别系统设计  

Design of Handwritten Signature Identification System Based on Improved DTW Algorithm

作　　者：王珂敏[1] 

机构地区：[1]武警工程学院通信工程系声像教研室,西安710086

出　　处：《自动化信息》2010年第7期50-52,71,共4页Automation Information

摘　　要：在线手写签名验证是一种基于生物特征的身份认证技术。为提高签名验证的效率，该文介绍了一种改进的在线签名识别算法。它优化了传统的动态时间弯折算法结构，提出了对最佳匹配路径的动态规划方法并将其应用于在线签名识别系统中。在模板较多时对匹配距离将适当限制，从而减少了系统运算量，提高了模板匹配速率。随着待识别模板数目的增多，该算法的效率优势更加明显。试验结果表明，该改进算法的运算效率高，误拒率和误纳率较低。On-line handwritten signature verification is a kind of biometric identity authentication technology. In this paper, an improved on-line signature identification algorithm is introduced to enhance the efficiency of signature verification; it optimizes the conventional DTW algorithm＇s structure, proposes the dynamic programming method for optimum matching path and applies the method to the on-line signature identification system. The matching distance will be restricted properly if the number of templates is more, thus reduces the computation workload of the system and increases the template matching rate. With the increment of templates which waiting for identification, the efficiency advantage of the new algorithm is more obvious. Experimental results show that the improved algorithm is of high efficiency in computation and lower false rejection rate （FRR） as well as false acceptance rate （FAR）.

关 键 词：签名认证 动态时间弯折 动态规划 匹配距离 模板匹配 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]