中文摘要：

为了对眉毛这种新颖的生物特征开展识别研究，提出了一种基于特征串比较的眉毛识别方法，其基本思想是采用离散傅里叶变换和K-均值算法进行特征串提取，并通过计算2个眉毛特征串之间的编辑距离来确定所识别的候选人．在22人的小规模眉毛图像数据库上所做的6组实验表明，该方法均达到了95．45％或100％的识别正确率，从而验证了眉毛识别用于个人身份鉴别的可能性和有效性．

英文摘要：

In order to study the novel biometric of eyebrow, this paper presents an eyebrow recognition method by comparison of feature strings, the basic idea of which is to extract feature strings using discrete Fourier transformation and K-means algorithm, and to recognize a given eyebrow image as the candidate person with the minimum edit distance between their feature strings. It has been shown that the method can reach an accuracy of 95.45 % or 100.00% in six experiments on an small-scale eyebrow database taken from 22 persons. Therefore, eyebrow recognition may possibly apply to personal identification, which can be valid.