基于Haar小波变换的快速指纹识别算法

Fast Algorithm for Fingerprint Identification Based on Haar-wavelet Transform

指纹中不同的空间频率和纹理方向代表了指纹图像的内在特征。为了提取这些特征,提出了基于Haar小波变换的快速指纹识别算法。该算法在提取指纹特征有效区域的基础上,直接对指纹灰度图像进行Haar小波变换以获取指纹特征向量。与传统的基于指纹细节特征的识别算法相比大大减少了计算量。另外,还将指纹识别算法移植到嵌入式安全领域。在基于Arm7处理器(主频100MHz)架构的嵌入式平台上,对FVC2000标准指纹数据库进行了实验,得到了很好的FAR,FRR关系图,并且整个识别过程耗时在1 s左右。实验结果表明,本文算法在快速指纹识别方面是成功的。

Ridge orientations and ridge spatial frequencies in various regions of fingerprints represent its intrinsic characteristics. In this paper, we proposed a new fast algorithm for fingerprint identification based on Haar-wavelet, which accurately extracted the fingerprint features based on the original gray scale image. The proposed algorithm requires less computational effort than conventional algorithms which are based upon minutia features extraction. The algorithm can achieve high recognition rates when tested in the standard FVC2000 fingerprint database, and complete a whole process of identification in about one second on an embedded platform based on Arm-7 （Primary Frequency： 100MHz） CPU.