摘要
为了快速提取不同频率空间和纹理方向上的指纹特征,本文采用了快速的5/3提升小波算法。该算法摆脱了传统的滤波器和傅立叶的频域概念,可以更直接地利用时域信号,具有运算复杂度低、实时性好的特点。将该算法与传统的基于细节特征识别的算法进行了比较。实验表明,该提升小波识别系统具有较高的识别率,并且具有匹配速度快、运算量小等特点。
In this paper,we use a new fast 5/3 lifting wavelet algorithm, which can extract different spatial frequencies and texture orientations in the various regions of a fingerprint. This algorithm gets rid of the conventional filter and the Fourier frequency domain. It can use the time domain signals directly, which have less complexity and better real-timeness. We compare this new algorithm with the conventional algorithm which is based on minutiae, and realize it based on TMS320DM642 DSP. The experimental results show that this lifting wavelet identification system have some good charac-teristics , such as high accuracy, fast matching, less computation, and so on.
出处
《计算机工程与科学》
CSCD
北大核心
2009年第2期46-49,共4页
Computer Engineering & Science
基金
福建省自然科学基金资助项目(A0610022)
国务院侨办科研基金资助项目(06QZR03)
关键词
指纹识别
DSP
提升小波
fingerprint recognition
DSP
lifting wavelet
