基于时频分析的指纹增强算法研究  

Research on Fingerprint Enhancement Algorithm by Time-Frequency Analysis

作　　者：彭章平[1] 彭小奇[1] 伍更新[1] 

机构地区：[1]中南大学物理科学与技术学院,长沙410083

出　　处：《计算机与数字工程》2007年第5期5-8,共4页Computer & Digital Engineering

摘　　要：指纹增强是自动指纹识别系统中的一个重要环节,良好的增强算法能够提高细节特征提取的准确率,从而提高整个指纹识别系统的鲁棒性。研究局部指纹图像的频谱特性,采用一种时频分析方法和一种新的基于概率统计学估计的方法求指纹的方向和频率。并提出一种鲁棒的方向和频率校正算法,构造一种带宽自适应的滤波器进行滤波。实验结果表明,该算法对指纹图像有显著增强效果,有助于指纹细节特征的提取。Fingerprint enhancement is a critical step in automatic fingerprint identification system, a good enhancement algorithm can improve the accuracy of minutiae extraction, which consequently improve the robustness of recognition system. The local fingerprint image frequency spectrum property is investigated,a time-frequency method and a new probabilistic approximation method for ridge orientation and frequency acquirement are adopted. A robust method of modifying ridge orientation and frequency estimation is proposed, an adaptive frequency filter is designed. Experimental results show that the method has prominent effect on fingerprint images, which is helpful to minutiae extraction.

关 键 词：指纹 指纹增强 时频分析 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]