基于非平稳信号频谱分析的指纹图像增强算法  

Fingerprint Enhancement Algorithm Based on Non-stationary Signal Frequency Spectrum 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：《数据采集与处理》2008年第1期35-39,共5页Journal of Data Acquisition and Processing

摘　　要：把指纹图像看作一种局部平稳总体非平稳的信号,研究了指纹图像的频谱特性,提出了一种基于统计学的指纹方向和频率估计方法。在此基础上,基于方向一致性分析和频率分析,提出了方向和频率校正算法。实验结果表明,该算法对指纹图像有显著的增强效果,并有效地提高了指纹细节特征提取的准确率。Taking fingerprint image as a signal, the signal is locally stationary and non-stationary as a whole. This paper studies its frequency spectrum property and puts forward a probabilistic approximation method for the ridge orientation and frequency. Furthermore, the acquired ridge orientation and the frequency are revised based on the orientation coherence and the frequency analysis. Experimental results show that the method has a prominent effect on fingerprint images, thus the accuracy of the minutiae extraction is improved.

关 键 词：指纹识别 指纹图像增强 频谱分析 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]