基于判别因子的指纹图像质量评估算法  

Fingerprint image quality evaluation based on discriminant factors

作　　者：郭新星[1] 彭小奇[1,2] 赵琪[1] 王健康[1] 

机构地区：[1]中南大学物理科学与技术学院,长沙410083 [2]湖南第一师范学院,长沙410205

出　　处：《计算机应用研究》2009年第11期4393-4394,4397,共3页Application Research of Computers

基　　金：国家自然科学基金资助项目(50374097);博士点专项科研基金资助项目(20030533008)

摘　　要：提出了一种新的指纹图像质量评估方法。该方法根据不同指纹特征对最终评估结果影响的大小,依次通过指纹图像有效面积比、位置偏移量、灰度均值、灰度方差、傅里叶频谱能量比等判别因子对指纹图像质量进行评估,并对低质量指纹提出改进建议。实验结果表明,该算法能够对不同质量的指纹图像进行准确评估,拒绝低质量指纹的输入,可显著提高自动指纹识别系统的准确率和识别效率。This paper proposed a new method to evaluate fingerprint image quality on the basis of discriminant factors such as effective area ratio, position offset, gray average, gray variance Fourier spectrum energy ratio. The order of discriminant factors was in accordance with the impact on the final evaluation results. Experimental results indicate that the method can accurately evaluate the different kinds of fingerprint images and also make improvement suggestions for user. Finally it rejects the low quality fingerprints input and improves the automated fingerprint identification system accuracy and efficiency.

关 键 词：指纹识别 图像质量评估 判别因子 图像预处理 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]