一种动态阈值加填补的指纹图像二值化算法  

Dynamic Threshold with Hole Padding Algorithm for Fingerprint Image Binarization

作　　者：计春雷[1] 冯伟[2] 黎明[1] 杨杰[2] 

机构地区：[1]上海电机学院电子信息学院,上海200240 [2]上海交通大学图像处理与模式识别研究所,上海200240

出　　处：《计算机仿真》2011年第7期258-261,289,共5页Computer Simulation

基　　金：国家863计划项目(2007AA01Z164)

摘　　要：针对整幅指纹图像进行判断和填补,消除可能存在的孔洞,采用指纹图像二值化是指纹识别的重要环节,通过直接识别指纹可以提高效果。传统指纹图像二值化方法处理速度慢、不适用于低对比度指纹图像,为了提高指纹图像二值化的效率和鲁棒性,提出一种动态阈值加填补的指纹图像二值化算法。将指纹图像分割为互不相交的若干个小窗口,在每个窗口中求得所辖像素的灰度平均值,经添加惩罚项后作为初步阈值进行二值化处理。利用MATLAB软件平台进行了仿真,将方法与传统指纹图像二值化方法进行了比较。实验证明,动态阈值加填补的二值化算法处理效率高、鲁棒性好。Image binarization is essential in fingerprint recognition,and it has direct impact on the success of fingerprint identification.Traditional image binarization algorithms either process too slowly to keep up with the actual requirements,or cannot be applied to low contrast image regions.In order to improve the efficiency and robustness of image binarization algorithms,a dynamic threshold with hole padding algorithm is proposed.Firstly,a fingerprint image is segmented into several small windows,within which the average gray value is calculated.Then a penalty term is added to the average gray value to serve as the threshold.Finally,the whole fingerprint image is analyzed to find the possible holes and a hole padding algorithm is then applied to eliminate the possible holes.Simulation experiments are performed on MATLAB platform,and comparisons are carried out with the proposed algorithm and traditional algorithms.Experimental results show that the proposed algorithm is more robust and has better binarization performance.

关 键 词：指纹 二值化 阈值 图像分割 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]