基于图像质量分区的指纹特征提取  

Fingerprint Feature Extraction from the Subareas Divided by Image Quality

作　　者：胡春风[1] 殷建平[1] 祝恩[1] 陈晖[1] 

机构地区：[1]国防科技大学计算机学院,湖南长沙410073

出　　处：《计算机工程与科学》2009年第8期4-6,48,共4页Computer Engineering & Science

基　　金：国家自然科学基金资助项目(60603015;60373023);高等学校全国优秀博士学位论文作者资助项目(2007B4);湖南省教育厅资助科研项目(湖南省优秀博士学位论文获得者资助项目)

摘　　要：低质量指纹图像的特征提取和变形指纹的匹配是当前指纹识别研究中的两个主要问题。很多算法在特征提取时不区分高、低质量区域,结果在高质量区域耗费了过多的运算时间和计算资源。本文提出了一种基于图像质量分区的指纹特征提取方法,先用一种简单的图像区域质量计算方法评价各区域的图像质量,然后对高质量区域直接从灰度图像跟踪纹线、提取节点,对低质量区域执行传统的方向计算、增强、二值化和细化后提取特征。实验结果表明,该方法不仅提高了特征提取的速度,在准确性上也有所提高。How to extract features from low-quality images and match badly deformed fingerprints are two main problems in fingerprint recognition. Most algorithms without any distinguishment between lowland high-quality subareas, use too much computation to extract the features in high-quality image subareas. A new method to extract fingerprint features from the subareas divided by image quality is proposed. The image quality of all subareas is estimated first. Ridges are traced and minutiae are extracted directly from the gray-level images for high-quality subareas. Orientation estimation, enhancement, binarization and thinning are executed for poor-quality subareas, and then minutiae are extracted. Experimental results show that the time of feature extraction is reduced and the accuracy is also improved.

关 键 词：指纹识别 方向计算 纹线跟踪 特征提取 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]