基于模糊均值聚类的自适应指纹图像分割  

Self-adaptive segmentation algorithm of fingerprint image based on FCM

作　　者：项明[1] 吴小培[1] 汤婷[1] 孙林森[1] 

机构地区：[1]安徽大学计算智能与信号处理教育部重点实验室,合肥230039

出　　处：《电子测量技术》2009年第5期77-81,共5页Electronic Measurement Technology

基　　金：国家自然科学基金资助项目(60375010)

摘　　要：指纹分割是指纹图像预处理中的重要一步,有效的分割能减少后续处理时间并提高特征提取的可靠性。提出一种基于FCM的指纹图像分割方法。利用指纹图像脊线和谷线的灰度方差、纹线的方向一致性进行指纹图像分级分割,由FCM算法自适应的选择分割阈值。避免了人为选择阈值的困难。实验结果表明,该方法具有较高的稳健性和有效性。Segmentation is an important step of fingerprint image preprocessing. Effective segmentation can reduce the time of subsequent processing and improve the reliability of feature extraction considerably. An algorithm based on FCM for segmentation of fingerprint image is proposed. Gray variance and direction coherence are used as features for segmentation and the threshold is selected self-adaptive, avoided the difficulty of artificial choice. Experimental results verify the feasibility and robustness of this algorithm.

关 键 词：模糊C均值 分割 阈值 指纹 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]