指纹图像快速细化的改进算法及其应用  

Improved fast thinning algorithm for fingerprint image and its application

作　　者：王晶[1] 李景萃[1] 

机构地区：[1]北京化工大学信息学院,北京100029

出　　处：《仪器仪表学报》2008年第7期1535-1539,共5页Chinese Journal of Scientific Instrument

基　　金：国家自然科学基金(60704011);北京市教育委员会共建项目建设计划(XK100100435)资助项目

摘　　要：针对现有指纹细化算法存在的细化不彻底、纹线吞噬、骨架偏离纹线中心等问题,提出了改进的快速细化算法,通过分析目标点周围纹线的走向趋势,选择去除或者保留周围的相连点,有效解决传统迭代算法细化不彻底的问题,并对细化图像采用方形模板进行纹线跟踪,克服了逐步递进的纹线跟踪算法过于复杂、不易实现等问题。最后将改进算法在以FPS200指纹传感器作为采集设备的自动指纹识别系统中进行验证。结果表明,该算法具有很好的细化效果,得到的细化图像满足进一步进行特征提取的要求。There is much trouble in thinning at present, such as not thinning to one pixel, ridge swallowing, thinning skeleton not in ridge center and so on. In order to overcome these disadvantages, an improved thinning algorithm is presented. By analyzing the trend of the ridge around the objectives, the algorithm selects removing or retaining the joint objectives around. This method can effectively solve the less thorough thinning problem in traditional iterative thinning algorithm. Then, the ridge tracking of thinned template to remove the fake features, and avoids the problem that fingerprint image is proposed, which uses a square gradually progressive ridge tracking algorithm is too complex and difficuh to achieve. Finally, this improved algorithm and the other parts of fingerprint image preprocessing are combined, and the whole algorithm is verified in the automatic fingerprint identification system （AFIS） based on fingerprint sensor FPS200. Results show that this algorithm has a good thinning effect and the thinned image meets the requirement of further feature extraction.

关 键 词：自动指纹识别系统 彻底细化 纹线跟踪 方形模版 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]