Abstract: It is difficult to get very high accuracy in fingerprint verification only by using solo method，such as minutiae based method，ridge based method，etc.The fusion of different methods is an important way to get higher accuracy compared to just using solo method.This paper proposes a novel serial fusion method.Firstly，minutiae based method is used to prejudge the input fingerprint.If it is able to determine the two fingerprints are impostor or genuine，this verification ends.Otherwise，ridge based method is used to match the two fingerprints.Then，the two scores of minutiae based method and ridge based method are integrated by fusion strategy.Now，it is able to determine the two fingerprints are impostor or genuine.Experimental results on FVC2002 DB2 indicate that，the Equal Error Rates（EERs） of minutiae based method，ridge based method，Sum fusion，serial fusion are3.0%，4.9%，2.0%，1.9% respectively.Also，the time consuming reduces 64.64% compared to Sum fusion.

Key words: fingerprint verification, serial fusion, minutiae, ridge, Sum fusion

摘要： 在指纹识别中，利用细节点或纹线等单一指纹特征的识别算法来得到一个很高的识别性能比较困难。将不同的指纹匹配算法进行融合来获得较高的准确率已经成为当前研究的热点。提出了一种新的序列化融合的指纹匹配方法。利用基于细节点的指纹匹配算法对待识别指纹进行预判，对于不能确定其为同源或异源的，将再利用基于纹线的指纹匹配算法进行匹配，对两次的匹配得分进行融合，基于融合结果判断其为同源或异源。在指纹库FVC2002 DB2上的实验结果表明，采用基于细节点的指纹匹配算法、基于纹线的指纹匹配算法、对上述两种算法的Sum融合方法、提出的方法得到的等错误率（EERs）分别为3.0%、4.9%、2.0%、1.9%，且相对于Sum融合方法，提出的方法在时间消耗上降低了64.64%。

关键词: 指纹识别, 序列化融合, 细节点, 纹线, Sum融合