基于多种特征融合的指纹识别方法

Fingerprint Recognition Method Based on Multi-feature Fusion

针对基于细节点特征的指纹识别方法和基于图像特征的指纹识别方法各自存在的一些问题,提出了一种基于多种特征融合的指纹识别方法。将基于方向场特征、基于灰度共生矩阵的纹理特征、基于LBP算子的纹理特征的方法和基于细节点特征的方法进行融合,弥补了各个方法的不足,提高了匹配的准确性。实验结果表明,提出的基于多种特征融合的指纹识别算法好于任一单个的指纹匹配方法,有效地提高了指纹识别系统的准确性。

To overcome the disadvantages and better the performance of minutiae-based and image-based fingerprint identification methods,a fingerprint recognition algorithm, consisting of the processes of classification selection and classifier fusion,was put forward. Three image-based methods,based on orientation field, texture features of gray level cooccurrence Matrix, histogram features of local binary pattern operator, were combined with minutiae-based method for fingerprint recognition. Experimental results indicate that the performance of fusion-based method is better than any method used alone in fusion and can be applied to automatic fingerprint identification system.