摘要
通过对多种指纹分类算法的研究和分析,提出了一种基于BP神经网络对指纹模板进行分类的新算法.首先在对指纹图象进行预处理后建立起指纹模板库,然后采用时间模拟退火函数进行学习因子修正的方法来减少BP算法迭代次数,以提高收敛速度及跳出局部最小.仿真证明:该算法与传统的指纹识别算法相比,分类速度明显高于传统的固定步长的BP算法.
By studying and analyzing many algorithms for fingerprint recognition,this paper proposes a new algorithm for fingerprint recognition based on BP neural network.First, we simply introduce the process of image precodition and fingerprint template store. Second, the simulated annealing algorithm is used to adjust the learning step and get a global optimum. The simulation result indicates that the new algorithm is more superior in convergent rate and precision than that of other traditional algorithms.
出处
《邵阳学院学报（自然科学版）》
2007年第1期54-57,共4页
Journal of Shaoyang University：Natural Science Edition
关键词
指纹识别
BP算法
神经网络
fingerprint recognition
BP algorithm
neural network
