指纹识别算法的研究与改进  

Study on the cost risk evaluation model of power transmission project based on bill-of-quantity model

作　　者：刘颖[1] LIU Ying(Shanxi Jiaotong Vocational and Technical College,Xi＇an Shanxi,710018)

机构地区：[1]陕西交通职业技术学院

出　　处：《自动化与仪器仪表》2018年第10期23-26,共4页Automation & Instrumentation

摘　　要：指纹识别是最常用的人体生物识别技术之一,其在图像处理和分类算法上进行传递。文中提出并测试了一个指纹识别算法框架,该框架使用一组图像预处理算法来实现指纹检测。关于特征提取,文中提出使用图像局部中的分叉数目,并使用人工神经网络（ANN）进行分类。最终,对指纹识别算法框架性能测试了三种不同的激活函数,其结果显示Sigmoid函数在较低的神经元数目下可以达到81%的准确度。Fingerprint recognition is one of the most commonly used human biometric recognition technologies for human body recognition.It passes on image processing and classification algorithms.In this paper,we propose and test a fingerprint recognition algorithm framework that uses a set of image preprocessing algorithms to implement fingerprint detection.With regard to feature extraction,we propose to use the number of bifurcations in the image locality,and we propose to use artificial neural networks（ ANN） for classification. In this paper,three different activation functions are tested for the performance of the fingerprint recognition algorithm framework.The results show that the Sigmoid function can achieve 81% accuracy with a lower number of neurons.

关 键 词：指纹识别 神经网络 细菌提取 图像预处理 SIGMOID函数 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]