首页 | 本学科首页   官方微博 | 高级检索  
     检索      

基于Map-Reduce并行构架和AE的大数据指纹监控识别算法
引用本文:史乙力,张诚一.基于Map-Reduce并行构架和AE的大数据指纹监控识别算法[J].应用声学,2015,23(3).
作者姓名:史乙力  张诚一
作者单位:海南师范大学 信息科学技术学院 海口 571158,海南师范大学 信息科学技术学院 海口 571158
基金项目:国家自然科学基金(71140008);海南特区发展基金(2011SFO3);海南教育厅科研项目(Hjkj2013-15)
摘    要:传统的指纹识别监控方法在应用到以海量、在线和增量为特征的大数据环境中时,不能自动高效地进行识别,且识别率和识别精度都不高。由此设计了一种基于Map-Reduce并行框架和AE的大数据智能监控指纹识别算法。首先,描述了Hadoop和Map-Reduce实现并行计算的原理;然后,设计了基于AE和BP神经网络的识别模型,采用AE进行指纹图像自动特征提取,并提出了一种基于比较差异算法对AE进行参数初始化的训练算法,采用BP神经网络进行具体识别;最后,将识别过程分解为Map函数和Reduce函数,并对函数的具体功能进行了详细定义。在Map-Reduce并行框架下对FVC2004中的指纹数据库进行实验,实验结果表明了文中方法能自动、高效地进行指纹识别,与其他方法相比,具有识别效率高和识别精度高的优点,具有较大的优越性。

关 键 词:指纹识别  自动编码器  受限玻尔兹曼机  智能监控  Map-Reduce

Big Data Fingerprint Monitoring Recognition Algorithm Based on Map-Reduce Parallel Framework and AE
Chengyi Zhang.Big Data Fingerprint Monitoring Recognition Algorithm Based on Map-Reduce Parallel Framework and AE[J].Applied Acoustics,2015,23(3).
Authors:Chengyi Zhang
Institution:college of information#$NBSscience and technology,Hainan Normal University,Haikou,college of information#$NBSscience and technology,Hainan Normal University,Haikou
Abstract:Traditional fingerprint monitoring method has the defects such as not able to automatic efficiently recognize with high recognition rate and accuracy, on the condition of applying to the big data environment on the characteristic of enormous, on-line and increment. Therefore, a big data fingerprint monitoring recognition algorithm is proposed based on Map-Reduce parallel framework and AE. Firstly, the principle for parallel computing of Hadoop and Map-Reduce is specified, then the recognition model based on AE and BP neural network is designed, the AE is used to extract the feature for fingerprint, and a training algorithm based on comparison difference and AE is obtained. The BP neural network is used to recognize the fingerprint. Finally, the recognition process is divided to Map function and Reduce function, the specified function is also defined. The experiment is implemented in the Map-Reduce framework and the FVC2004 is used as the dataset, the result shows the method in this paper can automatically and efficiently to achieve fingerprint recognition, and compared with the other methods, it has the priority of high recognition rate and accuracy.
Keywords:fingerprint recognition  auto encoder  RBM  intelligent monitoring  Map-Reduce
点击此处可从《应用声学》浏览原始摘要信息
点击此处可从《应用声学》下载免费的PDF全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号