基于SFAM网络的机械零件不变性模式识别  

Workpieces Invariant Pattern Recognition Based on SFAM Network

作　　者：冯淑杰[1] 马蓉[2] 唐智勇[3] 

机构地区：[1]天津大学计算机学院,天津300072 [2]天津滨海职业学院计算机系,天津300451 [3]长春理工大学光电工程学院,长春130022

出　　处：《长春理工大学学报（自然科学版）》2007年第3期51-52,50,共3页Journal of Changchun University of Science and Technology(Natural Science Edition)

摘　　要：本文对机械零件进行图像采集,计算零件的正交Fourier-Mellin矩,采用SFAM网络和ART2网络分别对三种差别较小的机械零件进行识别试验。结果表明,SFAM网络克服了ART2网络抗噪性和分类性能的矛盾,比ART2网络具有更高的识别精度。The images ofworkpieces are captured, and the orthogonal Fourier-Mellin moments of the workpieces are calculated, then SFAM and ART2 networks are introduced to recognize the workpieces in this paper. The SFAM network can get over the conflict of anti-noise and categorizing in the ART2 network. Experiments of recognition to three workpieces which are little different from each other are made. The results of experiments show that the SFAM network has a better precision of recognition than the ART2 network.

关 键 词：SFAM网络 正交Fourier-Mellin矩 机械零件 模式识别 

分 类 号：TP302[自动化与计算机技术—计算机系统结构] O235[自动化与计算机技术—计算机科学与技术]