中文摘要：

针对传统主成分分析（PCA）算法中图像数据单行或单列存贮给存贮和求解带来压力等问题,通过对PCA在图像降维和主特征提取应用中的算法分析,提出一种二维PCA和最近距离分类法相结合的特征对象提取算法,通过集成LAPACK库函数进行特征值求解,提高求解速度以及算法稳定性和可信度。离线扣件状态识别综合实验结果表明,采用联合特征识别方法能够有效区分左右扣件和丢失扣件,满足了扣件图像离线识别的要求。

英文摘要：

In traditional PCA algorithm,image data are restored in single row or column,which results in great pressure for data storage and algorithm solving.After applying PCA algorithm analysis in image dimension reduction and main component extraction,the two-dimensional PCA algorithm integrated with nearest distance classifying method was advanced,which was then resolved using LAPACK functions.It has high computing and resolving speed,high computing stability and credibility.In the experiment of off-line fastener recognition,left,right and lost fasteners are effectively distinguished through joint classification,and the requirement of offline recognition is met.