基于分块PCA和SVM的零件识别分类系统

Part Recognition and Classification System Based on Block PCA and SVM

针对工厂流水线零件的自动分拣,提出一种基于分块主成分分析和支持向量机(分块PCA-SVM)算法的零件识别分类系统。针对用PCA算法提取特征受外界噪声及光照环境影响较大的问题,本系统运用分块PCA算法,能够有效地抽取图像局部特征,进而提高识别率;采用基于高斯核函数的SVM(支持向量机)对特征数据进行训练,同时利用网格搜索法优化SVM参数。实验结果表明,改进的算法具有较高的性能,识别速度快且识别率达95%,能用于实际工厂流水线零件的自动分拣。

Aiming at the automatic sorting of factory assembly line parts,a part recognition and classification system based on block principal component analysis and support vector machine(block PCA-SVM)algorithm is proposed.Aiming at the problem that the extraction of features using the PCA algorithm is greatly affected by external noise and lighting environment,the use of the block PCA algorithm can effectively extract the local features of the image,thereby improving the recognition rate.SVM(Support Vector Machine)based on Gaussian kernel function is used to train the feature data,and the grid search method is used to optimize SVM parameters.Experimental results show that the improved algorithm has high performance,fast recognition speed and a recognition rate of 95%,which can be used for automatic sorting of parts in the actual factory assembly line.