小型零件边缘缺陷识别算法研究

Research on Recognition Algorithm of Plastic Parts Edge Defect

以注塑环形薄片零件为研究对象,对零件的边缘缺陷识别算法进行了研究和分析。综合运用几何特征描述子和形状特征描述子等多个特征值完成了零件缺陷的提取与分析,并基于聚类分析,提出了改进的有向无环图支持向量机(directed acyclic graph support vector machine,DAG-SVM)分类器用于解决四类缺陷分类问题,能准确识别零件缺陷。

This paper takes annular plastic sheet as the object,studying and analyzing the recognition algorithm of edge detection.Based on the features of the research object,it identifies and analyzes the efects of the parts by several feature values including the geometric feature descriptors and the shape feature descriptors.And an improved directed acyclic graph support vector machine （DAG-SVM） classifier is proposed to solve the 4-class defect classification problem and accurately identify the defects of the parts,based on cluster analysis.