摘要
为完成多个微小零件的识别,提出了一种改进的支持向量机分类算法.该算法应用基于边缘提取的不变矩获得特征属性,利用基于粗糙集的可辨识矩阵的启发式属性约简算法获得特征属性的约简,最后应用支持向量机进行目标识别分类.比较了使用支持向量机分类和使用提出的改进支持向量机分类对多个微小零件识别的效果.在显微视觉环境下的实验表明,提出的改进支持向量机分类方法能满足系统应用要求,分辨率达95%.
In order to identify multi micro parts, an improved SVM （support vector machine） algorithm is presented, which employs invariant moments based on edge extraction to obtain feature attribute and then presents a heuristic attribute reduction algorithm based on rough set＇s discernable matrix to obtain attribute reduction. At last, SVM is used to identify and classify the targets. The effect on identifying multi micro parts by SVM is compared with that by the proposed improved SVM. The experiment results under micro vision environment show that the proposed improved SVM classification method can meet the system application requirements, with the resolution of 95 percents.
出处
《机器人》
EI
CSCD
北大核心
2009年第2期118-123,共6页
Robot
基金
国家自然科学基金资助项目(60873032)
国家863计划资助项目(2008AA8041302).
关键词
不变矩
改进的支持向量机
粗糙集:属性约简
目标识别
invariant moment
improved support vector machine
rough set
attribute reduction
target identification
