基于粗糙集和支持向量机的火炮内膛疵病识别方法

Study of Gun Bore Flaw Classification Method Based on Fuzzy Rough Set and Support Vector Machine

火炮内膛疵病智能识别是火炮内膛窥测的最终目标,它涉及到内膛疵病的特征提取和疵病识别两方面。首先建立了包括疵病形状、纹理与颜色特征的火炮内膛疵病特征体系;并采用模糊粗糙集理论分析了各疵病特征对疵病识别的敏感性,由此优化了疵病特征体系,降低了疵病特征维数;建立了最小二乘支持向量机小样本、非线性数据特征的多疵病分类器,提高了疵病识别效率和质量。

Gun bore flaws intellective identification is final object of gun bore spying. It involves two aspects of feature extraction and flaw identification. In this paper,the gun bore flaw feature system which consists of shape,texture and color feature is built. Flaw identification sensitivity is analyzed based on fuzzy rough set,and the flaw feature dimensions are reduced by optimizing the flaw feature system. Using the small sample and non-linear data multi-classification organ of least-square support vector machine,the flaw identification efficiency and quality are heightened.