摘要
本文提出了一种基于模式类特征空间统计分布的模糊隶属度函数模型,可有效地反映模式在特征空间中的真实分布,用于模式分类器输入特征的模糊化可获取更好的识别性能。作者以带钢表面缺陷检测系统为应用对象,以采自上海宝山钢铁公司冷轧带钢生产线的实际样本为训练集和测试集,对本文提出的模糊隶属度函数性能进行了测试,并与基于模糊C均值的隶属度函数进行了比较,测试结果显示,本文提出的基于模式类特征空间统计分布的模糊隶属度函数模型抗噪能力强,在提高分类器识别率和降低错误率上有明显优势。
In this paper a model of discrete fuzzy membership function based on statistical distribution of features of pattern is presented. It is used for the fuzziness of input features of classifier. It can describe the distribution of patterns in feature space truly and get better recognition result. The model presented in this paper is compared with the membership function based on fuzzy C-Means in test. The sample set used comes from the 1500 strip steel line in Bao Steel Company. The result reveals that the model of discrete fuzzy membership function presented in this paper has a strong ability of anti noises and has evident privilege in improving the precision and decreasing the error rate.
出处
《信号处理》
CSCD
2004年第2期170-173,共4页
Journal of Signal Processing
基金
国家经贸委技术创新计划项目
宝山钢铁公司立项
