基于2DLDA与FSVM的人耳识别

Ear Recognition Based on 2DLDA and FSVM

针对人耳图像特征提取和识别方面存在的问题,提出一种将二维线性鉴别分析(2DLDA)和模糊支持向量机(FSVM)相结合的人耳图像识别方法。利用2DLDA将人耳图像直接投影,提取的人耳特征,可以保留人耳图像样本的大量类内和类间信息。同时,FSVM在支持向量机(SVM)的基础上引入隶属度参数,更加适合多类问题。实验结果表明,该方法与2DLDA相比具有更高的识别率。

The University enterprise R SeD Center of Measuring and Testing According to the existing problems on feature extraction and recognition of the human ear image,a recognition method of the human ear image is put forward,which combins the 2-dimensional linear discriminant analysis（2DLDA） and fuzzy support vector machines（FSVM）.2DLDA that directly projects to extract ear features can keep a large number of information of the human ear image on within-class and between-class samples.Meanwhile,FSVM based on the support vector machines（SVM） introductes membership parameters,which are more suitable for multi-class problems.The experimental results show that the method has higher recognition rate than 2DLDA.