多特征空间下的支持向量机及其在图像识别中的应用

分别在主成分分析(PCA)、独立成分分析(ICA)和线性鉴别分析(LDA)所构造的特征空间下用模糊支持向量机(FSVM)进行人脸识别。同时,提出了一种改进的FSVM方法,即利用FSVM和多叉决策树相结合的思想来设计人脸分类器,从而使FSVM分类器的速度得到了大幅度的提高。通过在ORL人脸库上的实验结果表明,该算法是有效的。

Improved fuzzy support vector machine based on different feature space and its application to face recognition

The face classifier design based on fuzzy support vector machine(FSVM) was presented.A comprehensive performance comparison of face recognition was made among three different feature spaces including PCA,LDA and ICA respectively.Meanwhile,an improved algorithm for face recognition was proposed,which combines the FSVM and triple decision tree.The experimental results concerning ORL face images prove the effectiveness of FSVM method,which is based on the fuzzy theory and decision tree and that the speed of classifier based on FSVM can be improved a lot.