摘要
当每个人只有一个训练样本时,最大散度差鉴别分析在人脸识别中的识别性能会降低,为了解决这一问题,提出了基于模糊决策和最大散度差鉴别分析的单样本人脸识别算法。通过对每个训练样本进行适当的分块,从而获得较多的训练样本个数,在这些新的训练样本集上应用类内中间值最大散度差鉴别分析算法得到最优投影矩阵,并基于这个最优投影矩阵可以计算训练样本和待测试样本的特征。对模糊决策方法进行分类。在著名的ORL和FERET人脸数据库上的大量实验结果表明,该算法可以提高识别率。
To improve the recognition performance of face recognition with single training sample,a face recognition method based on fuzzy decision and maximum scatter difference with single training sample is proposed in this paper.In this method,each training sample is portioned several blocks to increase the number of the element of the training sample set,on which the maximum scatter difference algorithm is performed to get the optimal projection matrix.Therefore,the features of the training sample and testing facial images can be obtained by projecting them on the optima projection matrix achieved above.During the recognition stage,the fuzzy decision is used to do the classification.Extensive experiment results on ORL and FERET illustrate the feasibility of the proposed method.
出处
《计算机工程与应用》
CSCD
北大核心
2011年第19期200-202,共3页
Computer Engineering and Applications
关键词
人脸识别
最大散度差鉴别分析
模糊决策
单训练样本
face recognition
maximum scatter difference
fuzzy decision
single training sample
