摘要
实际的人脸识别系统常常会面临小样本问题,为了提高在小样本情况下人脸识别的分类正确率,提出一种基于虚拟样本的高斯加权稀疏表示的人脸识别方法。该方法首先利用人脸的对称性来构造虚拟训练样本,扩充训练样本集;然后,对每个测试样本,利用高斯核距离度量该测试样本和各个训练样本的相似性关系,并将该高斯核距离作为训练样本的权值来形成加权的训练样本集:最后,利用稀疏表示方法进行人脸的识别分类。实验结果比较分析表明,该方法在小样本情况下可以获得更好的识别效果。
A real face recognition system often encounters small sample questions. In order to improve the classification accuracy in the case of small samples, we propose a Gaussian weighted sparse representation method based on virtual samples for face recognition. by exploiting the symmetry of the face to generate Gaussian kernel distance to measure the similarity First, the proposed method extends the training sample set virtual samples. Then, for a test sample, we exploit the relationship between the test sample and each training sample; and we put the Gaussian kernel distance as the weight of the training sample to form the weighted training sample set. Finally, we exploit the sparse representation method to perform classification. Experimental results show that the proposed method can obtain better recognition effects in the case of small samples.
出处
《控制工程》
CSCD
北大核心
2018年第3期488-492,共5页
Control Engineering of China
基金
国家自然科学基金资助项目(61261011)
关键词
人脸识别
小样本问题
虚拟训练样本
高斯核距离
加权的训练样本集
相似性关系
Face recognition
small sample question
virtual training sample
Gaussian kernel distance
weighted training sample set
similarity relationship