【摘要】 针对基于小样本集人脸图像的识别能力低,计算复杂度高的问题,提出了一种基于样本融合的核稀疏表示方法(KSRMSF).该方法首先通过在原始样本集中添加镜像训练样本和对称训练样本,扩大了原始样本集的规模,接着使用基于高斯核函数的算法从扩充后的训练样本集中挑选若干个最近邻训练样本,利用这组最近邻样本的线性组合表示待识别的测试样本,根据L2范式的结果对测试样本进行分类,通过修改最近邻样本数获得更高的分类精度.实验结果表明该方法比同类识别算法有更好的识别效果.更多还原

【Abstract】 To improve the recognition rate of face recognition method based on small sample set of face images and reduce the high computational complexity,a kernel sparse representation method based on samples fusion method( KSRMSF) is proposed. The proposed method first extends the training samples to form a new training set by adding some mirror virtual training samples and symmetrical ones,and uses a algorithm based on Gaussian Kernel Function to determine N nearest neighbors of the testing sample from the new training samples. Finally,the KSRMSF represents the testing sample as a linear combination of the determinated N nearest neighbors and performs the classification according to the L2 norm.Through the different values of N set,the classification is more accurate. Many experiments show that the KSRMSF can get a better classification result than the same type of algorithm.更多还原