摘要
由于矩阵奇异值(SVD)为依据的图像识别技术中,SVD与矩阵之间不是一一对应的。为此提出了同类图像的类SVD变换及其图像特征提取方法。并采用最近邻法进行分类识别,应用于人脸识别中,有较好的效果。
When applying SVD in image recognition, the SVD does not correspond to the matrix on a one-to-one basis. The thesis proposes a categorised SVD transform and its image feature extraction approach for categorised images, then applies the nearest neighbour classification and recognition. Experiments have shown good results when applying the method to face recognition.
出处
《计算机应用与软件》
CSCD
2010年第9期231-233,共3页
Computer Applications and Software
关键词
矩阵奇异值
图像的类SVD变换
图像特征提取
Matrix singular value decomposition （ Matrix SVD） Image categorised SVD transform Image feature extraction
