摘要
为了提取蒙古族人脸图像的民族特征,提出了基于分块双向二维主成分和KNN分类的人脸识别方法。该算法利用(2D)2PCA对人脸的各个子块提取特征然后投影到特征子空间,然后使用基于距离和余弦的KNN分类策略。该算法可以高效的提取局部特征,并精确地计算协方差矩阵的特征向量。使用自建的蒙古族人脸样本库进行实验。实验结果表明:该算法可以得到更高的识别精度。
Aiming at the realization of national feature extraction, an improved face recognition method is proposed in order to preserve some essential local features, which is a fusion of bidirectional 2DPCA and local feature. First, the method decomposes the image into non - overlapping sub - blocks, each sub - block contains important local information and is extracted its feature using （2D）2PCA and then projecting onto feature subspace. Then, a combined KNN classifier is designed in the feature space. Experimental results on Mongolia Nationality face database showed that the proposed method achieved better recognition accuracy compared with several other face recognition algorithms.
出处
《内蒙古农业大学学报（自然科学版）》
CAS
2016年第5期114-119,共6页
Journal of Inner Mongolia Agricultural University(Natural Science Edition)
基金
国家自然科学基金项目(61562067)
内蒙古自治区高等学校科学研究项目(NJZY068)
关键词
蒙古族人脸识别
双向二维主成分分析
局部特征
特征提取
Mongolia nationality face recognition
bidirectional 2DPCA
local feature
feature extraction
