【摘要】 文章的工作基于子空间分析框架,从特征融合的角度模拟人类视觉系统的自适应识别功能进行人脸识别.首先,利用主成分分析(PrincipalComponentAnalysis,PCA)提取人脸全局特征,在一个低维的“人脸子空间”中依照最近邻法则匹配测试样本;然后,针对人脸局部特征,提出了一种根据各局部子块(如眉、眼、鼻、嘴)的特征偏离程度进行自动加权的算法;最后,基于模糊综合的原理对全局与局部特征进行数据融合,给出最终识别结果.实验表明,该算法能很好地结合人脸图像全局和局部的互补信息,识别效果优于各单一模块的分类性能.更多还原

【Abstract】 In this paper, a method based on the fusion of global and local facial features in the framework of subspace analysis for face recognition is proposed. PCA (Principal Component (Analysis)) is performed to extract global features, and the results are then sent to a NN (Nearest-Neighbor) classifier for recognition. A special strategy is used to combine different local features such as eyes, eyebrows, nose and mouth according to their respective salience. The idea of FI (fuzzy integration) is adopted to fuse both global and local features and the final result is given. The experiments on the NLPR database demonstrate the effectiveness and feasibility of the proposed method.更多还原