【摘要】 文章提出一种基于人脸部件表示的级联线性判别分析人脸识别方法。该方法将人脸图像划分为具有交叠区域的多个部件,对每个部件应用线性判别分析以寻找该部件的判别方向,然后对所有部件应用线性判别分析以寻找总体最优判别方向。以从该级联线性判别分析提取的特征作为人脸描述。在FERET人脸库上的人脸识别和人脸确认的实验结果表明,该方法优于传统的基于全局图像的Fisherface方法。更多还原

【Abstract】 This paper presents a face recognition method based on cascade Linear Discriminant Analysis(LDA) of thecomponent-based face representation.In the proposed method,a face image is represented as multi-components with overlap at the neighboring area.LDA is conducted on each component to find the discriminant direction.Then,LDA is conducted on all the components to find the best discriminant direction.The feature extracted from the cascade LDA is the final face descriptor.Our experiments on the FERET face database have illustrated the effectiveness of the proposed method compared with the traditional global-based Fisherface method both for face recognition and verification.更多还原