人脸识别论文：pose adaptive LDA based face recognition.pdfVIP

Pose Adaptive LDA Based Face Recognition Zhenger Wang, Xiaoqing Ding, Chi Fang Electronic Engineering Department, Tsinghua University, Beijing, China {wangzhe, dxq, fangchi} @ Abstract recognition performance on faces with pose variations [1]. However, it needs the precise location of several In this paper, a novel method based on pose face key points and it is time-consuming for high adaptive linear discriminant analysis (PALDA) is computational complexity. Linear object class [3] is proposed to deal with pose variation problems in face another virtual image generating method by 2-D face recognition when each person has only one frontal model, which also needs the precise location of facial training sample. The basic idea of the PALDA method points. However, Chai et al [4] proposed Locally is described as follows: first, the pose style of the test Linear Regression (LLR) method which requires only sample is estimated by one pose classifier; then, the the location of the centers of two eyes. corresponding LDA feature, which is robust to the Eigen light-field (ELF) [5] is another effective variation between the estimated pose style and the method for pose robust face recognition. This frontal pose style, is extracted. Since human faces are algorithm first estimates the ELFs of each f