【作者】 王颖； 毋立芳； 关媛； 王涓涓；

【Author】 Wang ying Wu lifang Guan yuan Wang juanjuan School of electronic information and control engineering, Beijing University of Technology,Beijing,China 100022

【机构】 北京工业大学电子信息与控制学院；

【摘要】 论文研究基于LBP直方图的SVM人脸姿势估计方法。算法包括三部分:首先提取图像LBP特征直方图信息,它可以有效的降低图像特征的维数,然后将该特征用于训练SVM分类器,最后用得到的分类器进行姿势分类。用CMU-PIE人脸库中的9个姿势进行了分类训练和测试,实验结果表明算法能很好的区分这9类不同姿势并且对光照变化有一定的鲁棒性。更多还原

【Abstract】 In this paper,a SVM algorithm based LBP histogram is designed to estimate facial pose,the algorithm includes three parts:Firstly,the histograms of LBP features of the image are extracted.Then the the LBP(Local Binary Pattern) histogram is used to train 9 SVM classifiers for 9 poses,finaly the corresponding poses can be classified by the training SVM classifiers.9 poses of CMU-PIE face database are used for training and testing,The experimental result show that the algorithm can classify the nine poses effectively and it is robust to small illumination variation.更多还原