中文摘要：

设计了一种新颖的双向层级结构分类器，并将其应用于面部特征点的定位。该分类器在训练过程中对正例样本和反例样本交替进行重采样，因而与采用传统的单向层级分类器的面部特征定位方法相比，采用这种新的双向层级结构分类器的方法具有以下优点：可以应对大规模的数据集；可以处理存在复杂变化的正反例样本；而且无论是在训练过程还是在测试过程，其算法都能快速地过滤大量的“易分”样本，执行效率非常高。在两个公开测试数据库上的实验结果表明，采用双向层级结构分类器的方法可以实现准确、快速的特征点定位。

英文摘要：

A novel bidirectional cascaded classifier was designed and it was applied to facial feature localization. In the training phase, the classifier resamples the positive and negative samples alternately, so compared with the traditional facial feature localizaion method based on a unidirectional cascaded classifier, the method based on the new classifier has the following advantages： it can cope with large scale data sets, can deal with complicated variations of the positive and negative training samples, and either in the training stage or in the test process its algorithm can rapidly filter large number of simple samples, thus obtaining a very high efficiency. The results of the extensive experiment on two public face databases verified the effectiveness and efficiency of the method based on the new classifier.