多特征融合的AdaBoost快速训练人脸检测算法

Ada Boost Fast Training Algorithm Based on Multi-feature Fusion for Face Detection

为了提高人脸检测分类器的训练速度及人脸检测的速度与性能,提出一种基于多特征融合的AdaBoost快速训练人脸检测算法.首先介绍三种特征:Haar-like特征、三角积分特征、基于形态学梯度的边缘方位场特征.接着,给出多特征相融合的AdaBoost快速训练人脸检测算法.实验结果表明提出的算法可以减少人脸检测分类器的训练时间,能够提高人脸检测的速度和性能.

To improve the training speed of classifier and the speed and performance of face detection, the AdaBoost fast training algo-rithm based on multi-feature fusion is proposed in this paper. Firstly,the Haar-like features,the triangular integral features and theedge-orientation field features based on morphological gradient are introduced. Then, the AdaBoost fast training algorithm based on theabove three kinds of features is proposed. The results of the experiment show that the proposed algorithm could reduce training time ofclassifier and improve the speed and performance of face detection effectively.