节点文献
基于Boosting方法的人脸检测
Face Detection Based on Boosting Algorithm
【摘要】 该文提出一种基于Boosting方法的人脸检测算法。先用特征脸方法构造一个基于重建图像信噪比的阈值函数用于人脸检测,在此基础上,该文利用Boosting方法构造一个基于信噪比阈值的检测函数序列,然后以一定的方式将它们组合成一个总检测函数,据此判别一幅图像是否为人脸图像。实验结果显示,这种方法明显提高了检测性能。
【Abstract】 This paper presents a new algorithm based on Boosting for face detection.Firstly,it builds a threshold-based function for face detection,and the threshold is a signal-to-noise-threshold of an image rebuilt by eigenface-method.Then it uses boosting algorithm and build s a serials of threshold-based functions.At last,a final function based on boosted combination of those threshold-based functions is found and it is used in face detection.Experiment-results show good detective performance of the algorithm.
【关键词】 人脸检测;
特征脸;
信噪比;
Boosting方法;
【Key words】 Human face detection; Eigenface; Signal to noice ratio; Boosting algorithm;
【Key words】 Human face detection; Eigenface; Signal to noice ratio; Boosting algorithm;
【基金】 湖南省自然科学基金资助项目(编号:99JJY20058)
- 【文献出处】 计算机工程与应用 ,Computer Engineering and Applications , 编辑部邮箱 ,2004年03期
- 【分类号】TP391.41
- 【被引频次】21
- 【下载频次】302