摘要
针对单一分类器人脸检测非常耗时的问题,提出了一种由粗到精的融合分类器结构模式加速人脸检测。该系统分为3个阶段:前两个阶段,使用Adaboost级联分类器快速排除大量简单的非人脸图像;最后一个阶段,使用非线性的支持向量机分类器,将已通过前两个阶段检测的复杂图像准确归类为人脸或非人脸。实验结果表明系统性能良好。
Aiming at the problem that training time of face detection using single classifier is extremely long, this paper present a combination of simple-to-complex and three-stage classifiers to speed up a face detection system. In this proposed system, a large number of simple non-face patterns are rejected quickly by two first stage Adaboost cascaded classifiers while the last stage uses a non-linear SVM classifier to robustly classify complex patterns that have past two first stage as either faces or non-faces. Experimentalresults show that our system can achieve comparable results to state of the art face detection systems.
出处
《微电子学与计算机》
CSCD
北大核心
2007年第5期69-72,共4页
Microelectronics & Computer