摘要
KIM算法可以有效解决帧差法易产生空洞以及背景减法不能检测出背景灰度接近目标的问题,得到准确的运动区域,再利用基于级联分类器的Adaboost算法对运动区域进行人脸检测。实验结果表明,该方法可以更加快速、准确地实现人脸检测,具有较好的实时性。
The KIM algorithm can effectively solve the problems such as the empty hole produced by frame difference method and the bray level of the object closing to that of the background which cannot be detected by the background subtraction, and obtain the moving region accurately and detect faces on the motion region by Adaboost algorithm based on cascade structure. The experimental results show that it can realize the face detection more quickly and accurately with better real -time performance.
出处
《微处理机》
2014年第2期44-47,51,共5页
Microprocessors
