基于Adaboost的人脸检测技术

利用OpenCV图像处理软件,选取扩展的Haar like特征,实现了基于Adaboost的人脸检测器.阐述了Ad aboost人脸检测技术的基本原理,深入分析了强分类器的个数及排序对检测器性能的影响,并对3种分类方法得到的分级分类器的检测能力进行了比较,最后给出一种实用的分类方法.实验结果表明,该检测器检测速度快,开发周期短,具有可行性.

Research on face detection based on Adaboost

An adaboost-based face detection system is implemented by selecting expanded Haar_like characteristics with the help of Intel's OpenCV software. In this article,the theoretical basis is illustrated in detail, how the parameters such as number and arrangement of the stage classifiers affect the system's capacity is analysed, and the capacities of cascaded classifiers of three classification methods are compared.Finally,a practical classification methodology is given.Experimental results prove the efficiency and feasibility of the system.