基于AdaBoost的人脸检测算法

Research on Face Detection Algorithm Based on AdaBoost

着眼于图像识别技术的研究与探索,以MATLAB仿真软件为算法验证平台,采用AdaBoost算法进行人脸检测的研究。主要过程是利用Haar-like模型来表征人脸器官,在仿真软件中通过“积分图”转换来寻求Haar-like特征数值的快速计算。利用AdaBoost算法训练出一些具备人脸特性的弱分类器,通过设置不同权重的方式,把性能最佳的弱分类器群打造成为强分类器。实验结果表明,该算法能够实现人脸的快速有效检测,在智能安防系统中有广泛应用价值。

This paper focuses on the research and exploration of image recognition technology,using MATLAB simulation software as the algorithm verification platform,using AdaBoost algorithm for face detection research.The main process is to use the Haar-like model to represent the face organs,and find the fast calculation of Haar-like eigenvalue by"integral graph"conversion in the simulation software.AdaBoost algorithm is used to train some weak classifiers with face characteristics.By setting different weights,the best performance of the weak classifier group is built into a strong classifier.The experimental results show that the algorithm can realize the rapid and effective face detection,and has wide application value in intelligent security system.