基于Adaboost算法的人脸检测分类器的设计与实现  

The Design and Implementation of the FaceDetection Categorizer Based on the Adaboost Algorithm

作　　者：兰胜坤[1] LAN Sheng-kun （School of Computer Science Communication and Information System, Guizhou University, Guiyang 550025, China）

机构地区：[1]贵州大学计算机学院通信与信息系统,贵州贵阳550025

出　　处：《电脑知识与技术》2010年第11期8828-8830,共3页Computer Knowledge and Technology

摘　　要：人脸检测是人脸识别的第一步，人脸检测的速度直接影响到人脸识别的精度和速度。Adaboost算法是一种快速人脸检测算法，其主要思想是训练若干个弱分类器，将弱分类器集成为一个低于一定错误率的强分类器。对样本图像进行HaarLike特征提取，并用积分图快速计算Haar_Like特征值，利用Adaboost算法实现人脸分类器，最终实现一个能快速进行人脸检测的系统。实验证明，此设计方法的精确度高。Facedetection is the first step of facerecognition, the speed of facedetection impacts on the precision and the speed of the fac- erecognition. Adaboost algorithm is a rapid facedetection algorithm, the main ideas are several weak categorizer will be trained, and be as- sembled to a strong categorizer which has a Jess than a certain false rate. Get the haar_like features from the samples. And calculate the value of the haar hke features by checking the integral data, design the face detection categorizer based on the adahoost algorithm, at last a quick face detection system is reached. Experimental results show the Categorizer has higher accuracy.

关 键 词：人脸识别 人脸检测 类HAAR特征 ADABOOST算法 分类器 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术;自动化与计算机技术—计算机科学与技术]