基于Gabor变换和Adaboost算法的人体目标检测分类器

针对在图像中检测人体目标,提出一种基于Gabor变换和Adaboost算法的检测方法.首先利用二维Gabor小波变换进行特征提取,然后利用Adaboost算法对Gabor特征进行选取并训练强分类器.为了提高检测精度,提出采用单一正样本集合与多个负样本集合分别进行训练,形成多个强分类器级联的层级检测分类器.实验结果表明了该方法的有效性,同时显示该方法须与其它辅助手段相结合,才能提高检测的实时性.

Human detection classifier based on Cabor transform and Adaboost algorithm

A detection method based on Gabor transform and Adaboost algorithm is proposed to detect human objects in images. The basic idea of this algorithm is to perform feature selection by using the 2D Gabor transform and then obtain strong classifiers by using the Adaboost algorithm. To increase the detection rate, a single positive sample set is trained with multiple negative sample sets to generate a cascade classifier composed of several strong classifiers. Experimental results prove the validity of this method, and show that this method must be combined with other supplementary means to improve the real time performance of detection.