基于SVM-AdaBoost算法的行人检测方法

提出了一种基于SVM-AdaBoost算法的行人检测方法。该方法是通过改进和扩展Haar-like特征值,对AdaBoost级联分类器的算法加以改进,使用SVM作为AdaBoost的弱分类器,通过选择确定合适的核函数参数,提高分类精度,减少训练时间。实验结果表明,这种行人检测方法性能稳定,实时性和鲁棒性均优于传统的行人检测方法。

A pedestrian detection method based on SVM-AdaBoost algorithm

This paper presents a method for pedestrian detection based on SVM -AdaBoost algo-rithm. The method through expand and improve Haar-like features, to improve the AdaBoost cascade classifier algorithm use SVM as AdaBoost weak classifier, by selecting the appropriate kernel parameters to improve the classification accuracy and reduce training time. Experimental results show that the new pedestrian detection method stable performance, robustness and real-time are better than traditional pe-destrian detection methods.