一种基于支持向量机的行人识别方法研究OA北大核心CSCDCSTPCD
Research on pedestrian recognition method based on support vector machines
研究了基于支持向量机的车辆前方行人识别方法.通过提取样本的类Haar特征,采用AdaBoost算法训练得到了分割行人的级联分类器,实现了行人候选区域的快速分割;提取了样本的纹理特征、对称性特征、边界矩特征以及梯度方向特征,组成表征行人的多维特征向量,采用支持向量机训练得到了识别行人的分类器.试验结果验证了所提算法的有效性,获得约75%的行人检测率.
A support vector machine-based recognition method of pedestrian in front of automotive is introduced. The Haar-like characteristics of samples were selected and calculated, and the cascaded classifiers were trained using AdaBoost algorithm to lastly segment candidate pedestrian areas from the image. To form a multidimensional pedestrian describing vector, the texture and symmetry features, as well as boundary moment and gradient oriented features were abstra…查看全部>>
郭烈;张明恒;李琳辉;赵一兵
大连理工大学工业装备结构分析国家重点实验室,辽宁大连116024大连理工大学汽车工程学院,辽宁大连116024大连理工大学汽车工程学院,辽宁大连116024大连理工大学汽车工程学院,辽宁大连116024
信息技术与安全科学
汽车主动安全行人识别AdaBoost算法支持向量机
automotive active safetypedestrian recognitionAdaBoost algorithmsupport vectormachines
《大连理工大学学报》 2011 (4)
604-610,7
基金项目:中央高校基本科研业务费资助项目(20083013893308).
评论