Vision-Based Pedestrian Detection

摘要：

In environments where monocular camera is mounted on the moving vehicle, pedestrian detection becomes much more difficult. Especially, in cluttered scenes, the pedestrian detection is more challenging. A statisticsbased approach for pedestrian detection is proposed in such environment. Firstly, free driving space, pedestrian legs detection and local exhaustive search are combined to generate pedestrian candidate windows; symmetry and edge density features of pedestrian are used to remove nonpedestrian candidate windows. Then haar wavelets features and edge orientation histograms features are introduced.The SVM classifier, based on these features, is learnt from training images. Finally the trained SVM classifier is used for pedestrian candidate windows classification. Experimental results show the method has high performance in urban road and campus scenes. Index Terms C pedestrian detection. pedestrian segmentation. haar wavelet. edge orientation histograms. support vector machine.

作者: Ping Fu Dingyu Xue Mingxiu Lin Xinhe Xu

作者单位: Faculty of Information Science and Engineering, Northeastern University, Shenyang 110004, China

会议类型: 国际会议

会议名称: Fourth International Conference on Impulsive and Hybrid Dynamical Systems(ICIHDS 2007)(第四届国际脉冲和混合动力系统学术会议)

会议地点: 南宁

会议语种:英文

在线出版日期: 2007-07-20（万方平台首次上网日期，不代表论文的发表时间）