上海交通大学学报

上海交通大学学报（自然版）

• 自动化技术、计算机技术 • 上一篇    下一篇

一种基于激光雷达和视觉的行人检测方法

钱慧佳a，杨明a，李顥a,王春香b   

  1. (上海交通大学 a.电子信息与电气工程学院; b.机械与动力工程学院， 上海 200240)
  • 收稿日期:2009-07-24 修回日期:1900-01-01 出版日期:2010-07-28 发布日期:2010-07-28

A Pedestrian Detection Method Based on Laser Scanner and Camera

QIAN Huijiaa,YANG Minga,LI Haoa,WANG Chun xiangb   

  1. （a. School of Electronic, Information and Electrical Engineering; b. School of Mechanical Engineering, Shanghai Jiaotong University, Shanghai 200240, China)
  • Received:2009-07-24 Revised:1900-01-01 Online:2010-07-28 Published:2010-07-28

摘要： 针对一般城市环境的道路交通安全问题，提出了一种联合激光雷达和视觉的行人检测方法.利用激光雷达和摄像头坐标之间的透视变换关系，将环境的深度信息映射到图像中，并使用航位角推算方法同步激光雷达和摄像头，然后提取兴趣区域.在激光雷达聚类过程中，提出了一种基于行人宽度信息的目标分割方法，能有效分割并排行人.在基于视觉的行人检测中，首先根据行人边缘对称性的特点，预处理兴趣区域，然后采用基于Hausdorff距离的模板匹配方法，根据激光雷达的深度信息匹配行人上半身模板.实验结果表明，该方法可以取得较理想的效果.

关键词: 激光雷达, 视觉, 透视变换, 同步, Hausdorff距离, 模板匹配

Abstract: For urban traffic safety, a camera and laser scanner codetection method was presented. A proper ROI (region of interests) can be located based on the perspective transform between the camera coordinate and laser scanner coordinate, and a method of synchronization is induced from dead reckoning. In the process of laser scanner’s cluster, a simple rule is proposed for object segmentation based on pedestrian’s width. It can separate pedestrian side by side effectively. Then, vision based method is used for identification of pedestrian. First, a method of pretreatment for ROI is introduced according to the characteristics of the symmetry of pedestrian’s edge. Second, a template matching method based on Hausdorff distance is applied to match the upperbody template by laser scanner’s distance refinement. The experimental results are relatively ideal.

中图分类号: