Prescan-based Simulation for Pedestrian Detection
Dong Yongku · Journal of Transport Information and Safety · 2013
This paper introduces a method of pedestrian detection in real traffic scenarios,by using a fusion of data from both laser scanner and camera installed on a vehicle.The ROI(region of interests)induced from laser scanner through data clustering can be transformed into the image coordinate by multiplying aperspective transform matrix derived from the co-calibration between laser scanner and the camera.Then,the analyses are performed by looking into degree of symmetry,the pixels of vertical edge,and a match between the ROI and template based on Hausdorff distance,to determine whether a pedestrian is present in the image.This system is modeled in the Prescan traffic simulation software to test the robustness of the pedestrian detection algorithm introduced in this paper.