Acquisition of Pedestrian Trajectory around Vehicle Using On-board Camera

Abstract

This paper presents a general algorithm for pedestrian detection by on-board monocular camera, which can be applied to cameras of various viewing angle. The foreground objects which have discriminative motion-difference compared to that of background buildings are extracted as Regions of Interest (ROIs). Then, those ROIs are tracked by Spatio-Temporal MRF (S-T MRF) model as a possible pedestrian. While tracking process, ROIs are verified by HOG/Fisher cascade. Before classification stage, some ROIs are quickly rejected using geometric constraints. Relative trajectory between on-board camera and pedestrian is acquired with high accuracy based on coordinate information estimated at the stage of geometric constraint.