A robust pedestrian detection based on corner tracking

How to detect pedestrian quickly and accurately in complex traffic scenes is the key to pedestrian detection. In contrast to most standard approaches for pedestrian detection and tracking, the approach in this paper has better robust and accuracy. The core part of the approach is to extract local corner features of the objects using Moravec algorithm in video image and achieve tracking these corner features in different image sequence by block matching. Experiment results show the capacity of the approach to detection and tracking is effective in different complicated traffic scenes.