Multi-sensor approach for people detection

摘要

Human detection is an important research topic for many researchers who are working with surveillance, safe driving, military and security applications. It is now becoming more and more appropriate with the global increase in the terrorism related activities. This paper presents an algorithm for people detection using information gathered from a composite moving sensor incorporating a camera and a laser range finder. Laser range finder is used to identify a region of interest (ROJ) where a moving object is likely to be present. Corresponding ROI in the visual image is then analyzed and a hierarchical template matching strategy is used to confirm the presence of a moving human. This approach improves the robustness of the template matching and the computational efficiency as the matching is only done in the resized ROI. The proposed strategy is evaluated through experimentation.