Robust human detection from complex background by radial reach filter
Yutaka Satoh, Hideki Tanahashi, Y. Niwa, Kazuhiko Yamamoto · 2004
A novel statistical approach is proposed for detecting the human region from a complex background, named radial reach filter (RRF). The RRF evaluates local texture between a reference image (background) and the current scene by using radial reach correlation (RRC). The RRC is a new robust dissimilarity measure and has a well-formed probabilistic model of binary or normal density, and the model is effective for designing some threshold values of the RRF or tuning the monitoring system based on it in real world applications. Experimental results on real images show the effectiveness of the proposed method.