Robust background subtraction on traffic videos
Éldman de Oliveira Nunes, Aura Conci, Ángel Sánchez · International Conference on Systems, Signals and Image Processing · 2011
Background subtraction involves processing of a video sequence from a static camera to detect the foreground objects in all frames. This paper introduces a robust background subtraction technique, the Adaptive Local Threshold (ALT) algorithm, which it is based on the Approximate Median Filter (AMF). It has been applied to accurately extract the moving vehicles on complex weather traffic videos (i.e. fog and snow scenes). Experimental results have shown that the proposed algorithm produces similar qualitative detection results (based on the Jaccard coefficient) than AMF for the tested videos. Additionally, our method has the advantage of not needing any threshold parameter to detect the foreground targets.