A Real-Time Video Noise Reduction Algorithm in the Dusk Environment

This paper is dedicated to achieving real-time video noise reduction in the dusk environment. For coping with the inherent problem of blur moving object, the basic idea behind the proposed method is to use different filtering strategies for background and foreground (i.e., moving object). Firstly, background and foreground are separated by changing detection technique and then the temporal averaging filter and spatial filter are utilized for recovering the background and foreground, respectively. Thus, detail preserving can be performed as well as noise reduction in the realtime process. Experimental results show that the proposed method manifests a better visual quality and provides a compressing ratio of above 90% and at least 15% over other filters.