Fast bilateral filtering using motion detection
Zengrong Lei, Zhuo Su · 2011
In this paper, considering the information redundancy in video, a new fast bilateral filtering based on motion detection is proposed. The main idea of the method is using frame difference to detect the change area between two consecutive frames in the video, and then only filter the changed area; the unchanged part can directly use the filtering result of the previous frame at the corresponding position. Unlike traditional fast bilateral filtering, the algorithm in this paper focuses on how to accelerate in the time domains in video. Previous methods concentrated on accelerating in the space domain in single-frame. Experimental results show that the method proposed can greatly accelerate bilateral filtering while achieving high quality in video processing.