Parallax-Robust Surveillance Video Stitching

Parallax-Robust Surveillance Video Stitching

Blog Article

This paper presents a parallax-robust video stitching technique for timely synchronized surveillance video.An efficient two-stage video stitching procedure is proposed in this paper to build wide Field-of-View (FOV) videos for surveillance applications.In the stitching model calculation candy coated licorice pastels stage, we develop a layered warping algorithm to align the background scenes, which is location-dependent and turned out bunny spa headband to be more robust to parallax than the traditional global projective warping methods.

On the selective seam updating stage, we propose a change-detection based optimal seam selection approach to avert ghosting and artifacts caused by moving foregrounds.Experimental results demonstrate that our procedure can efficiently stitch multi-view videos into a wide FOV video output without ghosting and noticeable seams.