TY - JOUR AU - Kurz, Christian AU - Thormählen, Thorsten AU - Seidel, Hans-Peter PY - 2011 DA - 2011// TI - Visual Fixation for 3D Video Stabilization JO - JVRB - Journal of Virtual Reality and Broadcasting VL - 8(2011) IS - 2 KW - camera motion estimation KW - camera shake KW - structure-from-motion KW - video stabilization KW - visual fixation AB - Visual fixation is employed by humans and some animalsto keep a specific 3D location at the center ofthe visual gaze. Inspired by this phenomenon in nature,this paper explores the idea to transfer this mechanismto the context of video stabilization for a handheldvideo camera. A novel approach is presentedthat stabilizes a video by fixating on automaticallyextracted 3D target points. This approach is differentfrom existing automatic solutions that stabilizethe video by smoothing. To determine the 3D targetpoints, the recorded scene is analyzed with a stateof-the-art structure-from-motion algorithm, which estimatescamera motion and reconstructs a 3D pointcloud of the static scene objects. Special algorithmsare presented that search either virtual or real 3D targetpoints, which back-project close to the center of theimage for as long a period of time as possible. The stabilization algorithm then transforms the original imagesof the sequence so that these 3D target points arekept exactly in the center of the image, which, in caseof real 3D target points, produces a perfectly stable resultat the image center. Furthermore, different methodsof additional user interaction are investigated. It isshown that the stabilization process can easily be controlledand that it can be combined with state-of-thearttracking techniques in order to obtain a powerfulimage stabilization tool. The approach is evaluated ona variety of videos taken with a hand-held camera innatural scenes. SN - 1860-2037 UR - http://nbn-resolving.de/urn:nbn:de:0009-6-28222 DO - 10.20385/1860-2037/8.2011.2 ID - kurz2011 ER -