Design And Implementation Of Image Stitching Method For Immersive 3D Animation Making

Immersive 3D animation technology begins to grow fast in movies,education,entertainment and other visualization areas.Compared to other technology,utilizing immersive 3D animation technology can bring the viewers a more visualized experience.Although there is huge potential in immersive 3D animation field,researchers mainly focus on hardware device development.So there is not much related content and the market potential is enormous.At the same time,stereo 3D technology has been rapidly developed and is rich in content.Thus,a transferring method from stereoscopic 3D animation scenes into immersive 3D panoramas is in need recently.Only in this way can we utilize 3D animation resources for second time and form a huge immersive panorama content database.Based on the stereo-vision principle,we design and develop a stereo image stitching system to ensure the viewing comfort of viewers when transferring 3D animation scenes into stereo panorama image pairs.In the article,we analyze the functions and performance need of the system in detail.Based on the stereo vision principle,we utilize MEL script language to sample the 3D animation scenes and form the image database.Then we combine SIFT algorithm and RANSAC algorithm to stitch panorama image pair of the 3D animation scene.After that,we refine semi-global matching algorithm using weighted median filter and get the parallax map of panorama image pair.Combining with comfort parallax range,we refine and smooth the parallax of seams.Finally,we corporate left eye panorama and refined parallax map to polish right-eye panorama so that the viewers can get comfortable viewing experience.After applying functional testing and performance testing,we realize all the functions of the system.