From Images to Video: View Morphing of Three Images

This paper presents a novel image-based approach to generate a video of a scene based on only three wide-baseline uncalibrated images without the use of a 3D model. After automatically recovering an accurate trifocal plane amongst the original cameras, we compute the correct dense disparity maps between each rectified image pairs by using our feature-based trinocular-stereo algorithm. Employing the barycentric warping scheme with the computed disparity, an arbitrary novel view located in the trifocal plane is generated. Furthermore, after the self-calibration of these three cameras, we correctly augment 3D objects into the virtual environment synthesized by the triview morphing algorithm. We extend our approach to dynamic scene synthesis using three still images, where several rigid objects may move in any orientation or direction. After segmenting three reference frames into several layers, the novel views in the dynamic scene can be generated by applying our algorithm. The experiments are presented to illustrate that a series of photo-realistic virtual views can be generated to fly through a virtual environment covered by several static cameras.