Computer Science · Computer Vision and Pattern Recognition
StereoPasting: Interactive Composition in Stereoscopic Images
Ruofeng Tong, Yun Zhang, Keli Cheng
See on the mapAbstract
We propose "StereoPasting," an efficient method for depth-consistent stereoscopic composition, in which a source 2D image is interactively blended into a target stereoscopic image. As we paint "disparity" on a 2D image, the disparity map of the selected region is gradually produced by edge-aware diffusion, and then blended with that of the target stereoscopic image. By considering constraints of the expected disparities and perspective scaling, the 2D object is warped to generate an image pair, which is then blended into the target image pair to get the composition result. The warping is formulated as an energy minimization, which could be solved in real time. We also present an interactive composition system, in which users can edit the disparity maps of 2D images by strokes, while viewing the composition results instantly. Experiments show that our method is intuitive and efficient for interactive stereoscopic composition. A lot of applications demonstrate the versatility of our method.