2univsersity of Rennes 1/IRISA
IEEE Transactions on Image Processing (2015)
In this paper, we propose a new method for inpainting missing areas in static or free-moving camera videos. The method can be used for object removal, error concealment and background reconstruction applications. To limit the calculation time per frame, a frame is inpainted by considering a small number of neighboring pictures which are grouped into a group of pictures (GOP). More specifically, to inpaint a given target frame, the method starts by aligning all the frames of the GOP. This is achieved by a region-based homography computation method which allows to strengthen the spatial consistency of aligned frames. Then, from the stack of aligned frames, the inpainting process is performed by globally minimizing an energy function considering both space and time coherency terms. This energy function is efficient enough to provide high quality results even when the number of pictures in the GoP is rather small, e.g. 20 neighboring frames. This drastically reduces the algorithm complexity and makes the approach well suited for near real-time video editing applications as well as for loss concealment applications. Experiments with several challenging video sequences show that the proposed method provides more visually pleasing results than the most recent and efficient state-of-the-art approaches.