PROJECT TITLE :

Video In painting With Short-Term Windows: Application to Object Removal and Error Concealment - 2015

ABSTRACT:

In this paper, we tend to propose a brand new video inpainting methodology that applies to each static or free-moving camera videos. The tactic can be used for object removal, error concealment, and background reconstruction applications. To limit the computational time, a frame is inpainted by considering a small number of neighboring footage that are grouped into a group of footage (GoP). Additional specifically, to inpaint a frame, the tactic starts by aligning all the frames of the GoP. This is achieved by a region-based mostly homography computation methodology that permits us to strengthen the spatial consistency of aligned frames. Then, from the stack of aligned frames, an energy function primarily based on both spatial and temporal coherency terms is globally minimized. This energy function is efficient enough to produce prime quality results even when the quantity of footage in the GoP is quite small, e.g. twenty neighboring frames. This drastically reduces the algorithm complexity and makes the approach well suited to near real-time video editing applications along with for loss concealment applications. Experiments with many challenging video sequences show that the proposed technique provides visually pleasing results for object removal, error concealment, and background reconstruction context.