| Abstract: | We describe a fast, reliable and automatic algorithm for image sequence inpainting that combines spatio-temporal interpolation with fine texture preservation inside missing areas. The algorithm provides an estimate of the inpainting error by using an automatic geometric recognition of missing regions. Computational kernels are sparse linear systems solved using Generalized Minimum RESidual iterative method equipped with AMG multigrid preconditioner. Experiments on synthetic and real data are discussed. |
