Riemannian Accelerated Spectral Conjugate Gradient Method for the Tensor Completion Arising in the Color Image Restoration
摘要
In this paper, a novel tensor completion model is constructed by making use of the first order forward finite difference tensors, which can improve the local piecewise smoothness and preserve the edges of the color image. Based on the acceleration factor technique, an efficient Riemannian accelerated spectral conjugate gradient method is proposed to solve it. Its convergence theorem is also given. Numerical experiments with synthetic data and real images demonstrate the feasibility and effectiveness of the new method.