A Non-Local Tensor Completion Algorithm Based on Weighted Tensor Nuclear Norm

In this paper, we proposed an image inpainting algorithm, including interpolation step and a non-local tensor completion based on weighted nuclear norm. Specifically, the algorithm adopts triangular linear firstly to preform observation image. Second, extract similar patches in using patch match rearrange them tensor. Then, use norm recover tensors. Finally, regroup all these recovered tensors obtain final From experiments color RGB images, can see that performance of peak signal-to-noise ratio (PSNR) relative standard error (RSE) are significantly better than other methods.