Image Completion Algorithm Based on γ Nuclear Norm Tensor Train

This paper innovatively proposes a new image completion algorithm, introducing the γ nuclear norm as a substitute for the rank function. The algorithm utilizes Tensor Train technique, further enhancing the computational efficiency and model performance. In the experimental part, this paper tests a large amount of image data on the conversion of low-order tensors to high-order tensors using Ket Augmentation technique. The results show that the proposed method is superior to the existing traditional methods in both the accuracy and robustness of data recovery, and highlights obvious advantages in terms of PSNR and SSIM values. Furthermore, experiments on the removal of local artifacts and small-scale noise also demonstrate the effectiveness of the proposed method.