Adaptive thresholding hosvd algorithm with iterative regularization for image denoising

In this paper, we propose a very simple 3D patch stack based image denoising method by Higher Order Singular Value Decomposition (HOSVD). We used the idea of iterative regularization from spatially adaptive iterative singular-value thresholding(SAIST) to design our algorithm, which indicates more faster convergence speed than some other methods. By using the parallel computing technique for implementing the algorithm, the computational complexity is highly reduced. The experiments also show good PNSR result with different noise levels.