SVD-Based Image De-nosing with the Minimum Energy Model

This paper proposes a new solution integrating energy function into singular value decomposition (SVD) for image de-noising. By selecting the proper singular values that represent signal and discarding the ones that represent noise, the additive noise of an image can be eliminated effectively. In order to obtain the optimal number of the singular values for image reconstruction and to eliminate the noise, the paper presents a minimum energy model. The experiment results show that the established model is effective in the circumstance that the image has regular pattern.