Non-linear Anti-Geometric Diffusion based Adaptive Singular Value Decomposition Method for Efficient Image De-noising

˗Image de-noising is an important challenging issue in image pre-processing. Two popular methods to the problem are partial differential equation (PDE) based nonlinear diffusion method and singular value decomposition (SVD) method. Various image de-noising algorithms based on these two methods have been independently developed. This paper proposes an approach for image de-noising by performing SVD filtering in detail sub blocks of the image subspace, where SVD filtering is adaptive to the inhomogeneous nature of the natural images. The method utilizes nonlinear antigeometric diffusion method for the fast segmentation of the image into detail sub blocks and for the noise attenuation from the singular vectors of the image subspace. Comparisons were made with respect to both PDE based nonlinear diffusion method and wavelet based adaptive SVD filtering method. The results indicate that the proposed method is efficient and fast for image de-