Anisotropic Tensor Averaging Matrix Based Structure-Preserving Image Smoothing

摘要

Filtering is considered as a basic but a defiant task in most image processing areas because of noise. Structures of the image like edges would be blurred when a filter was applied to filter the noises. A novel structure-preserving smoothing algorithm was proposed in this paper. A nonlinear local structure tensor based anisotropic averaging matrix was presented to prevent the structures from filtering the noise. The effectiveness of the presented method was testified by the comparison results between the presented methods and other smoothing algorithms.