A Comparison of PDE-based Non-Linear Anisotropic Diffusion Techniques for Image Denoising
Sisira K. Weeratunga, Chandrika Kamath, Sisira K. Weeratunga, Rika Kamath · University of North Texas Digital Library (University of North Texas) · 2002
PDE-based, non-linear di#usion techniques are an e#ective way to denoise images. In a previous study, we investigated the e#ects of di#erent parameters in the implementation of isotropic, non-linear di#usion. Using synthetic and real images, we showed that for images corrupted with additive Gaussian noise, such methods are quite e#ective, leading to lower mean-squared-error values in comparison with spatial filters and wavelet-based approaches. In this paper, we extend this work to include anisotropic di#usion, where the di#usivity is a tensor valued function which can be adapted to local edge orientation. This allows smoothing along the edges, but not perpendicular to it. We consider several anisotropic di#usivity functions as well as approaches for discretizing the di#usion operator that minimize the mesh orientation e#ects. We investigate how these tensor-valued di#usivity functions compare in image quality, ease of use, and computational costs relative to simple spatial filters, the more complex bilateral filters, wavelet-based methods, and isotropic non-linear di#usion based techniques.