Asadduzzaman Babu, September 2011--April 2013, "Ratio-based Edge Detection Inspired Speckle Reducing Anisotropic Diffusion", Computer Science Department, Faculty of Science, University of Western Ontario, Canada.
M.Sc. Thesis Abstract
Speckle Reducing Anisotropic Diffusion, SRAD, is a multiplicative speckle noise reduction method. In highly speckled environment, SRAD occasionally produces over-smoothed, dislocated/broadened edge lines and inadequate de-noising on homogeneous image regions where the speckles are well developed. Moreover, the performance of SRAD is highly dependent on the initial selection of a good homogeneous area. To overcome these weaknesses, we propose two different ratio-based edge detection inspired extensions to SRAD. One of the proposed extensions incorporates an edge-sensitive boosting factor to guide the gradient and Laplacian operator based edge detector of SRAD. The edge-sensitive boosting factor is defined by the global edge information provided by a ratio based edge detector. The other proposed extension introduces a weighted diffusion function in the original diffusion model of SRAD. The proposed diffusion function is a weighted sum of two components – (1) a global ratio-based edge detection inspired component and (2) the original diffusion function of SRAD. A common scaling function selection strategy for both extensions and the use of a larger window size for gathering local statistics have also been proposed. The proposed filters show significant improvement in speckle de-noising and edge preservation.