Impact of bilateral filter parameters on medical image noise reduction and edge preservation

The objective of this paper is to examine the bilateral filter and its application to digital medical imaging. The bilateral filter is a non-linear adaptive smoothing filter which is capable of preserving the edges of image structures. Selection of the bilateral filter's parameters greatly influence how well an image is smoothed and how well edges are preserved.;The parameters of the bilateral filter were analyzed individually. The noise reduction for each parameter is related to the much simpler box (averaging) filter. Edges were also analyzed for each parameter. Techniques for proper parameter selection are studied and general guidelines are suggested. After the properties of each parameter were examined, they were recombined and applied to CT images. Solutions addressing the limitations of this filter for high noise CT images are also considered.;It was found that the bilateral filter is capable of providing effective smoothing and at the same time preserves edges which are otherwise blurred by other filters. The viability of this filter greatly depends on the selection of its parameters.