infocus #20 December 2010 Extending the Unsharp Mask Image Processing Filter

John C Russ demonstrates macrographically how filters found in many image processing software, can be used, and these same techniques can be used on images of microscopic samples.

DOI: 10.22443/rms.inf.1.62

The unsharp mask filter is widely used to enhance images. Several modifications and extensions to the conventional implementation of the method are illustrated
which improve the results. These include application to only the brightness information in colour images, use of the Difference of Gaussians to improve noise sensitivity,
using ratios rather than differences so that enhancement is more uniform across the full range of brightnesses and conditional or blended mixing of bright and dark
edge enhancement. Replacing convolution-based blurring with rank-based detail isolation (either using a median filter or morphological opening and closings), can eliminate the halos that appear next to edges in the conventional implementation. The examples include a variety of non-micrographic images to emphasise that
these techniques are applicable to all images and at all imaging scales.

Return to listing