For some time now I am looking into the field of night sky image processing. It is a huge topic and there are already a lot of different approaches and solutions to many of the existing problems. I have collected many star images over the time. To me it is very interesting to try different algorithms with my own data. This article starts with a problem every night sky photographer and astronomer is aware of: noise. I did some research and came across “anisotropic diffusion”.

What is anisotropic diffusion?

In short, anisotropic diffusion is a non-linear and space-variant transformation i.e. the transformation depends on the image content. The effect is that the resulting images preserve linear structures while at the same time smoothing is made along these structures. Additional information on anisotropic diffusion is available for example on wikipedia and here. Furthermore I found this youtube explanation quite helpful.

Continue reading →