Remove stripes combo#

Description

This stripe removal algorithm is developed by Dr. Nghia Vo and is a part of the module httomolibgpu.prep.stripe for data correction. This filter works in the sinogram space to minimise stripes leading to Ring Artifacts in the reconstruction. See more about the filter in the original publication [12]. Also more information about the method can be found on the author’s software Sarepy.

Remove stripes combination uses four algorithms applied sequentially to the data in order to remove different types of stripe artefacts as presented in the paper [12].

Where and how to use it:

Remove stripes combination should be applied when different types of stripe artefacts present in the data (e.g. small or large or partial stripes). It is useful to have a look at this classification of different ring/stripe artefacts beforehand. The method can be used blindly, however, applying multiple pre-processing algorithms can distort the data, resulting in artefacts in the reconstruction. Meaning that if there are not to many stripes in the data and the method is still applied, could result in suboptimal reconstructions.

What are the adjustable parameters:

As this method based on multiple algorithms, including Remove stripes by sorting, it has a number of free parameters. The snr parameter is used to detect and remove large stripes. The associated to that method parameter la_size is used to remove large stripes, also by means of sorting. Parameter sm_size defines the size of the median filter when sorting is performed to remove thinner stripes.

Practical example:

In this example we demonstrate how to apply remove stripes combo algorithm to the normalised projection data.