Figure 2
Example showing the processing steps each projection goes through in order to identify spots generated by diffraction of the sample. The data from the scan (a) are first divided by the data from a reference scan without the sample, known as the flat-field image (b), and then multiplied by 100, providing an image without the intensity resulting from the direct beam (c). A rolling median correction is applied to (c) to remove additional noise, resulting in (d); the spots are finally segmented by using a Laplacian of Gaussian method (e). The result of the aforementioned steps is a binary image where the identified spots are 1 and the background 0. The scale of each image is related to the size of the detector. Note that the sample projection most clearly visible in (c) has been magnified by a factor of 2. |