Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
Using Gimp to reduce noise in series of X-Ray Dataset?
#1
Hello dear Gimp Pros,

bit of an unusual topic Big Grin

I'm doing chemistry/material research. In part of my research, X-Ray data is created in the form of gray scale images. I wonder if I could use image processing in Gimp (or maybe other software) to enhance my data quality.

The data in the images consists of 3 parts:
1) Reflection points: seen as dark spots in the images. The position, size and shape (round, oval) is part of the data acquisition.
2) Fluorescence: seen as a centered dark shadow.
3) random Noise: single points all overall the image. This noise has a preferential direction (up & down), which is due to the detector type used.


The images can be created in different "Cutoff" levels, which refer to an internal algorithms of the device to remove data from the image, based on the intensity of the recorded signal for each pixel. There are 10 levels.

In one image I marked very weak signal points, that are barely visible. The position of these points can be calculated from the other data points, so there is a point to be expected.

The position and shape of the reflection points is used for further research, so this is the data I would like to extract more precisely.

Is there a possibility to reduce the amount of the noise and fluorescence from the different cutoff levels? The process would have to be a fixed routine without manual work, so it can satisfy scientific standards.

Best regards,
DuroHeci

answer for more images

and the marked image


Attached Files Image(s)
                                           
Reply
#2
So, you have data, and this data is transformed into an image so that a human can evaluate it. That doesn't make that image a good basis for further processing; if you want to scientifically enhance your data you should work from the initial data. In addition, in Gimp, image processing is designed to process pictures and can be polluted by concepts that are nice for general images (gamma correction) but irrelevant for raw data.

There are signal processing libraries that can work wonders. With Python, stat with NumPy and SciPy.
Reply


Forum Jump: