Being new in this forum, would like to get some hints for solving the following problem. In the attached frame the short vertical line segments are emission lines. As can be clearly seen, these lines are grouped in horizontal curved bands. I need to get of this frame the +/- 24 bands. So far I would already be happy if I could get masks. As a next step I would then scan through each band (called order), calculate the centre points and respective widths, and fit for every order this centre line.
What I did already:
Order detection in every single column of the frame via typical signal processing. The problem is that I get rather good results in the bottom region until row 800. Above row 800 the orders are interrupted and so close to each other that it gets very confusing.
Fourier filtering: My intension was to separate horizontal from vertical information. For the time being (for the order extraction) I am not interested in the high number of vertical line segments, just in the bands being horizontal. The result for the top region (above row 800) seems to be better than before but below the intensities are too low so that no structure is detected.
Wavelet transformation: I hoped to be able to extract a wavelet component containing only the order structure but that does not seem so straight forward.
Before providing more details, I would like to ask for some directions in which I can search for solving this issue. The overall aim is to have 100% automatic processing. I am not an image processing professional but feel selfsufficient in Python. In that context I was looking into Astropy and scikit-image but without good results. Thanks in advance, Stefan
