SHG (spectroheliography) software release version 3.1

The place for all your solar software; capture, stacking, post processing, ancillaries etc. It's all up for discussion here!
Post Reply
thesmiths
Almost There...
Almost There...
Posts: 1053
Joined: Mon Jun 04, 2012 8:13 pm
Location: London, England
Has thanked: 42 times
Been thanked: 1485 times

SHG (spectroheliography) software release version 3.1

Post by thesmiths »

We have taken account of a lot of feedback from users and contributors (JFP, JBB, Valerie, Matt C, and Ken H) and made a number of small improvements on our previous software version. We decided to call this version 3.1 and have put it into a release repository on GitHub:
https://github.com/thelondonsmiths/Sole ... s/tag/v3.1

A fairly significant change is how we represent the geometric distortion. We now consider that horizontal lines should remain horizontal (previously, vertical lines remained vertical). This likely has more physical significance with regards to how the image is taken (i.e. scanning parallel to either the RA or DEC axis). The practical effect is the image is now rotated from how it was presented before. Mathematically, the transformation matrix is now forced to have a zero in the bottom left (instead of the top right).

New (left) vs old (right) geometry orientation
New (left) vs old (right) geometry orientation
140123 new vs old.JPG (89.54 KiB) Viewed 1238 times

The image as generated now (on the left) takes up a bit less real estate so looks a little nicer (and probably saves a step in later image processing). There were other suggestions how to improve the details of the geometric correction but we decided not to make a decision on these for now.

H-alpha geometry correction graphic
H-alpha geometry correction graphic
140123_shift=10_ellipse_fit.png (195.19 KiB) Viewed 1238 times

We also now save the geometry correction graphic (as long as "Show graphics" is turned on and "Save CLAHE.png only" is turned off). This can be helpful in ascertaining what problems there may be in the scanning (fps vs scan rate, instrument tilt). Also, if the edge detection is completely failing, it may be possible to diagnose the issue from this graphic (a typical failure mode is a cloud appearing during the scan).

Calcium H protuberances: new (left) vs old (right)
Calcium H protuberances: new (left) vs old (right)
14255 protus new vs old.JPG (55.07 KiB) Viewed 1238 times

We have also spent some time improving the edge detection as this helps with circularisation of the image and is particularly noticeable in the protuberance image. As well as doing the +10 pixel shift introduced in the previous version (which "normalises" all the spectral lines to continuum images), we now create a histogram of the intensities, make a curve fit to a cubic equation, and find the local minimum of the cubic equation (a cubic equation will have at most one local minimum). We find this method gives much more robust and precise results. In the image above of Calcium H, the circularisation is more exact with the new method (on the left) and the black disk overlay can be made closer to the edge of solar limb.

Calcium H edge detection
Calcium H edge detection
144255_shift=10_ellipse_fit.png (196.85 KiB) Viewed 1238 times


Post Reply