Next we are going to make some copies of the light mask to be used for deconvolution and noise reduction. Make 3 copies of the light mask by draging on the tab in the left bar to somewhere on the PI desktop. Rename them: deconv_mask, tgv_mask and mmt_mask.
Masks are used to protect certain parts of an image when manipulating it. The brightest parts of the mask allow the image to be altered while the dark parts protect it. For example, if you want to do noise reduction but don’t want to affect the stars or brighter parts of your image you can take the light mask and apply it to the image. In this case the mask is actually opposite of what we want, so we can invert it with a click of the ‘Invert Mask’ button which is on the menu bar (or through the Mask menu). You can also enable or disable the visibility of the mask so you can see what’s happening to your image while processing it with the ‘Show Mask’ button.
Let’s start on the deconvolution mask. Normally when I do deconvolution I use the light_mask without modification, but for this image it was difficult to control the effects so I decided to reduce the brights of the mask. In this case I simply cut the histogram range of the image in half, so instead of ranging from 0 to 1 it now ranges from 0 to 0.5. I used the CurvesTransformationTool and clicked on the upper right end point inside the histogram displan and drug it down to 0.5 (See Figure 12). Apply this to the deconv_mask image and you will see it get darker.
Figure 12: Decreaseing the range of the deconvolution mask
We are going to do something similar for the TGV mask. In this case I want to compress the range from the top and bottom. TGVDenoise can use an image for support, but I find that restricting it’s application to our image is useful. I want to protect the bright portions of the image more than the darker parts but not by too much which is why I compress the range of the mask. Figure 13 shows the CurvesTransformation window before applying it to the tgv_mask window. After this I want to shift the peak of the histogram to the mid point. I used the HistogramTransformation tool with a change to the midpoint and applied it to the tgv_mask (see Figure 14). When used with the TGV process this will give us close to a 50% blend between the original image and the noise reduced image with a slight preference to more NR data being allowed for the background than the high signal areas.
Figure 13: Modifying the TGV mask
Figure 14: Further modification of the TGV mask
I also use the MultiscaleMedianTransform for large scale noise reduction. I use very aggressive settings for this tool so our mask needs to be very protective to keep more of the original image. In this case I just want to shift the midpoint of the histogram up quite a ways. I usually center it around the ¾ mark, but sometimes go all the way to the 7/8 mark. Use the HT tool as showin in Figure 15 and apply it to the mmt_mask.
Figure 15: HistogramTransformation modification of the mmt_mask
Continued...
Regards,
David
