Here I want to show you an example of the tool B3Estimator (black body estimator) first, to simulate an H-alpha filter when you don't have one, and second, to enhance some features that are not apparent in your data (but there they are). Maybe you remember that this tool is able to generate a synthetic image in any desired filter from two images of a given region obtained with two different filters, supposing that every pixel in the image behaves as a black body (as stars do). It can be used to have a good estimation of the continuum emission.
Our example starts with the following images of (a section of) the Stephan's Quintet, obtained by our colleague
Vicent Peris with threebroadband R, G, B filters plus a narrowband H-alpha filter from the Calar Alto Observatory:
As can be seen, there is a strong H-alpha signal in the bottom-right galaxy (NGC 7320) which, in fact, is much closer than the other galaxies. Only four of the five galaxies in Stephan's Quintet form a physical association (Hickson Compact Group 92), but not the nearer NGC 7320 with a small redshift (790 km/s, while the other four exhibit large redshifts, near 6600 km/s). Do these other galaxies lack of H-alpha emission? No. What happens is that, due to the higher redshift, their H-alpha emission lines are shifted outside the standard H-alpha filter. Can we recover this information?
Here we have an opportunity to use the B3Estimator tool. Using the B and G images, we are going to generate a synthetic R image (according to a blackbody law) and subtract it to the real R image. The H-alpha emission (shifted or not) should appear as an excess of red emission over the blackbody. And so any kind of reddening. As in the B3Estimator tool the mathematical derivation is done for spectral emission, we should first divide with pixelmath every image by the width of the filter. This is mandatory if the filter widths are very different, but as in this case the three filters have a similar bandwidth we do not need to do this. Once generated the synthetic R image (introducing in B3Estimator the central wavelength of each filter), we subtract it to the real R image without rescaling. We are just interesteed in the excess over the continuum. What we get is the following:
The black background is noisy because the algorithm is trying to fit to a blackbody curve what in fact is only poissonian noise without a real signal. But we can see how most of the stars have been efficiently removed (as they are nearly a blackbody) and also the main body of the galaxies. The excesses over the red continuum inside the galaxies are representative of the H-alpha emission, indeed: the coincidence with the regions detected in NGC 7320 in the real H-alpha image, is very good.
Note also that there is a strange feature in the region between the upper galaxies (NGC 7319 and NGC 7318A/B) with the shape of a bow. Is it real? Compare it with this image, where the light blue feature is X-ray emission obtained with the Chandra space telescope:
It is a bow shock produced by the collision of the galaxies, that heat up the gas producing this emission in X-ray, and apparently also an excess (probably H-alpha emission) in the red zone of the spectrum. To finish, the final color image, with real R + estimated H-alpha in the red channel:
