[Previous] | [Session 76] | [Next]
N. Wherry, M. R. Blanton, D. W. Hogg (NYU), R. H. Lupton (Princeton)
We present an algorithm for producing detailed RGB composites (\textit{i.e.,} color pictures) from 3-band astronomical images. The value of each RGB composite pixel is modified monotonically with astronomical image intensity, but the ratios between R, G and B (\textit{i.e.,} hue and saturation) are fixed by the colors of the astronomical image pixel. Even pixels that saturate the dynamic range of the RGB composite are kept fixed at their correct hue and saturation. The algorithm is: R = r\,f(r,g,b), \quad G = g\,f(r,g,b), \quad B = b\,f(r,g,b), \mathrm{where} \quad f(r,g,b)=\mathrm{arcsinh}\left[\beta(r+g+b)\right]/\beta(r+g+b)\max(r,g,b) and r, g, b are scaled image intensities, and \beta is a user-chosen nonlinearity parameter that controls the degree to which the stretch is logarithmic (\mathrm{arcsinh}[\beta\,I] \approx \beta\,I for \betaarrow 0; \mathrm{arcsinh}[\beta\,I]\approx \ln[2\,\beta\,I] for \betaarrow\infty). More logarithmic stretches increase the apparent dynamic range; they can be used to simultaneously show faint spiral arms and the bright cores of star clusters. Most importantly, the stretch is always applied to the R, G and B values together in order to keep the color ratios (\textit{i.e.,} hue and saturation) fixed.
Images produced by our method are both better looking and more informative. For example, one can see low surface brightness and tidal features in the outskirts of a galaxy, and color gradients in the nucleus. We compare these images to equivalents made by traditional methods and show them to be far superior.
The author(s) of this abstract have provided an email address for comments about the abstract: inkwherry@hotmail.com
[Previous] | [Session 76] | [Next]
Bulletin of the American Astronomical Society, 36 #2
© YEAR. The American Astronomical Soceity.