AIP4Win 2.0 supports "astronomical" pixel values ranging from 1x10 to
-1x10 and microscopic pixel values as small as 1x10 . This wide dynamic range means that you can sum images and easily work with pixel counts in the millions and billions, or average and work on images that express pixel values to three or four decimal places. Those decimals mean more dependable dark matching, more accurate flat-fielding, and more precise histogram shaping. In addition, AIP4Win fully supports floating-point FITS formats, so everything is saved and nothing lost when you save an image to disk. Every user will benefit from AIP4Witfs dedication to robust low-noise internal number-crunching.
In this new edition of the Handbook, we treat noise theory more fully than we did in the first. The current generation of CCDs performs so well that the random arrival times of photons—shot noise—is the primary limitation in their performance. In a new chapter titled "Counting Photons," we explore what Poisson noise means to astronomy. Serious imagers, educators, and students need to be familiar with this fundamental noise source, and those interested in making beautiful images will understand more fully how and why the long integration times that capture millions of photons yield the most beautiful images.
I became fascinated with noise three summers ago, while teaching a course in digital imaging at Portland State University. One practical benefit for the observers, teachers, and students is that A!P4Wiri s stellar photometry tools now perform a noise analysis based on the count of photons detected and pixels measured, all reported as a routine part of measuring stellar brightness. This analysis will alert observers to the precision they have attained in measuring stellar magnitude, and we trust that the feedback will assist them in making even more precise mea-
surements. For students and educators, it's an object lesson that noise statistics impose fundamental limits on what we can learn about the Universe.
Inevitably, however, one studies noise in search of ways to eliminate it. It was clear that spatial methods (such as pixel averaging) always blur image detail, and frequency methods fail because random noise is random in frequency space. However, a hybrid spatial-and-frequency analysis called the wavelet transform allows one to determine whether a pixel or cluster of pixels is probably real or whether it is just random. AIP4Win's wavelet tools grew out of this, as I boiled theory into practice. Because they are totally noise free, wavelet-processed images look strange to noise-accustomed eyes—so smooth and silky—yet every detail of a noisy image is faithfully replicated. If you were raised on a diet of film grain and random noise, wavelet processed images may not look right to you; but teachers and students are sure to find wavelets a powerful tool for the analysis of severely noise-limited images.
For the image-makers, we developed a noise-averaging spatial filter called the Smooth Background Tool. This tool smooths noise by averaging pixels in the dark sky background parts of an image, but it leaves bright areas untouched. Instead of hiding noise by making sky backgrounds jet black, imagers can lighten the sky to show the faint outer parts of galaxies and nebulae that are lost when the sky is solid black.
Another feature of the new AIP4Win is a suite of tools for creating synthetic images. These are intended for the educators and students to explore image constituents—bias, dark current, vignetting, image, and noise. Using these tools, you can build images with precisely known constituent parts and use them to explore the relationships between the sky background, stars, objects, and detector characteristics.
The other significant addition is enhanced coverage of color imaging. Five years ago, when we developed AIP4Wirfs first set of color tools, color imaging was quite new. Use of spectral class G2V solar-analog stars to attain accurate color balance had just caught on, and luminance-overlay color (LRGB) was a cutting-edge technique. After a lot of reading, thinking, and experimenting, we took the somewhat scary step of processing color images with AIP4Win in a luminance/ chrominance (LCh) color space rather than the traditional RGB color space. For astronomy, where luminance varies over a million-to-one range (black sky to brilliant star), the choice of LCh color space offers significant advantages over the limited scale of RGB. There are now two chapters in the Handbook devoted solely to color—and processing color images is now "native" to AIP4Win.
Three factors appear to be driving the widespread interest in color imaging: the availability of really good color-separation filters for monochrome CCD cameras, a new generation of high-quality Bayer-array CCD cameras, and the appearance of excellent digital single-lens reflex (DSLR) cameras. We developed new routines and procedures for loading, decoding, and displaying images from these cameras. I bought a Nikon D70 to learn the ins and outs of DSLR imaging and Jim and I have both experimented with a Canon 10D to get everything possible from the images they capture, in the processing making thousands of raw images, dark frames, and flat-field frames.
For the CCD imagers, we revised and updated AIP4Witfs Join Colors Tool. The software code behind this tool translates your filtered monochrome plus luminance images into color. Working in LCh, RGB, and Lab color spaces, we sought new ways to build images with bright, clean colors and crisp, clear detail. We also added a new Color Effects Tool that operates directly on color images from DSLR cameras and astronomical CCD cameras.
Finally, to prevent even the slightest loss of color information, AIP4Win can save color images in a special 96-bit FITS file (32-bit floating-point data in each color channel) as well as the traditional 48-bit TIFF format with 16-bits integer data in each color channel. If you save your images in this format, no information is lost, so you can always pick up and continue processing where you left off.
In addition to color tweaking, every one of A!P4Wiri s tools can access and process the luminance component of an image without disturbing its color balance or saturation, and the suite of color effects tools can alter chrominance while leaving image brightness untouched. By cleanly splitting image chrominance from image luminance, AIP4Win allows you to apply deconvolutions, correct brightness gradients, fix uneven sky backgrounds, replace bad star images, smooth sky backgrounds, and otherwise process any image—color or black and white—with any tool.
Was this article helpful?