Calibration of Images

The sequence of operations on an electronic image, just after the acquisition and before it can be effectively analyzed, is very well documented. The first requirement is offset subtraction, which is equivalent to determining the zero point for each pixel (termed bias or offset). The second is the removal of the thermal signal produced in the sensor itself (termed dark current or dark-frame). The last is achieving pixel uniformity across the entire image field (termedflat-field). These operations are all elementary, but it is critical to carry them out optimally so as not to waste the precious information we have patiently gathered at the telescope. They are also very repetitive as they have to be applied to every electronic image. Let us see how IRIS solves these problems.

From the very beginning, IRIS has had a command-line user interface. When mouse clicks and pop-up windows did not exist, it was the only way to give instructions to a program. Even though IRIS has now moved to a Windows-oriented style, where basic commands are available in drop-down menus, IRIS still makes available a "console mode" where we can enter and edit commands (see Figure 6.2). This method of controlling software might be regarded as old-fashioned (and perhaps obsolete) but in reality it is more flexible and powerful than a mouse-click navigation style, where windows, menus and dialog boxes can completely cover the screen. Consider what professional astronomers use. They almost exclusively use command-line scripts in such standard professional software as IRAF or MIDAS. What is best for them is surely good enough for us! A common shortcut, when using a command-line interface, is not to type new commands in full but to edit previous ones. So when using a similar series of commands, only the first has to be typed in full - subsequent commands are edits of the previous one. This makes processing a sequence of images quick and straightforward. By default, the images are initially loaded and stored in a working directory. This directory is set up from a drop-down dialog box, as are all the general program parameters. As might be expected, the setup is automatically saved for the next usage.

Figure 6.2. Screen shot of IRIS in action. Richardson-Lucy Deconvolution has been carried out via the command line (RL2 15 1) and the result viewed with Modified Equalization selected from a drop-down menu.

The second stage of processing makes use of the operation SUB (i.e., subtract IMAGE1 - IMAGE2). The two parameters are the operand file names and a third is a constant added to each pixel after subtraction. For example, for a null value constant the full command is:


The result of the subtraction is in memory and is displayed on the screen. To keep the result it needs to be saved on the hard disk with the SAVE command. IRIS is also able to process a sequence of images. As an example, to subtract an offset image from the five images RAW1, RAW2, RAW3, RAW4, RAW5, only one command is used:


The first parameter is the generic name of the sequence to be processed, i.e., the name of the image file without the index number. IRIS adds the index itself to find the correct file on the hard disk. The second parameter is the offset image name to be subtracted from the sequence of raw images. The third parameter is the generic name of the resulting image sequence; again IRIS will add the index. The fourth parameter is a constant value to be added to each pixel of the resulting images. Finally, the fifth parameter is the number of images to be processed in the sequence. The command SUB2 therefore executes all the operations: RAW1-OFFSET=RESULT1, RAW2-OFFSET=RESULT2, etc.

This syntax is very common in IRIS. The "2" appended to a command (e.g., SUB2) generally denotes it is for processing a sequence. Similarly, for dividing by a flat-field:


This performs the operations: IMAG1/FLAT=IMAG1, IMAG2/FLAT=IMAG2, IMAG3/FLAT=IMAG3. Here, the output sequence has the same name as the input sequence, which saves space on the hard disk but makes the operation nonreversible: Use with care! The command DIV2 performs a little bit more than a simple division as it normalizes the median level of the input images, which then preserves the mean level of the processed images.

IRIS includes many complex commands (which are also available in dropdown menus). As an example, to compute the median sum of an image sequence, it is sufficient to enter from the command line:


SMEDIAN (literally stack median) computes the median pixel from the pixels in an image sequence of 15 images with a generic name "IMAG." The result is displayed on the screen; again, it can be saved in a file with the SAVE command.

IRIS includes more than 200 commands. When it is considered that many of them have multiple suboptions, the total number of available functions, from the console, is estimated to be close to 5000! Their description is beyond the scope of this chapter. For an exhaustive list of IRIS functions it is better to refer to the Web page:


Let us go a bit deeper into the key steps of the astronomical imaging process by first considering how best to remove the thermal component from the raw image. The thermal signal, also called the dark signal or current, is an artifact due to the spontaneous generation of electrical charges within the sensor itself, triggered by temperature. Despite considerable progress over the last decade in sensor technology (MPP technology, for instance) the thermal signal is still significant in deep-sky imaging, where the exposure duration is generally quite long. In some circumstances images can be taken with a noncooled device, which is the case with a digital camera. The issue of the elimination of the dark signal is tricky because it is not usually a constant, as temperature could have varied from one exposure to the next. Under these circumstances, if a unique dark image, acquired in darkness with the same exposure length, is subtracted from all the images, the result will not be reliable. IRIS uses an optimal function of the thermal signal, which compensates for these variations in producing the calibrated image. The basic idea is to multiply the dark image by a value, which is computed to minimize the noise in the resulting image. The constant value c is given by the formula:

With Si,j the pixel intensity at i,j coordinates on the processed image, Di,j is the corresponding intensity in the dark image and n is the number of pixels concerned. The user selects the area where the computation is to be carried out by simply dragging a box with the mouse. As an example, if the raw image is already loaded in memory, the command OPT DARK computes the optimal coefficient using the master dark-frame image (DARK) stored on the hard disk, multiplies each pixel in the image by this constant and subtracts the result from the raw image. As you may have guessed, the related command OPT2 is available to automatically process an image sequence.

Was this article helpful?

0 0
Get Paid to Take Digital Photos

Get Paid to Take Digital Photos

Reasonable care has been taken to ensure that the information presented in this book isĀ  accurate. However, the reader should understand that the information provided does not constitute legal, medical or professional advice of any kind.

Get My Free Ebook

Post a comment