Using Filters to Enhance Detail

MaxIm DL offers a wide variety of filters for image processing.  These include Kernel Filters, FFT Filters, Unsharp Mask, Digital Development, Rank Filter, Local Adaptive Filter, and Remove Gradient.  By far the most commonly used are the Kernel and FFT Filters.  Perhaps the most useful is Digital Development Processing, or DDP.  This is given its own section on the Advanced Processing with MaxIm DL page.  Below is an overview of the filters available in MaxIm.

MaxIm Filters

• Kernel Filters - These are generally used for basic sharpening or smoothing.  There are also kernel filters for removing dead pixels (black spots in the image) and hot pixels (from cosmic rays or excessive dark current).

• FFT Filters - These filters give you more control over sharpening and smoothing than the kernel filters.

• Unsharp Mask - Despite the "unsharp" in the name, this is a sharpening technique.  An unsharp mask filter simulates the effect of a true unsharp mask which involves subtracting a blurred (or smoothed) version of an image from itself. 

• Digital Development - This technique combines gamma stretching with unsharp masking, making it an excellent filter for bringing out detail within the bright cores of galaxies.  (See Advanced Processing section for this filter.)

• Rank Filter - Enhances faint detail but can significantly increase background noise.

• Local Adaptive Filter - Sharpens low-contrast details in an image but tends to increase background noise.

• Rotational Gradient Filter - Enhances low-contrast radial structures such as cometary jets.

Above:  An unfiltered image of the Whirlpool Galaxy.  Compare this to the filtered images below.

Kernel Filters

Two of the basic types of filters in any image processing program are the High-Pass and the Low-Pass filters.  High-Pass filters sharpen an image, and Low-Pass filters soften the image.  Why soften the image?  Because a low-pass filter tends to remove noise, and sometimes a less sharp but less noisy image is preferable to a sharp but noisy one.  High-pass filters tend to increase noise.

Kernel filters in MaxIm DL offer a couple simple ways to use high-pass or low-pass filters on an image.  Several other filter types are offered as well, but their use is more limited.  To use a kernel filter, select Filter > Kernel Filters from the main menu, or click the Kernel button on the toolbar.

Above:  The Kernel button is the leftmost filter toolbar button.  It looks like a popcorn kernel.

This will open the Kernel Filter window where you can select the type of filter to use.  Some filters have adjustable parameters, but in general the kernel filters are somewhat limited in user control.  This is a good thing if you desire a quick enhancement.  If you like more control, use the FFT filters (described below).

Above:  Kernel Filters window set for a simple Low-Pass filter.

Above:  A low-pass-filtered image of the Whirlpool Galaxy.

• High-Pass/High-Pass More:  These filters sharpen the image but add a little noise.  The More option provides a stronger filter.
   

• Low-Pass/Low-Pass More:  These filters smooth the image and reduce noise.  The More option provides a stronger filter.
   

• Dead Pixel/Hot Pixel:  Dead pixels are unresponsive pixels in the CCD and appear dark.  Hot pixels are caused by cosmic rays or pixels with high dark current and appear bright.  These filters remove these defects by measuring the brightness of surrounding pixels.  The Threshold value controls how bright the surrounding pixels are before the Dead or Hot pixel is detected and removed.
   

• Average/Median:  These are used to smooth the image and reduce noise.  In this sense they are similar to Low-Pass filters.
   

• Dilation/Erosion:  Dilation replaces a pixel with the highest of the surrounding pixels, while Erosion replaces a pixel with the lowest of the surrounding pixels.  Erosion tends to make stars disappear in deep-sky images.  These are probably not the most useful filters, but you never know when they might come in handy!
   

• Kernel Size:  Adjusts the number of surrounding pixels used in the filter calculations.  A larger kernel size tends to have a stronger effect on the image.

FFT Filters

FFT stands for Fast Fourier Transform, which is something your college calculus professor mentioned and you figured you would never use, so you didn't pay attention.  Well, fortunately for your CCD images, someone else did pay attention and came up with FFT filters.  These are High-Pass and Low-Pass filters, similar to those used in the Kernel Filters but with much more user control.

To use the FFT Filters, select Filter > FFT Filters from the main menu, or click the FFT filter button on the toolbar.

Above:  The FFT Filter button is the second in the filter toolbar.  It looks like a little graph with a blue line.

Above:  FFT Filter window set for a Mild High-Pass filter.

Above:  A mild high-pass-filtered image of the Whirlpool Galaxy.

• Filter Type:  This is where you select High-Pass (sharpening) or Low-Pass (smoothing) filtering.
   

• Filter Hardness:  This controls the values of the Cutoff and Weight parameters and determines the effect the filter will have.  Mild filters have little effect for a subtle sharpening or smoothing.  Hard filters have a strong effect for more dramatic sharpening or smoothing.  Custom allows you to manually enter values into the Cutoff and Weight parameters.
   

• Cutoff/Weight:  Decreasing the Cutoff gives the filter a stronger effect.  Increasing the Weight makes the filter stronger.

Unsharp Mask

Unsharp mask is a type of sharpening filter.  It works by first creating a low-pass-filtered version of the image, then subtracts this blurred picture from the original to provide a sharpening effect.  Since a blurred image will have larger stars than the original, subtracted a heavily blurred mask from the original can result in black haloes around stars.  Unsharp masking can easily be overdone if not applied carefully.

To use the Unsharp Mask, select Filter > Unsharp Mask from the main menu, or click the Unsharp Mask button on the toolbar.

Above:  The Unsharp Mask button is the third in the filter toolbar.  It looks like a little black ninja mask.

Above:  Unsharp Mask window set for an FFT filter, mild hardness, and 50% weight.

Above:  Unsharp masked image of the Whirlpool Galaxy.

• Filter Type:  Allows you select which type of Low-Pass filter to use for the unsharp mask.  These work in the same way as the FFT and Kernel Filters described above.  The FFT gives more user control.
   

• FFT Hardness:  If you select FFT Low-Pass for a filter type, you can adjust the settings for the filter as you would for the standard FFT Low-Pass filter.  A Mild FFT produces a subtler effect in the final unsharp mask, whereas a Hard FFT produces a stronger effect.
   

• Cutoff:  Has the same effect on the FFT filter mask as the Cutoff setting does in the regular FFT Filters.  A lower value has a stronger effect.
   

• Mask Weight:  Determines how strongly the mask image is subtracted from the original.  A higher value produces a sharper final image.

Rank Filter

This filter tends to amplify faint detail.  However it also increases noise significantly, since faint detail is typically composed of noise!  A very noise-free image, well-exposed and composited from many individual images, may benefit from this filter, but its use is not common.

Local Adaptive Filter

The local adaptive filter preferentially enhances areas of low contrast.  Again this tends to increase noise, but can work well for low-contrast planetary detail.

Next, Combining Files....

Next Page