Image Processing Tips



Removing Diagonal Bands

Diagonal Bands

Diagonal Bands

I find that my Logitec QC 4000 and SAC7A cameras both tend to produce diagonal bands. In this tutorial I will show the reader how to safely remove the bands without sacrificing image detail. The first step in the process is to do the normal alignment and stacking with Registax. Instead of applying wavelets you will want to save the image as a 16 bit TIFF image after the stacking process. Do not apply any wavelets at this point. Next, open the image with AstraImage 2.5MAX and split the image into its RGB components.

Registax with wavelets turned off

Registax with wavelets turned off prior to saving the image for FFT editing.

Saving the image in TIFF format.

Saving the image in TIFF format.

The image opened in AstraImage 2.5MAX.

The image opened in AstraImage 2.5MAX.

Splitting the image into RGB components.

Splitting the image into RGB components.

RGB components ready to be edited.

RGB components ready to be edited.

After splitting the image into its RGB components the next step is to edit each color plane with the Fast Fourier Transform (FFT) editor. The FFT of each color plane will contain the bulk of the image detail in the central portion of the two dimensional graph. The image noise )banding) will be found in smaller regions outside the central portion of the image. The FFT editor allows us to edit out those outlying features (noise). You will want to experiment with the editor to see what happens when you edit different parts of the FFT.

The FFT Editor option in AstraImage 2.5MAX.

The FFT Editor option in AstraImage 2.5MAX.

The FFT of the blue color plane.

The FFT of the blue color plane. Notice the outlying spots along the diagonal of the graph.
These represent the noise that we want to remove.

>Using the ellipse tool.

Using the ellipse tool I have edited out the noise.

Once the noise has been edited out of the FFT click the OK button. AstraImage will now do an inverse transformation giving you a relatively noise free image. Repeat this process for the green and red color planes.

The FFT of the green color plane.

The FFT of the green color plane.

The FFT of the green color plane with the noise edited out.

The FFT of the green color plane with the noise edited out.

The FFT of the red color plane.

The FFT of the red color plane.

The FFT of the red color plane with the noise edited out.

The FFT of the red color plane with the noise edited out.

After each of the color planes has been edited with the FFT editor and the inverse transformations have been applied it is time to recombine the edited versions of the color planes back into a color image. Select the RGB Combine option under the Process menu. In the RGB Combine dialog select each of the edited color planes and recombine them.

The recombined color image

The recombined color image.

Now you are ready to save the image as another TIFF image that you will open with Registax for wavelet processing.

Saving the image as in TIFF format

Saving the image as in TIFF format. Notice that each time I save the image
I append the file name with the name of the processing step that was
just completed.

Wavelet processing

Wavelet processing in Registax.. Notice that the diagonal bands are now gone.

After wavelet processing you can adjust the color balance and do any other post processing that you would normally do. What started out as a hopeless looking image has turned out to be a fairly nice image.

automatic color balance tool

Applying the automatic color balance tool in Paint Shop Pro.

Removing random noise

Removing random noise with Neat Image.

before

after

Before

After