New XMAN Processing Method

Owned by jr_mills
Apr 05, 2022
3 min read

New XMAN Processing Method

!worddavde54bb1ebeccd424b9187ab8389d790c.png|height=440,width=224!VS
This is a quick note on the new image processing code installed on the XMAN software during expedition 392. The method was developed for the XSCAN line scan imager and has been adapted to XMAN. Much less artifacts and it is very fast.

PROCESSING

As before we still calculate an attenuation profile and subtract it from the original image and then expand the results (using a LUT function) to cover the dynamic range of a U16 grayscale image.
There are two methods for calculating the attenuation profile:

  • APC: This can be used for any core type that has a continuous profile thru the length of the image (Including XCB and RCB). Line by line, across the image, a profile is extracted. A mean profile is then calculated from these profiles and is used as the attenuation correction.

  • ROTARY: This method is used for discontinuous pieces found in XCB and RCB cores. It uses 200 lines before and after the line being processed to calculate the attenuation correction. This allows it to adapt to changing profiles. But to be honest, we still have not found a good method of processing these types of core.

As before, you use the Image Processing Utility to determine the best correction variables. Your setting will be used for processing during active acquisition. And as before, you can use the utility to refine your processing later using batch mode processing.
Keep in mind when you subtract the attenuation profiles from the original image you are left with a very "mono tone" image. Of the possible 65000 grayscale values in your original image you may only have 3000 values in the corrected imag. To see anything, we must expand the 3000 to cover the 65000 range. Here are the steps:

  1. Find the mean and standard deviation of the values in the corrected image.

  2. Take the difference between the MIN and MAX (set by the user) and divide it by five

  3. Create a series of 5 values starting with MIN and increasing by the value in step 2. For example:

    • MIN = 2 and MAX = 11,

    • MAX-MIN = 9,

    • Divide 9/5 =1.8

    • Create the series (3.8, 5.6, 7.4, 9.2, 11)

  4. Take each value in the array and multiple the standard deviation by this value.

  5. Create a range by subtracting the value in step 3 from the mean (LOW RANGE) and then adding the value in step 3 to the mean (HIGH RANGE)

  6. Extract the values between the LOW and HIGH.

  7. Expand them where LOW = 0 and HIGH = 65000 using a LUT function.

  8. Repeat this for each number in the series.

  9. Add the LUTs and calculate a mean LUT. This will be used to expand the attenuation corrected image.

What going on with all of that?
The small value in the series will only extract the highest contrast values in the image. This is good for seeing small details in the core. Higher values will see more broader detail but will "wash" away the fine details. The purpose of the series to create an image that capture the full range of details.

CREATING THE LUT

On the Processing Utility window, you will see this control:
The MIN and MAX values are used to create the series that in turn create the LUTS. As you change the values the image will immediately show you the results. Very easy to finds the correct values for your sediments.
MASK Threshold is percentage of the image max value and when Trim to Black = True, all values below that value will be set to 0. This can help enhance the image but it can also result in erosion as shown.