4. Example: De-noising an image

In this example we remove instrument noise from an image. See also Filtering Tool.

Negative spikes (-33) are visible due the to instrument missing an acquisition cycle.

1. Using the Filtering tool select the appropriate filter.

   Available filters are Rolling Mean and Rolling Median.

2. Select appropriate filter size.

   Typical window sizes are 3, 5 or 7. A larger window size will be more sensitive in detecting outlying data but is more likely to introduce artifacts.

3. Select the appropriate filter threshold.

   An ideal threshold will change invalid data while leaving valid data untouched.

A rolling median filter replaces the invalid values with the local median.