Sigma Clipping is a technique for combining data, including images, which rejects "outlier" data. A simple average is the most effective method of removing random noise if the noise is purely Gaussian in nature. If there are outlier values, perhaps bad pixels caused by cosmic rays or other transient event, the Sigma Clip algorithm will remove them prior to averaging the remaining data.
The Sigma Factor is a number that sets a "threshold" beyond which a pixel value is declared to be an outlier and rejected. The higher the Sigma Factor, the less data is rejected; lower values will cause more data to be rejected.
Sigma Clip can be used to remove hot/cold pixels from an image stack, if combined with dithering.
See also SD Masking.