Sigma filter

Use the Process > Smoothing Filters > Sigma... command, to reduce statistical noise in your image. The Sigma filter belongs to the smoothing filters. Smoothing reduces image noise, while maintaining coarse structures.

The dialog box that is opened when you use an image processing operation is made up in the same way for every operation. Click here, to switch to a description of this dialog box.

Parameters for the Sigma filter

Define the Size and Sigma parameters in order to apply the filter:

Enter the diameter of the adjacent area in units of pixels in the Size field. You can select diameters of 1-50 in steps of 1. All pixels within the defined neighborhood will all be considered during the averaging.

Enter the value for the full width at half maximum σ of the used Gaussian distribution into the Sigma field. The Sigma parameter influences the form of the Gaussian distribution. A smaller Sigma value creates a narrow distribution with a large maximum value. A larger Sigma value creates a flat and wide distribution.

Exactly what happens?

The idea of the filter is based on the assumption that the noise in the image is distributed bell-shaped, i.e., Gaussian shaped. When using a monochrome image with lots of noise, the frequency of the intensity value to the left and right of the mean value also drop away in a bell-shaped curve, i.e., Gaussian curve.

The Sigma filter is also identified as a neighborhood filter. It replaces the intensity value of the current pixel by the average of all of the intensity values whose distance is smaller than Sigma to the current intensity value. The Sigma value therefore defines the intensity interval within which the pixels in the neighborhood are supposed to be.

1727 05032013