# Median Filter

The median represents the middle data point that half of data is smaller and half of data larger than this point. This is a robust estimator to remove outliers from a data set. Consequently, this tool can be useful as a naive denoiser, effective against impulse noise.

The layout of the window dialog is quite simple and few settings are available.

**Kernel size**: From \(3\times 3\) to \(15\times 15\), this defines the size of a squared kernel that is used to apply the filter. The larger the kernel, the more blurred the result will be.**Iterations**: This defines the number of passes of the kernel.**Modulation**: In Siril, modulation is a parameter between 0 and 1 mixing the original and processed images. A value of 1 keeps only the processed image, a value of 0 does not apply any median filter at all.

Siril command line

```
fmedian ksize modulation
```

**ksize**x

**ksize**(

**ksize**MUST be odd) to the loaded image with a modulation parameter

**modulation**.