Filter types

Pattern filters

 

With a pattern filter you can detect: areas where pixels have more or less the same value, outliers where the values of all neighbours are largely different from the center pixel, and the directions in which differences between a center pixel and its neighbours are found. A pattern filter always works in a 3x3 environment and works on images and other raster maps with a value domain.

Whether or not any of the 8 neighbours is considered to have more or less the same value as the center pixel, is determined by the threshold value that you have to specify. When the absolute difference between a neighbour and the center pixel is smaller than or equal to the threshold value, the answer is true. For each true neighbour, a certain bit is set. The value assigned to the center pixel in the output map is the bit-wise combination of all true neighbours. For more information, refer to Example pattern filters.

This will give you the following results:

Standard pattern filters:

There is no standard pattern filter stored on disk.

User-defined pattern filters:

A pattern filter always works in a 3x3 environment. In the Filter dialog box as well as on the command line, you can define a pattern filter; specify the threshold value which has to be used. For more information and some examples, refer to User-defined pattern filters.

See also: