Applet Kernel Gaussian Steerable
From BoofCV
Steerable filters are used to describe regions inside of an image. For example, once the orientation of a region has been determined the filters response to first, second, and third order Gaussian derivatives can be computed. The main advantage to using a steerable filter is that they can be computed at an arbitrary orientation fast an efficiently.
In this applet the basis kernels and the output at different angles is shown. The output is computed by summing together the basis kernels with different amounts of weights to produce an output facing a different direction.
