An isotropic operator in an image processing context is one which
applies equally well in all directions in an image, with no particular
sensitivity or bias towards one particular set of directions (*e.g.*
compass directions). A typical example is the zero
crossing edge detector which responds equally well to edges in any
orientation. Another example is Gaussian smoothing. It
should be borne in mind that although an operator might be
isotropic in theory, the actual implementation of it for use on a
discrete pixel grid may not be perfectly isotropic. An example of this
is a Gaussian smoothing filter with very small standard deviation on a
square grid.