next up previous contents index
Next: Metropolis Algorithm Up: No Title Previous: Median Filter

Metric

  A metric or distance function is a function d(p,q) of two points p and q which satisfies:

Frequently used examples are:

The Euclidean distance: in two dimensions,

In a digital image, the elements of p and q are row and column numbers. Generalized to any number of elements in p and q, one can write

Points with equal from p form a circle (sphere, hypersphere) of radius around p.

The city block distance: in two dimensions,

with obvious generalization to more dimensions. Points (pixels in an image) with equal from p form a diamond around p; in an image:

Points with from p are called the 4-connected neighbours of p.

The chess board distance: in two dimensions,

Points with equal from p form a square around p; in an image:

Points (pixels in an image) with from p are called the 8-connected neighbours of p. e.g. [Rosenfeld76].

A metric can also be defined in a binary space, e.g. as the distance between two bit patterns ( Hamming Distance).



Rudolf K. Bock, 7 April 1998