First let us summarize the previous discussion by giving the
definitions of connected sets of lattice
points, or hypercube vertices, of
and its complement. These definitions
are reminiscent of both graph connectivity and
digital topology (See [11],[12], [13]).
In the following, we assume that f interpolates the extended
.
For a given threshold
,
two High (respectively Low) vertices are adjacent
iff they are edge adjacent or diagonally adjacent according to the disambiguation
rule at
.
Path connectivity is the transitive
closure of the adjacency relation.
A set
(of lattice vertices) is connected iff there is path,
completely contained in the set, between each
pair of vertices.
A connected set of High (resp. Low) vertices is defined
to be a path-connected set containing only High (resp. Low) vertices.
A connected component of Highs (resp. Lows) is just a maximal path connected set of Highs (resp. Lows).
By a d-dimensional hypercube we mean a hypercube defined by 2d adjacent integer lattice points, so that the boundary of a d-dimensional hypercube will consist of d-1-dimensional hypercubes.
Let C be a d-dimensional hypercube and M a component of
.
A (hyper)surface patch of M, with respect to C, is a d-1-dimensional connected
component of
.
Toward developing a Digital Morse Theory,
let us make the following assumptions (or axioms) on the components (iso-surfaces) of
,
for each
not in the range of the extended
.
If an interpolation function satisfies the axioms it is called a Digital Morse interpolant.