Image Processing Reference
In-Depth Information
Fig. 2.2 Edge map after the enlargement
cell Y 2 i , 2 j + 1 must be evolved to a cell that will produce a light intensity which is
more closer to the value of the edge pixel and less close of the non-edge pixel. In
addition, if no cell in its immediate neighbor is marked as a non-edge cell, its next
state must correspond to a value which leads to a homogenous area.
It must be mentioned that all non-edge and edge cells of the enlarged map are
surrounded by eight (8) unknown cells based on the aforementioned selected Moore
neighborhood with r
1 and after the evolution; they are marked with states which
leads to equal values with the corresponding pixels of the low resolution image.
In addition, every cell stated as “2” appears either two known cells (horizontal or
vertical cells) or four known cells (central cell) in its Moore neighborhood which
will eventually designate the next state. The total number of the used states and the
constructed transition rules are twenty four (24) therefore, each CA cell is marked
with one discrete number in the range of
=
after the evolution of the CA. Finally,
null boundary conditions are used in order to evolve the state of the boundary cells.
Figure 2.3 presents simple cases of the above rationality.
[
0
,
23
]
2.3.3
Remapping Process
At this stage of process, every pixel value of the resized image is defined based on
the state of the corresponding cell. Let us consider that f i , j and F 2 i , 2 j correspond to
Search WWH ::




Custom Search