Where
A
and
B
are image matrices, and
(x, y)
are the spatial coordinates within
A
and
B.
Before pixel-by-pixel between the template and ROI, the segmented characters have
to be resized to the same size with the templates stored to avoid the tracing errors. The
recognition process cannot be carried out if the extracted characters are not same size
with the defined template. In this report, the sizes of templates stored which are of 24
pixels width and 42 pixels height. Figure 3.9 shows the defined templates of this project.
Table 3.6 shows the character extraction and recognition pseudocode.