S.B. Shokouhi (Iran)
Boundary tracing, boundary network, structured light
Feature extraction of a structured light pattern means
to find and correct the broken binary and single pixel
thickness image. In most applications for correcting the
pattern, a perfect boundary network should be available.
In this paper, it’s shown that how we can implement the
boundary network using the boundary-tracing algorithm
based on the directional coding. Both methods of the
difference code and neighborhood windowing are
applied for implementing the classification algorithm.
All possible ambiguities on the boundary can be found
and compensated for.