Computational Geometry in the Human Brain - Discrete and Computational Geometry and Graphs - page 157

Fig. 14. Straight skeleton constructed by Algorithm 2.

( (

Fig. 15. Degenerate straight skeleton: (a) straight skeleton; (b) close-up diagram of
the central part.

computations were performed in single-precision floating-point arithmetic. Our
algorithm was able to compute the straight skeleton, as shown in Fig. 15 (a).
However, if we expand the central part by 10 5 , we get the diagram in Fig. 15 (b),
where we can see many vertices instead of a single vertex. This kind of distur-
bance is not surprising; we note that the algorithm gave a topologically consistent
output even though the polygon was highly degenerate.
This algorithm is similar to the computations in the human brain in the
sense that both are persistent once a decision has been made, regardless of its
accuracy. In this way, both are able to achieve robustness against imprecise
numerical computations.