capture and decode are based on a variety of papers. they're not
structured very well, and i don't recommend using them as a reference.
however, FastThreePhase is fairly clean. it's based very directly on
"Multilevel quality-guided phase unwrapping algorithm for real-time
three-dimensional shape reconstruction" by Song Zhang, Xiaolin Li, and
Shing-Tung Ya (2007).
it also contains some optimizations that i developed independently, but
are similar to ones described in "Dual-frequency pattern scheme for
high-speed 3-D shape measurement" by Kai Liu, Yongchang Wang, Daniel L.
Lau, Qi Hao, and Laurence G. Hassebrook (2010).
the best way to understand theory of individual chunks of code s is to
look at the commit log:
hope that helps,