My own work has been on "theoretical" mesh generation, meaning abstract problems motivated by the real issues. Most of this work studies the tradeoffs among the various mesh generation goals. Here is an example theorem from a paper by myself, Scott Mitchell, and Jim Ruppert: any n-sided polygon can be triangulated with O(n) nonobtuse triangles. Check out the triangulated pig below.

For more on my mesh generation work, see my list of publications. Most of my mesh papers are collaborations with David Eppstein of UC-Irvine; his page of publications on mesh generation and optimal triangulation includes short abstracts. For much more on mesh generation, both practical and theoretical, I recommend David Eppstein's Geometry in Action, Paul Heckbert's Collection of Links, Jonathan Shewchuk's Triangle, Robert Schneiders' Mesh Generation on the Web, Steve Owen's Meshing Research Corner, and Roger Young and Ian MacPhedran's Internet Finite Element Resources.