Author  - Elmar de Koning
    Support - edeko.nosp@m.ning.nosp@m.@gmai.nosp@m.l.co.nosp@m.m
    Website - http://psimpl.sf.net
    Article - http://www.codeproject.com/KB/recipes/PolylineSimplification.aspx
    License - MPL 1.1

psimpl

    'psimpl' is a c++ polyline simplification library that is generic, easy to use, and supports
    the following algorithms:

    Simplification
    + Nth point - A naive algorithm that keeps only each nth point
    + Distance between points - Removes successive points that are clustered together
    + Perpendicular distance - Removes points based on their distance to the line segment defined
      by their left and right neighbors
    + Reumann-Witkam - Shifts a strip along the polyline and removes points that fall outside
    + Opheim - A constrained version of Reumann-Witkam
    + Lang - Similar to the Perpendicular distance routine, but instead of looking only at direct
      neighbors, an entire search region is processed
    + Douglas-Peucker - A classic simplification algorithm that provides an excellent approximation
      of the original line
    + A variation on the Douglas-Peucker algorithm - Slower, but yields better results at lower resolutions

    Errors
    + positional error - Distance of each polyline point to its simplification

    All the algorithms have been implemented in a single standalone C++ header using an STL-style
    interface that operates on input and output iterators. Polylines can be of any dimension, and
    defined using floating point or signed integer data types.

changelog

    28-09-2010 - Initial version
    23-10-2010 - Changed license from CPOL to MPL
    26-10-2010 - Clarified input (type) requirements, and changed the behavior of the algorithms
                 under invalid input
    01-12-2010 - Added the nth point, perpendicular distance and Reumann-Witkam routines; moved all
                 functions related to distance calculations to the math namespace
    10-12-2010 - Fixed a bug in the perpendicular distance routine
    27-02-2011 - Added Opheim simplification, and functions for computing positional errors due to
                 simplification; renamed simplify_douglas_peucker_alt to simplify_douglas_peucker_n
    18-06-2011 - Added Lang simplification; fixed divide by zero bug when using integers; fixed a
                 bug where incorrect output iterators were returned under invalid input; fixed a bug
                 in douglas_peucker_n where an incorrect number of points could be returned; fixed a
                 bug in compute_positional_errors2 that required the output and input iterator types
                 to be the same; fixed a bug in compute_positional_error_statistics where invalid
                 statistics could be returned under questionable input; documented input iterator
                 requirements for each algorithm; miscellaneous refactoring of most algorithms.