/** @file RELEASE_NOTES.TXT
This file gives a high level overview of recent changes.

Revision 2.3

- WARNING: VC++ build files haven't been updated in a while, so if
  you're building on MS Windows, you'll have to update them, and 
  maybe do some light porting of the new code.
- Added convenient "take()" funtion to utilities.h.
- Added solveCubic() & solveQuartic() functions, and extended
  solveQuadratic().
- Added subpixelInterpolate() to do quadratic interpolation of array
  (and by extension, computerVision::Image) values within 3x3 and 3x1
  neighborhoods.
- Added class Index3D.
- Added indexing operators to Vector2D, Vector3D, and Index2D classes
- Several minor upgrades and bugfixes.

Revision 2.2

- WARNING: VC++ build files haven't been updated in a while, so if
  you're building on MS Windows, you'll have to update them, and 
  maybe do some light porting of the new code.
- Added clear() methods to Vector2D and Vector3D.
- Added class ArrayND to represent arrays of arbitrary dimension.
- Added simple special-case convolution routine for ArrayND.
- Added (unstable) MaxRecorder class to make it easier to keep 
  track of extrema in loops.
- Some code cleanup and other minor updates.

Revision 2.1

- Added several new classes:
  - BoxIntegrator2D allows fast (amortized) integration over 2D arrays.
  - NormalizedCorrelator compares 1D signals.
  - BSpline2D is still unstable.  Use with caution.
- New functions normalizedCorrelation(), argmin2D(), and argmax2D() in
  utilities.h.
- Added missing return statement in AmanatidesWoo?D::operator=().
- Minor updates and comment changes.

Revision 2.0

- Through version 1.28, dlrNumeric was part of the larger dlr_libs
  package.  This release makes dlrNumeric into a stand-alone
  package.  Version 2.0 of dlrNumeric is identical to the version in
  dlr_libs version 1.28.

**/